ccdoc home page
Ccdoc Issues

All
Summary
Joe Linoff
$Date: 2004/09/30 16:09:24 $
$Revision: 1.35 $

This page keeps track of issues related to ccdoc as they come up. If you click the Summary link at the top of the page it will take you to a summary page. If you click on the status types (such as OPEN) it will take you to a page that lists only issues with that status.

The status field contains either CLOSED, FIXED or OPEN.

These pages were automatically created by a perl script called make_issues.pl that is freely available by clicking here the input to the script is a text file.


Status Summary
Status Number
CLOSED 8
FIXED 144
OPEN 31
Total 183

Entry Summary
Id Title Status
0001 The -rootfile switch parses relative names improperly. FIXED
0002 The ccdoc_top tage is incorrectly placed in the HEAD section. FIXED
0003 Need a new switch, similar to -head, that inserts a file after the head tag to insert custom META elements. FIXED
0004 HTTP server problem with uncommon extensions like .exe_gz and .taz. FIXED
0005 Log output is not flushed properly on solaris platforms. FIXED
0006 Tests weren't working properly. FIXED
0007 WARNING: File '*.h' cannot be read so it will be ignored. FIXED
0008 Turn off the ccdoc generation of special members. FIXED
0009 Compile problem in gcc 3.0 on SuSE linux platform. FIXED
0010 Random character appears between package name and function/macro/etc. name in the HTML title. FIXED
0011 Instantiations of class template methods are reported as global functions. CLOSED
0012 Ccdoc should ignore template method instantiations. FIXED
0013 Add ccdoc to source forge. FIXED
0014 The typedef id for function pointer doesn't get extracted properly by ccdoc. FIXED
0015 Methods with throw clauses surrounded by parentheses don't get extracted properly by ccdoc. FIXED
0016 Source file names of the form ../folder/file.ext don't work properly in the cmd.exe shell under windows. FIXED
0017 Can't differentiate between different revisions of ccdoc 0.8. FIXED
0018 Directory separators in -root names cause run time errors. FIXED
0019 The @see directive no longer accepts user defined links. FIXED
0020 ccdoc crashes FIXED
0021 Report the version/revision information in the default trailer. FIXED
0022 Centralize the version/revision information. FIXED
0023 DOC++ vs ccdoc CLOSED
0024 Missing comma separator in @see output for the alternative URL form. FIXED
0025 Undocumented 0.7a feature for explicitly specifying @pkgdoc URLs is no longer supported. FIXED
0026 Provide an expanded package contents view. FIXED
0027 Provide a way to turn off package source. FIXED
0028 Provide a way to turn off the default author on the package page. FIXED
0029 Provide a way to turn off the default version on the package page. FIXED
0030 Provide a way to turn off default description comments on the page. FIXED
0031 Provide a way to change the "package" string in the title on package pages. FIXED
0032 Users guide and on-help report than -norptpub is the default. FIXED
0033 This construct is incorrectly recognized as a struct: void foo(struct bar* spam). FIXED
0034 For -rpthpc only show packages. FIXED
0035 ccdoc 0.7a to 08 quick reference FIXED
0036 Comments get lost. FIXED
0037 Change the asterisk column header to "Inherited From". FIXED
0038 Change the Path column header to "Entity". FIXED
0039 Performance problem in phase3. FIXED
0040 Protected inherited methods reported. FIXED
0041 Improve 0039 performance enhancement. FIXED
0042 ccdoc bug notification - not correctly determining variable name FIXED
0043 Old hash mark syntax for @see does not work. FIXED
0044 The summary report could use a little more indenting. FIXED
0045 The variable width font in the Code: section is not as readable as the fixed width font used in 0.7a. FIXED
0046 The @see index doesn't seem to work. FIXED
0047 The indenting on the contents page should be the same as the summary report. FIXED
0048 Derived classes in namespaces are not recognized correctly. FIXED
0049 Non-template class method instantiation not recognized correctly. FIXED
0050 Ccdoc fails to parse methods with fct args properly. FIXED
0051 External class method implementations with default arguments are not correctly recognized. FIXED
0052 Backslash character definition ('\\') confuses the ccdoc scanner. FIXED
0053 No line numbers for internal comment parsing warnings. FIXED
0054 Serious problem with typedef of struct FIXED
0055 ccdoc generates default special members for C style structs CLOSED
0056 Incorrect line numbers reported in warning messages. FIXED
0057 "Undocumented" as default short desc. in tables is confusing. CLOSED
0058 Provide a way to turn off the default author and version for packages. CLOSED
0059 Validate ccdoc generated HTML. FIXED
0060 Group entries by type. OPEN
0061 Typedef struct statement does not include struct comments. OPEN
0062 Long entry name deforms HTML table. OPEN
0063 Allow user defined defaults for author, short description and version. FIXED
0064 Friend classes in namespaces are not recognized correctly. FIXED
0065 Links from documentation in namespaces are not recognized correctly. CLOSED
0066 The <sstream> header is not needed in phase3_html.cc. FIXED
0067 Ccdoc does not correctly reset the access specifiers for methods in base classes that are derived as protected or private. OPEN
0068 Core dump on solaris in the index phase. FIXED
0069 Cannot handle construct: "Float operator/(const Float & rval)" FIXED
0070 Users guide documentation error. FIXED
0071 Operator keyword duplicated for method operators. FIXED
0072 New switch to turn off sorting of class entities. FIXED
0073 Can't tell ccdoc to ignore a privately declared default constructor. FIXED
0074 Ccdoc cannot handle international character sets. FIXED
0075 Ccdoc cannot handle function typedef of the form: typedef void fct(int arg1); FIXED
0076 Conflict between -meta switch and ccdoc defined meta variables. FIXED
0077 Ccdoc incorrectly identifies a class attribute name. FIXED
0078 Ccdoc incorrectly identifies a variable as a function. FIXED
0079 platform.pl cannot handle embedded parentheses. FIXED
0080 Trim trailing spaces in path name for get_stmt_no_pkgs(). FIXED
0081 Multi-space blank lines between SHORT and LONG descriptions cause them to merge. CLOSED
0082 Add the @since directive per the javadoc 1.2 specification. FIXED
0083 Modify ccdoc to create a separate links page (as was done in v0.7a). FIXED
0084 Alias @exception to @throws per javadoc 1.2. FIXED
0085 Ccdoc not recognize default constructors with member initialization. FIXED
0086 Comment form to handle single line suffix comments. FIXED
0087 Exceptions were not reported as links. FIXED
0088 An extra invalid link reported for @link references to classes. FIXED
0089 Ccdoc short description syntax is not compliant with javadoc. FIXED
0090 Support the {@link ...} javadoc syntax. FIXED
0091 Constant values are not hyperlinked in the documentation for function arguments. FIXED
0092 The @link syntax does not support the # scoping operator. FIXED
0093 Add javadoc style doc comment inheritance. OPEN
0094 Comment reuse for overloaded functions. OPEN
0095 Comma's inserted for each line in @returns and @deprecated comments. FIXED
0096 Provide an @repeat operator to avoid having to re-type comments. OPEN
0097 Private method is reported incorrectly for overloaded methods with different scopes. FIXED
0098 It would be a help if you made the reference red if the class cannot be found for @link. FIXED
0099 Ccdoc does not correctly follow the standard for default constructors. FIXED
0100 If using switch -files with a wildcard expression the first file is not processed. CLOSED
0101 Template problem. FIXED
0102 Returned nested classes and types in namespaces are not recognized correctly. FIXED
0103 Invalid processing of operator~(). FIXED
0104 Invalid processing of operator&(). FIXED
0105 The javadoc @see tag defines the label part as optional. OPEN
0106 The javadoc spec for @see allows unqualified names as well as qualified names. OPEN
0107 javadoc # syntax is not supported in the {@link...} directive. FIXED
0108 -rptdefasd, -rptdefsd and -rptdefv problems and suggestions OPEN
0109 Print warnings to stderr. FIXED
0110 Invalid processing of operator[]. FIXED
0111 Multiple link file overflow problem. FIXED
0112 Phase 3 exception for operator|. FIXED
0113 Merge separate ccdoc databases. OPEN
0114 Error processing of a operator unsigned long(). FIXED
0115 Mangled links for .\doc\ references on the command line. FIXED
0116 The string constant "\\" isn't parsed correctly. FIXED
0117 Class method reported as global function. FIXED
0118 Resolve partially scoped references. FIXED
0119 Inconsistent command line syntax for -html paths under DOS. OPEN
0120 Add support for the proposed todo tag in javadoc. FIXED
0121 Reuse namespace comments between packages. OPEN
0122 Enhance template reporting. FIXED
0123 {@link ...} not reported correctly in the @deprecated directive FIXED
0124 Scanner condition evaluates to a constant value. FIXED
0125 Hash mark displayed for default {@link...} reference. FIXED
0126 CcDoc does not recognize ccdoc c++ short comment in a "typedef struct" definition. OPEN
0127 CcDoc does not recognize recognize template derivations. FIXED
0128 Order class entities the same way that were ordered in v07. OPEN
0129 All subdirectories are picked up on NT when the * wildcard is specified. FIXED
0130 Code section wraps in a strange way for small browser windows. FIXED
0131 Support comma separated declarations. OPEN
0132 support single-line prefix comments via //@+ OPEN
0133 Anonymous namespaces are not handled correctly. FIXED
0134 Need a forward link to the contents section for classes and packages. FIXED
0135 Fix the copyright notice in the code. FIXED
0136 Ccdoc gets confused by a nested struct. FIXED
0137 The {@link} directive does not work @param clauses. OPEN
0138 The @see and {@link} directives do not correctly reference entities in different packages. FIXED
0139 Core dump when an invalid package name of the form "A::B:" appears. FIXED
0140 Core dump when a friend declaration of the form: foo&operator<< appears. FIXED
0141 Architecture is not reflected in the bin directory name. FIXED
0142 No man page for linux ports. FIXED
0143 Support javadocs frames format. OPEN
0144 Mark static members and methods in the class overview. FIXED
0145 It is impossible to reference an operator in the @see and {@link..} directives. FIXED
0146 Namespace processing doesn't work properly in certain cases. FIXED
0147 Lots of compiler warnings when building ccdoc using the Borland C++ compiler. FIXED
0148 Allow users to customize ccdoc output using a CSS file. OPEN
0149 Template method implementations confuse ccdoc. FIXED
0150 Make it easier to update the revision number in the source files. FIXED
0151 Translate the generated html to other languages. OPEN
0152 Ignore duplicate macros names. FIXED
0153 Add a heuristic to ignore include guards and DLLIMPORT macros. FIXED
0154 Eliminate the requirement that the -rptmac and -rptmac1 switches appear in phase 1. FIXED
0155 Incorrect documentation when there is only one ccdoc namespace comment for namespace declarations in multiple files. FIXED
0156 More namespace problems in r35. FIXED
0157 In some cases operator code is being processed incorrectly. FIXED
0158 Class method implementations are reported as global functions. FIXED
0159 Very long names make the contents table very hard to read. FIXED
0160 Empty function names appear under certain conditions. FIXED
0161 Strange function names like "max)" appear sometimes. FIXED
0162 Doxygen compatibility - support single line comments using doxygen syntax. FIXED
0163 Doxygen compatibility - ignore the @endlink directive. FIXED
0164 Doxygen compatibility - ignore @file comment blocks. FIXED
0165 Doxygen compatibility - support virtual grouping of definitions. OPEN
0166 Variable values are not reported properly when braces occur in the value. FIXED
0167 In German we want to use embedded dots in the short description. OPEN
0168 Slow g++ 3.2 compile of help.cc because of the large help string. FIXED
0169 Man page has formatting problems. FIXED
0170 Bug on NT platforms parsing command line with spaces and wildcards. FIXED
0171 Constructors with constructed default arguments are reported with the wrong method name. FIXED
0172 Template friend declarations are not resolved properly when packages are used. FIXED
0173 The md5hash macro is being improperly ignored by the rptmac1 heuristic. FIXED
0174 Bug#195282: ccdoc: FTBFS with g++-3.3 FIXED
0175 Generate an index in HTML and XML. OPEN
0176 Enhance the table of contents for easier navigation with letter based links at the top. OPEN
0177 Invalid link from parameter variable name to member variable. OPEN
0178 Namespace/anonymous enum interaction bug. OPEN
0179 Add a new @entity directive to allow referencing. OPEN
0180 Limit the length of generated file names. OPEN
0181 Copy constructor documented incorrectly under some conditions. OPEN
0182 Add directory tree/outline expansion (+/-) capabilities to better organize the information. OPEN
0183 Ccdoc started failing with a segment violation in phase 3 when processing a large system (thousands of classes) after a small change to one of the header files. FIXED

Title: The -rootfile switch parses relative names improperly.
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/08/22
Resolved By: Joe Linoff
Resolved On: 2001/08/22
Id: 0001
[Top]
[Summary]
[Next]
Problem Description:
Little bug: Switch '-rootfile': If there is no parh given, the first letter of filename will missing from the HTML hrefs. Eg:
    '-rootfile index.html' => '<a href='ndex.html'>@root</a>' ('i' missing)
but
    '-rootfile ./index.html' => '<a href='index.html'>@root</a>'
Solution Description:
Found a problem in make_file_url. The separator iterator was being incremented when the path was completely relative.

Broken Code: phase3_html.c:1845

  // ================================================================
  // Make file URL, strip off the leading directories.
  // ================================================================
  void ccdoc::phase3::html::make_file_url(string& url,string& fn) const
  {
    // Strip off the leading directories to
    // make the relative reference work.
    string::iterator itr = fn.begin();
    string::iterator relpath_itr = fn.begin();
    for(;itr!=fn.end();++itr) {
      if( '/' == *itr ) {
        relpath_itr = itr;
      }
    }
    url = "";
    for(++relpath_itr;relpath_itr!=fn.end();++relpath_itr) {
      url += *relpath_itr;
    }
  }

Fixed Code

  // ================================================================
  // Make file URL, strip off the leading dSeptember 18, 2001 (Tuesday)

The new release of ccdoc is available: "ccdoc v0.8 r18 2001/09/18".

This release fixes issues 0052 and is the first release that contains the linux port.





irectories.
  // ================================================================
  void ccdoc::phase3::html::make_file_url(string& url,string& fn) const
  {
    // Strip off the leading directories to
    // make the relative reference work.
    string::iterator itr = fn.begin();
    string::iterator relpath_itr = fn.begin();
    for(;itr!=fn.end();++itr) {
      if( '/' == *itr ) {
        relpath_itr = itr;
        ++relpath_itr; // ISSUE 001 FIX
      }
    }
    url = "";
    for(;relpath_itr!=fn.end();++relpath_itr) {
      url += *relpath_itr;
    }
  }

Title: The ccdoc_top tage is incorrectly placed in the HEAD section.
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/08/22
Resolved By: Joe Linoff
Resolved On: 2001/08/22
Id: 0002
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Very little bug: <a name=ccdoc_top></a> is placed into the HEAD, which is not correct (according to HTML 4.0 at least, but I don't think its right for HTML 3.x). It should be pleced just *before* that <table border=0 width="100%"> in the BODY. However I don't know a browser which is confused by this.
Solution Description:
Moved <a name=ccdoc_top></a> to the BODY section.

Title: Need a new switch, similar to -head, that inserts a file after the head tag to insert custom META elements.
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/08/22
Resolved By: Joe Linoff
Resolved On: 2001/08/22
Id: 0003
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
What is missing, a switch similar to '-head' which inserts the specified file just after the <head> tag. This is required to insert custom META elements, primary for set the charset. (Note that most browsers accepts META http-equivs in the BODY (!) so I can use -head, but its not correct HTML.)


Perhaps a -meta <file> switch?
Solution Description:
Added a -meta <file> switch that allows the user to specify a file that contains custom meta variables.

Title: HTTP server problem with uncommon extensions like .exe_gz and .taz.
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/08/22
Resolved By: Joe Linoff
Resolved On: 2001/08/23
Id: 0004
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The HTTP server on www.joelinoff.com gives back text/html content-type for those files with uncommon extensions, like ccdoc.exe_gz, doc.tgz. It confuses many browsers. (I think the solution is to set DefaultType in the httpd.conf to application/octet-stream, unless u serve html or plain text documents with odd or no extension.)

(btw, if some users will report CRC errors in the gzipped files, this is because there is some conflict between the HTTP server and some (or only one?) proxy-es. Dont ask why, I got partial-content (HTTP 206) without the last few kbytes of files (=> gunzip detects CRC error), then decided to bypass the cache-proxy.)


I completely agree. I have been trying to work with my domain hosting service to correct this (Earthlink) but they are very slow to respond.
Solution Description:
I resolved this by renaming the archives that were causing problems. Apparently the ".gz" extension does the right thing.

Title: Log output is not flushed properly on solaris platforms.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/08/22
Resolved By: Joe Linoff
Resolved On: 2001/08/22
Id: 0005
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Part of the log file is missing on the solaris platforms because the s_log global variable is not flushed at the end of the program.
Solution Description:
This was resolved by flushing the s_log object before the program exits.

Title: Tests weren't working properly.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/08/22
Resolved By: Joe Linoff
Resolved On: 2001/08/22
Id: 0006
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Tests are being reported as successful when they are not.
Solution Description:
There was a bug in the test.pl file that caused all tests to pass.

Title: WARNING: File '*.h' cannot be read so it will be ignored.
Status: FIXED
Reported By: Yuriy Stul
Reported On: 2001/08/23
Resolved By: Joe Linoff
Resolved On: 2001/08/23
Id: 0007
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Wildcards do not work in my cmd.exe shell. Here is my example:
    C:> ccdoc.exe -db LB.db -pkg LBTools *h
    WARNING: File '*.h' cannot be read so it will be ignored.
Solution Description:
Fixed in r10.

I modified the MSVC version of the program to attempt to process wildcards if the attempt to open the specified file fails. The -files switch still works.

When I originally closed this issue, I used the following reasoning.

Wildcards do not work in the cmd.exe shell on Windows platforms because it does not expand them. Instead it expects the application to do the expansion.

Ccdoc cannot do that because a user can run many different shells on Windows platforms, such as: ksh, rsh and tcsh (I use tcsh). Most of those shells (in fact all of them execept cmd.exe) expand wildcards before passing the the program.

Unfortunately, ccdoc can't tell the difference a legitimate file name with a * or ? in it from the tcsh and a wildcard request from the cmd sheel so you have to tell it using the -files switch.

You can work around this problem by using the new -files switch as follows:

    C:> DIR \B *.h > hfiles.txt
    C:> ccdoc -db foo.db -pkg foo -files hfiles.txt

Title: Turn off the ccdoc generation of special members.
Status: FIXED
Reported By: Martin Trappel
Reported On: 2001/08/23
Resolved By: Joe Linoff
Resolved On: 2001/08/23
Id: 0008
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc seems to generate automatic comments for Standard Copy Constructor, Destructor and Assignment Operator.

Can I disable this, since I have a few classes where I do not have these and I don't quite see the point in having doc for non-existant members?


This was originally added because of request from Lou Sanchez-Chopitea because, according to the standard, these special members always exist whether you declare them or not.

I agree with your point that you should not be forced to see them unless you want to and will look into providing a switch that gives you the control to turn the reporting of them on or off.

Solution Description:
Added the -[no]cdsm switches to phase 1 to allow users to control whether or not default special members are created.

Title: Compile problem in gcc 3.0 on SuSE linux platform.
Status: FIXED
Reported By: Michael Davis
Reported On: 2001/08/23
Resolved By: Joe Linoff
Resolved On: 2001/08/23
Id: 0009
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
We use gcc 3.0 on a SuSE linux platform here, and your code compiled with only one glitch. Here is a patch should you want it:
*** phase3_html.cc.orig Thu Aug 23 09:52:22 2001
--- phase3_html.cc      Thu Aug 23 09:36:43 2001
*************** void ccdoc::phase3::html::write_class_de
*** 476,482 ****
        --xitr;
        make_tag_id(*xitr,prev_tag);
        }
!       if( itr ) {
        vector<statement::base*>::iterator xitr = itr;
        ++xitr;
        if( xitr != contents.end() ) {
--- 476,482 ----
        --xitr;
        make_tag_id(*xitr,prev_tag);
        }
!       if( itr != contents.end() ) {
        vector<statement::base*>::iterator xitr = itr;
        ++xitr;
        if( xitr != contents.end() ) {
Solution Description:
Thank you. I have updated and released the fix.

Title: Random character appears between package name and function/macro/etc. name in the HTML title.
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/08/23
Resolved By: Joe Linoff
Resolved On: 2001/08/23
Id: 0010
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Random character appears between package name and function/macro/etc. name in the HTML title.

Eg. this is the title for function RcMalloc in rcheap.h:

  <title>ccdoc RcMalloc rcheap.h:C:function</title>
'C' should not be there.
Solution Description:
This problem occurred because the lineno was being appended to the output string used for the title as an integer rather than a string in phase3_html.cc at line 2169.

Title: Instantiations of class template methods are reported as global functions.
Status: CLOSED
Reported By: Joe Linoff
Reported On: 2001/08/24
Resolved By: Joe Linoff
Resolved On: 2001/08/24
Id: 0011
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In the following example, the instantiation of the template constructor is incorrectly reported as a global function.
    template <class T>
    class A {
    public:
      A();
    };

    // Instantiation is incorrectly reported as
    // a global function.
    template <class T>
    A<T>::A()
    {
    }
Solution Description:
This occurs because ccdoc attempts to do a name resolution to determine where the instantiation belongs so it tries to look up A<T> and can't find it.

The best way to work around this problem is to tell ccdoc to ignore the template method instantiations as shown below:

    template <class T>
    class A {
    public:
      A();
    };

    #if !defined(__ccdoc__)
    // Instantiation is incorrectly reported as
    // a global function.
    template <class T>
    A<T>::A()
    {
    }
    #endif
Meanwhile, I am looking at adding a feature to ccdoc to tell it to ignore template method instantiations (without ignoring template functions), see issue 0012.

Title: Ccdoc should ignore template method instantiations.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/08/24
Resolved By: Joe Linoff
Resolved On: 2003/02/17
Id: 0012
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
See issue 0011 for details of the problem. Ccdoc should have a switch that allows users to tell it to turn off (or on) the recognition of template method instantiations.
Solution Description:
Fixed in r33 and verified by test 69.

See issue 0149 for more details.


Title: Add ccdoc to source forge.
Status: FIXED
Reported By: Scott MacDiarmid
Reported On: 2001/08/24
Resolved By: Joe Linoff
Resolved On: 2002/01/02
Id: 0013
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I am sure you have been asked this before, but have you ever considered moving this project to SourceForge (http://www.sf.net)? It provides bug tracking, forums, and version control. I find it useful from both sides of the coin. Meaning as a developer and an end-user. As a developer it allows people to easy submit bugs to me, ask me question and it also provides CVS for good version control. As an end user I can follow what a project is doing very easily. The only catch is you would have to put it under an open source license. Anyway, just a thought.


I think that this is a great idea and have initiated the process. I will close this issue when the transfer is complete.
Solution Description:
I finally got around to it. See http://ccdoc.sourceforge.net.

Title: The typedef id for function pointer doesn't get extracted properly by ccdoc.
Status: FIXED
Reported By: Chi-Fung Fan
Reported On: 2001/08/24
Resolved By: Joe Linoff
Resolved On: 2001/08/24
Id: 0014
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The typedef id for function pointer doesn't get extracted properly by ccdoc. Example:
    /** Zero-argument const method type for Node */
    typedef Node* (Node:*NodeConstMeth) () const;
Ccdoc will generate an entry with 'const' instead of NodeConstMeth.


This is a serious problem, I will look into it as soon as possible.
Solution Description:
The problem was in phase1_parser.c. The parse_typedefs id heuristic did not how to deal with a trailing const keyword.

Title: Methods with throw clauses surrounded by parentheses don't get extracted properly by ccdoc.
Status: FIXED
Reported By: Chi-Fung Fan
Reported On: 2001/08/25
Resolved By: Joe Linoff
Resolved On: 2001/08/27
Id: 0015
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Methods with throw claused don't extracted properly by ccdoc. Example:
  void PlotDown(const doc::CC* pCC) const
    throw (RunTimeError);


I will fix this as soon as possible. Meanwhile, here is a workaround that you can use.
  void PlotDown(const doc::CC* pCC) const
#if !defined(__ccdoc__)
    throw (RunTimeError)
#endif
    ;
Solution Description:
Fixed in r5.

This problem occurred because the get_fct_id() heuristic in phase1_parser assumed that the trailing set of outermost parentheses defined the function argument list. I changed it look for the trailing set of outermost parentheses before a "throw" keyword.


Title: Source file names of the form ../folder/file.ext don't work properly in the cmd.exe shell under windows.
Status: FIXED
Reported By: Yuriy Stul
Reported On: 2001/08/26
Resolved By: Joe Linoff
Resolved On: 2001/08/28
Id: 0016
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Source file names of the form ../folder/file.exe don't work properly. The error message that it report is:
  The instruction at "0x00402330" referenced memory at "0x6e64658f".
  The memory could not be read.

I need your help. I cannot reproduce this bug. I have created half a dozen test cases and they all work.

Here are two examples of the command lines I was testing:

  ..\bin_opt_msvc_MSWin32-4.0\ccdoc -db ..\test\test.db -html ..\test\html\ ..\sources\s1.h ..\sources\s2.h

  AND

  ..\bin_opt_msvc_MSWin32-4.0\ccdoc -db ../test/test.db -html ../test/html/ ../sources/s1.h ../sources/s2.h
I tried multiple levels of directories. I tried different source file extensions and everything else I could think of.

All of this testing was done under Win2K using the MS-DOS command interpreter (cmd.exe).

Can you help figure out what am I doing wrong so that I can track down and fix this problem?

Solution Description:
Fixed in r8.

First, my thanks to Yuriy for helping me track this down.

This was a different manifestation of issue 0020. The fix was the same.


Title: Can't differentiate between different revisions of ccdoc 0.8.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/08/26
Resolved By: Joe Linoff
Resolved On: 2001/08/26
Id: 0017
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
It is difficult to distinguish between different versions of ccdoc 0.8.


I need to a -version switch to report the version/revision of the program.
Solution Description:
Fixed in r5.

Added the -version switch. Deprecated the -cid switch.


Title: Directory separators in -root names cause run time errors.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/27
Resolved By: Joe Linoff
Resolved On: 2001/08/27
Id: 0018
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Directory separators in -root names cause run time errors. If you use a directory separator in a -root name, it confuses ccdoc into thinking there is a directory when there isn't which causes a run-time error.
Solution Description:
Fixed in r6.

Added '/' and '\' filtering in ccdoc::phase3::html::format_name().


Title: The @see directive no longer accepts user defined links.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/27
Resolved By: Joe Linoff
Resolved On: 2001/08/27
Id: 0019
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The @see directive no longer accepts user defined links. In the previous version of ccdoc I could put in a direct link as follows:
@see <a href=mywebpage.html>Useful Stuff</a>


This is an oversight. It will be fixed immediately.
Solution Description:
Fixed in r7.

Modified the scanner to accept @see arguments that begin with a '<'. Modified the html generator to output these without interpretation.


Title: ccdoc crashes
Status: FIXED
Reported By: Chi-Fung Fan
Reported On: 2001/08/27
Resolved By: Joe Linoff
Resolved On: 2001/08/28
Id: 0020
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Both MSVC and cygwin versions of ccdoc crashed on generating docs for my program source. Attached is the debug messages before ccdoc dumped core.

I noticed that the offendng file that caused the core dump is the last file 'src/ScribDoc.h'. If I move it to be the first input file, everythinkg runs ok. This is also the workaround for me.

What information do you need to debug this problem?

ccdoc -v -db doc/webdocs/ccdoc.db -pkg Util include/mathpack/convolution.h include/mathpack/geometry.h include/ctl/iterator.h src/plot.h
db: read begins
db: does not exist
db: read ends
phase1: begins
phase1: parsing 'include/mathpack/convolution.h' ...
phase1: parsing 'include/mathpack/geometry.h' ...
phase1: parsing 'include/ctl/iterator.h' ...
phase1: parsing 'src/plot.h' ...
phase1: ends
db: write begins
db: writing 146 statements
db: write ends
ccdoc -v -db doc/webdocs/ccdoc.db -pkg Scribble src/navigator.h src/pager.h src/charannotvw.h src/drawtool.h src/presenter.h
db: read begins
db: read ends
phase1: begins
phase1: parsing 'src/navigator.h' ...
phase1: parsing 'src/pager.h' ...
phase1: parsing 'src/charannotvw.h' ...
phase1: parsing 'src/drawtool.h' ...
phase1: parsing 'src/presenter.h' ...
phase1: ends
db: write begins
db: writing 406 statements
db: write ends
ccdoc -v -db doc/webdocs/ccdoc.db -pkg Scribble src/definitions.h src/annottempl.h src/annotation.h src/chardefinitions.h src/dynamics.h
src/doctypes.h src/visitor.h src/docreader.h src/docwriter.h src/tablet.h src/glyph.h src/cc.h src/token.h src/page.h src/ScribDoc.h
db: read begins
db: read ends
phase1: begins
phase1: parsing 'src/definitions.h' ...
phase1: parsing 'src/annottempl.h' ...
phase1: parsing 'src/annotation.h' ...
phase1: parsing 'src/chardefinitions.h' ...
phase1: parsing 'src/dynamics.h' ...
phase1: parsing 'src/doctypes.h' ...
phase1: parsing 'src/visitor.h' ...
phase1: parsing 'src/docreader.h' ...
phase1: parsing 'src/docwriter.h' ...
phase1: parsing 'src/tablet.h' ...
phase1: parsing 'src/glyph.h' ...
phase1: parsing 'src/cc.h' ...
phase1: parsing 'src/token.h' ...
phase1: parsing 'src/page.h' ...
phase1: parsing 'src/ScribDoc.h' ...
phase1: ends
db: write begins
db: writing 1994 statements
db: write ends
      0 [main] ccdoc 94223783 open_stackdumpfile: Dumping stack trace to CCDOC.EXE.stackdump
/cygdrive/c/CHIFUNG/BIN/mkscribdocs.sh: line 63: 94223783 Segmentation fault      (core dumped) ccdoc -db doc/webdocs/ccdoc.db
root.txt
      0 [main] ccdoc 603763 open_stackdumpfile: Dumping stack trace to CCDOC.EXE.stackdump
/cygdrive/c/CHIFUNG/BIN/mkscribdocs.sh: line 64: 603763 Segmentation fault      (core dumped) ccdoc -db doc/webdocs/ccdoc.db -index
      0 [main] ccdoc 94223007 open_stackdumpfile: Dumping stack trace to CCDOC.EXE.stackdump
/cygdrive/c/CHIFUNG/BIN/mkscribdocs.sh: line 70: 94223007 Segmentation fault      (core dumped) ccdoc -db doc/webdocs/ccdoc.db
-html doc/webdocs/ -header root_hdr.txt -srcurl ../../ -root Scribble
Solution Description:
Fix in r8.

Dangling prefix comment statements at the end of db caused an iteration to extend past the end of a collection in database.cc around line 400.


Title: Report the version/revision information in the default trailer.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/08/28
Resolved By: Joe Linoff
Resolved On: 2001/08/28
Id: 0021
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Report the version/revision information in the default trailer. This makes error reporting easier.
Solution Description:
Fixed in r8.

Title: Centralize the version/revision information.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/08/28
Resolved By: Joe Linoff
Resolved On: 2001/08/28
Id: 0022
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Currently there are several references to "0.8" in phase3_html.c and in switches.cc. These should be centralized to make maintenance easier.
Solution Description:
Fixed in r8.

All references to version are now funneled through ccdoc::switches::version().


Title: DOC++ vs ccdoc
Status: CLOSED
Reported By: Bragas
Reported On: 2001/08/28
Resolved By: Joe Linoff
Resolved On: 2003/02/18
Id: 0023
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I'm wondering if you couldn't merge your project with DOC++. They look very similar and joining your efforts could help creating even better tool (best in its category ?).

I'm not DOC++ developer and this is just my private impression.


That is an interesting suggestion. I never really gave it much thought because there are a large number of documentation generation tools out there (see my other tools page) and I have no idea which ones are suitable for merging.

As far as I am concerned, the DOC++ developers (or anyone else for that matter) are free to take the ccdoc code and incorporate it into their system(s). I would be happy to support them in that effort. Furthermore, if they showed sufficient commitment to my goal of improving documentation in the C++ community for free, I would be happy to hand it over completely.

Solution Description:
Surveyed current ccdoc users and asked whether they wanted this project merged with Doxygen or DOC++. All (100%) of the respondents said no because they felt that the tools served different needs. Ccdoc was viewed as an interface documentation tool and doxygen was viewed as a source code documentation tool. In many cases, folks used both.

Title: Missing comma separator in @see output for the alternative URL form.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/08/28
Resolved By: Joe Linoff
Resolved On: 2001/08/28
Id: 0024
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The @see directive does not correct separate entries with commas if there an alternative form URL specified.
Solution Description:
Fixed in r9.

Forgot to output leading comma's for the alternative URL form of the @see directive.


Title: Undocumented 0.7a feature for explicitly specifying @pkgdoc URLs is no longer supported.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/28
Resolved By: Joe Linoff
Resolved On: 2001/08/28
Id: 0025
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In version 0.7a, ccdoc supported an undocumented syntax for the @pkgdoc directive that allowed the user to specify a custom URL. The syntax looked like this:
    /** @pkgdoc A.B ../A_B.html*/
Ccdoc would create the link using the standard package name and substitute in the specified link. We are using that feature extensively. Will it be supported in v0.8?


Yes. This is an oversight. All features that ccdoc can reasonably support will be supported.
Solution Description:
Fixed in r9.

Added support for the two argument form of @pkgdoc and documented it in the on-line help and in the users guide.


Title: Provide an expanded package contents view.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/29
Resolved By: Joe Linoff
Resolved On: 2001/08/29
Id: 0026
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
This would replace the tree view in the previous version. Basically it would expand the contents path hierarchically (the same way it is done in the class summary) with the type, scope and short description.


I think that this is a great idea. The switch name will be something like -[no]rpthpc (hierarchical package contents).
Solution Description:
Fixed in r10.

I implemented the -[no]rpthpc. It reports the hierarchical class structure of each package. I liked the format so much that I made it the default.

I also added the -[no]rptcsd to allow you to control whether details are reported on the class summary page (report class summary details).


Title: Provide a way to turn off package source.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/29
Resolved By: Joe Linoff
Resolved On: 2001/08/29
Id: 0027
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The source entry on the package page always points to a ccdoc db file. This is not useful to casual readers.


I agree. I will remove this from the next release.
Solution Description:
Fixed in r10.

Completely removed source statements for packages. The db reference does not make sense for users.


Title: Provide a way to turn off the default author on the package page.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/29
Resolved By: Joe Linoff
Resolved On: 2001/08/29
Id: 0028
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Reporting unascribed as the default package author when the author is not specified is not really useful in our environment. Can I turn this default handling off?


Not right now, I will add a switch called -[no]rptdpa (default package author) to give you that control.
Solution Description:
Fixed in r10.

Added the -[no]rptdpa switches. The default is -norptdpa.


Title: Provide a way to turn off the default version on the package page.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/29
Resolved By: Joe Linoff
Resolved On: 2001/08/29
Id: 0029
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Reporting unknown as the default package version when the version is not specified is not really useful in our environment. Can I turn this default handling off?


Not right now, I will add a switch called -[no]rptdpv (default package version) to give you that control.
Solution Description:
Fixed in r10.

Added the -[no]rptdpa switches. The default is -norptdpv.


Title: Provide a way to turn off default description comments on the page.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/29
Resolved By: Joe Linoff
Resolved On: 2001/08/29
Id: 0030
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Reporting undocumented as the default package comment when the comment is not specified is not really useful in our environment. Can I turn this default handling off?


Not right now, I will add a switch called -[no]rptdpd (default package comment) to give you that control.
Solution Description:
Fixed in r10.

Added the -[no]rptdpd switches, made -norptdpd the default.


Title: Provide a way to change the "package" string in the title on package pages.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/29
Resolved By: Joe Linoff
Resolved On: 2001/08/29
Id: 0031
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In our environment packages have different semantics, some of them are subsystems, others are packages and still others are layers. I would like to be able to change the title on the package page to reflect this.


This is a good idea. I will add a @pkgdoctid directive that allows you to specify the title tag that ccdoc writes out. This was not possible in earlier versions because they use GIF images but in the new version it is straightforward.
Solution Description:
Fixed in r10.

Added the @pkgdoctid directive. See the on-line help or the users guide for additional details.


Title: Users guide and on-help report than -norptpub is the default.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/08/29
Resolved By: Joe Linoff
Resolved On: 2001/08/29
Id: 0032
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Minor documentation nit, help and the users guide report that -norptpub is the default when the real default is -rptpub.
Solution Description:
Fixed in r10.

Title: This construct is incorrectly recognized as a struct: void foo(struct bar* spam).
Status: FIXED
Reported By: Wawszczak, Jarosław
Reported On: 2001/08/30
Resolved By: Joe Linoff
Resolved On: 2001/08/30
Id: 0033
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
This construct is incorrectly recognized as a struct: void foo(struct bar* spam).
Solution Description:
Fixed in r10.

This occurred because the parsing heuristic didn't recognize the the struct declaration was inside an argument list.


Title: For -rpthpc only show packages.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/09/02
Resolved By: Joe Linoff
Resolved On: 2001/09/04
Id: 0034
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I have one more request.... I'd like the 'hierarchical contents' to show 'packages only', and not classes, namespaces, etc. Basically, I want to limit the depth to packages. The 'classes' link can still be the full glory. Without this, the current contents with classes is too detailed for the top-level index. With my mod, it will be a direct replacement of the old 'tree'.


I think that this is a good idea. I will release it in r11. Unfortunately, I won't be able to work on it until next week.
Solution Description:
Fixed in r11.

Title: ccdoc 0.7a to 08 quick reference
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/09/02
Resolved By: 2001/09/04
Resolved On: Joe Linoff
Id: 0035
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I think that it would be useful to have a summary of the changes between 0.7a and 0.8 for users that are familiar with 0.7a.


I think that this is a good idea. I will add something as soon as I can.
Solution Description:
Fixed in r11.

Added a quick reference guide at the top level.


Title: Comments get lost.
Status: FIXED
Reported By: Yuriy Stul
Reported On: 2001/09/03
Resolved By: Joe Linoff
Resolved On: 2001/09/04
Id: 0036
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Comments get lost under the following conditions:
  /**
   * This is a dangling comment at the head of a file.
   */
  /**
   * This is a class comment that gets lost.
   */
  class A {};
You can work around this problem by changing the initial comment to a non-ccdoc style comment as follows:
  /*
   * This is a dangling comment at the head of a file.
   */
  /**
   * This is a class comment that gets lost.
   */
  class A {};
Solution Description:
Fixed in r12.

This problem occurred because the database reader assumed that comments could not exist in sequential records. The fix was in database.cc around line 399.


Title: Change the asterisk column header to "Inherited From".
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/09/05
Resolved By: Joe Linoff
Resolved On: 2001/09/05
Id: 0037
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Change the asterisk column header to "Inherited From". The asterisk is not very meaningful.
Solution Description:
Fixed in r13.

Title: Change the Path column header to "Entity".
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/09/05
Resolved By: Joe Linoff
Resolved On: 2001/09/05
Id: 0038
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Change the asterisk column header to "Inherited From". The asterisk is not very meaningful.
Solution Description:
Fixed in r13.

Title: Performance problem in phase3.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/09/05
Resolved By: Joe Linoff
Resolved On: 2001/09/07
Id: 0039
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The new version of ccdoc seems to generate HTML files much slower than v0.7a. In our tests it was running about 10X slower.
Solution Description:
Fixed in r14.

Found an O(N^2) complexity algorithm in databases::get_stmt_no_pkgs(). Converted the algorithm to O( n log(n) ). The test case performance improved from 49 minutes to 30 seconds but the real case is still slow.

Found another O(N^2) complexity algorithm in databases::get_stmt_no_pkgs() related to tree traversal for id lookups. Converted the algorithm to O( n log(n) ). The real test case performance improved from 18 hours to 12 minutes.

Made the generated HTML slightly more compact. The resulting output in the browers is identical.

Summary of performance improvements:

Summary of other miscellanous changes:


Title: Protected inherited methods reported.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/09/05
Resolved By: Joe Linoff
Resolved On: 2001/09/06
Id: 0040
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In some cases, it looks like protected inherited methods are reported.
Solution Description:
Fixed in r13.

The make_class_contents() method in phase3_html.cc was not verifying that the inherited methods were accessible.


Title: Improve 0039 performance enhancement.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/09/07
Resolved By: Joe Linoff
Resolved On: 2001/09/08
Id: 0041
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The following changes would improve the issue 0039 performance fix even more:

Solution Description:
Fixed in r15.

Title: ccdoc bug notification - not correctly determining variable name
Status: FIXED
Reported By: Paul Kohler
Reported On: 2001/09/10
Resolved By: Joe Linoff
Resolved On: 2001/09/10
Id: 0042
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Version: ccdoc v0.8 r14 2001/09/07 bin_opt_msvc_MSWin32-4.0
Shell: cmd.exe
Platform: Windows 2000

It seems to be reporting a link and link text in the contents of a package as the value (it is a static global variable). There are two and the comments are generated correctly, but the link goes to the first one.

Here is the snippet of code with comments:

typedef int   Bool;
#define true  1
#define false 0

/**
 * Global boolean value to signal if an account is loaded.
 *
 * @author  Paul Kohler
 */
static Bool g_bAccountInit = false;

/**
 * Global boolean value to signal if the connection has been established.
 *
 * @author  Paul Kohler
 * @see IConnect
 */
static Bool g_bConnected = false;
And this is the resulting html snippit:
false(href="ccdoc.lib.libivr.false.var.html")   variable   public   Global
boolean value to signal if the ICSS connection has been established.
false(href="ccdoc.lib.libivr.false.var.html")   variable   public   Global
boolean value to signal if an account is loaded.


Excellent report! I was able to duplicate the problem in a few minutes and am currently working to fix it. My hope is that the fix will be released in r15 as early as today.
Solution Description:
Fixed in r15.

The variable parsing heuristic was not correctly recognizing '=' as an id separator in some cases. Look for "Issue 0042" in a comment in phase1_parser.cc.


Title: Old hash mark syntax for @see does not work.
Status: FIXED
Reported By: Lou Sanchez-Chopitea
Reported On: 2001/09/10
Resolved By: Joe Linoff
Resolved On: 2001/09/10
Id: 0043
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The old @see #foo syntax no longer works.

This was an oversight. I inadvertently deprecated it. It will be fixed in the next release.

Solution Description:
Fixed in r15.

Title: The summary report could use a little more indenting.
Status: FIXED
Reported By: Lou Sanchez-Chopitea
Reported On: 2001/09/10
Resolved By: Joe Linoff
Resolved On: 2001/09/10
Id: 0044
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
It would be nice to be able to control the indenting level for reporting hierarchical objects on the summary page..

Perhaps I could add a switch that would allow you to control the level of indenting.

Solution Description:
Fixed in r15.

Added the -rptcsi (Class Summary Indent) switch.


Title: The variable width font in the Code: section is not as readable as the fixed width font used in 0.7a.
Status: FIXED
Reported By: Lou Sanchez-Chopitea
Reported On: 2001/09/10
Resolved By: Joe Linoff
Resolved On: 2001/09/11
Id: 0045
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The variable width font in the Code: section is not as readable as the fixed width font used in 0.7a.
The mono-space font was removed because of a general consensus that the variable width font was easier to read as demonstrated by Bjarne Stroustroup changing the font for all C++ examples in his recent books.

I have no opinion one way or the other.

I will add a switch that allows the user to control this. of indenting.

Solution Description:
Fixed in r17.

Added the -rptfwcf (Fixed Width Code Font) switch.

I originally reported this as fixed in r15 but the generate HTML still seemed to be proportional so I changed the generated output.


Title: The @see index doesn't seem to work.
Status: FIXED
Reported By: Chris Meuzelaar
Reported On: 2001/09/10
Resolved By: Joe Linoff
Resolved On: 2001/09/10
Id: 0046
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
ccdoc fails to generate links for @see directives when generating documentation for a large number of files / classes when multiple instances of the function exist in a class (see below). The sample below will generate HTML with the link back to the first function CSampleTray::AddSampleAt 0, but not the other two. Attached are the batch files used to generate the documentation as well as the database file. I am generating documentation using ~500 files.

Here is a bit of sample code that doesn't work.

class A {
public:
  /**
   * Do stuff.
   * @see A::B 1
   * @see A::B 2
   */
  void B();
  void B(int);
  void B(double);
};
Solution Description:
Fixed in r16.

I thought that this bug was important enough to warrant a re-release.


Title: The indenting on the contents page should be the same as the summary report.
Status: FIXED
Reported By: Lou Sanchez-Chopitea
Reported On: 2001/09/11
Resolved By: Joe Linoff
Resolved On: 2001/09/11
Id: 0047
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The indenting on the contents page should be the same as the summary report.
Solution Description:
Fixed in r17.

Modified the -rptcsi (Class Summary Indent) switch to also change the contents indenting.


Title: Derived classes in namespaces are not recognized correctly.
Status: FIXED
Reported By: Dmitry A.Steklenev
Reported On: 2001/09/11
Resolved By: Joe Linoff
Resolved On: 2001/09/11
Id: 0048
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In the following example, the HTML for class_two does not correctly link to class_one.
namespace ccdoc_test {
  class class_one {
  };
  class class_two : public class_one {
  };
}
Solution Description:
Fixed in r17.

Partially scoped names were not being expanded in the extends clause processing in phase3_html.cc.


Title: Non-template class method instantiation not recognized correctly.
Status: FIXED
Reported By: Dmitry A.Steklenev
Reported On: 2001/09/11
Resolved By: Joe Linoff
Resolved On: 2003/02/17
Id: 0049
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In the following example, some_method() is reported as a separate function.
namespace ccdoc_test {
  class class_one {
  public:
    int some_method();
  };
  inline int class_one::some_method()
  {
    return 0;
  }
}

This is not quite as serious as the previous bug because it can be worked around as follows:
namespace ccdoc_test {
  class class_one {
  public:
    int some_method();
  };
  #if !defined( __ccdoc__ )
  inline int class_one::some_method()
  {
    return 0;
  }
  #endif
}
Solution Description:
Fixed in r33 and verified by test 69.

This was fixed as part of the issue 0149 work.


Title: Ccdoc fails to parse methods with fct args properly.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/09/11
Resolved By: Joe Linoff
Resolved On: 2001/09/11
Id: 0050
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The following example is reported incorrectly as a variable of name fp:
class A {
public:
  int fct2( void ( * fp ) ( ) ) ;
};
Solution Description:
Fixed in r17.

This bug resulted from the phase_parser not being able to correctly identify methods under certain conditions.


Title: External class method implementations with default arguments are not correctly recognized.
Status: FIXED
Reported By: Lou Sanchez-Chopitea
Reported On: 2001/09/17
Resolved By: Joe Linoff
Resolved On: 2003/02/17
Id: 0051
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
External class method implementations with default arguments are not correctly recognized. Here is an example:
class A {
public:
  void fct(int x=0);
};
void A::fct(int x)
{
}
Ccdoc treats the implementation as a separate function.
This occurs because ccdoc does not understand default arguments and thinks that the =0 makes two different argument sequences. You can workaround this in two different ways, both are shown below:
Workaround #1 Workaround #2
class A {
public:
  void fct(int x=0);
};
#if !defined(__ccdoc__)
void A::fct(int x)
{
}
#endif
class A {
public:
  void fct(int x /**@#-*/ =0 /**@#+*/ );
};

void A::fct(int x)
{
}
Because there are workarounds, this a low priority bug.
Solution Description:
Fixed in r33 and verified by test 69.

This was fixed as part of the issue 0149 work.


Title: Backslash character definition ('\\') confuses the ccdoc scanner.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/09/18
Resolved By: Joe Linoff
Resolved On: 2001/09/18
Id: 0052
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The following definition confuses the ccdoc parser.
const char X = '\\';

This is a serious bug that will be fixed as soon as possible.
Solution Description:
Fixed in r18.

There was a scanner problem for detecting the ends of character constants in some cases.


Title: No line numbers for internal comment parsing warnings.
Status: FIXED
Reported By: Ray Rizzuto
Reported On: 2001/09/18
Resolved By: Joe Linoff
Resolved On: 2001/09/21
Id: 0053
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Error during parse of header file with a package comment:
WARNING: Internal comment parsing error.
 Unexpected EOF for ''.
 This comment will be ignored.
WARNING: Internal comment parsing error.
 Unexpected EOF for ''.
 This comment will be ignored.
Solution Description:
Fixed in r19.

Title: Serious problem with typedef of struct
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/09/20
Resolved By: Joe Linoff
Resolved On: 2001/09/21
Id: 0054
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The following struct typedef conduses ccdoc.
typedef struct _TEST {
    int a;
    int b;
} TEST1;
It generates the following documentation:
package ccdocTest

Contents

    Entity    Type       Scope    Short Description
    a         typedef    public   undocumented
    b         variable   public   undocumented

This is definitely a bug. When I revisited the test suite I noticed that there were no cases for struct and class typedefs.

You might be able to workaround this problem as follows:

  struct _TEST {
    int a;
    int b;
  };
  typedef _TEST TEST1;
Solution Description:
Fixed in r19.

The ccdoc::phase1::parser::parse_typedef() method did not correctly recognize embedded structures.


Title: ccdoc generates default special members for C style structs
Status: CLOSED
Reported By: Dékány Dániel
Reported On: 2001/09/20
Resolved By: Joe Linoff
Resolved On: 2001/09/20
Id: 0055
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc should never create default special members (constructor, destructor, operator=) for structs in C programs, since there is no such thing in C.
This is true. Unfortunately, ccdoc does not officially support C.

Perhaps this problem can be solved in another way by allowing you to specify -nocdsm in phase1 for certain files.

Solution Description:
Client has a workaround. I will re-open this issue if I get additional requests.

Title: Incorrect line numbers reported in warning messages.
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/09/20
Resolved By: Joe Linoff
Resolved On: 2003/02/18
Id: 0056
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Try this:
---[ccdoctest.h]---
/** @pkg ccdocTest */

int x;
Ccdoc will say, that x is in ccdoctest.h:4, but it is in ccdoctest.h:3. It seems to me, that the extra 1 line number is related somehow to the ccdoc comment.
Solution Description:
Fixed in r34, verified by test 13.

This was extremely tricky because the line numbers were correct most of the time so it was hard to find a test case. The only time they were incorrect were when the first line of the file had a ccdoc directive that required a push/pop of the tokens that included the new line.


Title: "Undocumented" as default short desc. in tables is confusing.
Status: CLOSED
Reported By: Dékány Dániel
Reported On: 2001/09/20
Resolved By: Joe Linoff
Resolved On: 2001/09/21
Id: 0057
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
"Undocumented", the default short description can be confusing in HTML tables when most entries have a short description, since "Undocumented" entries can be easily overlooked. I think the solution is to give switch to override this text, then I can replace it with "-".
Solution Description:
Ccdoc already supports this. The resolution is described in detail in issue 0058.

Title: Provide a way to turn off the default author and version for packages.
Status: CLOSED
Reported By: Dékány Dániel
Reported On: 2001/09/20
Resolved By: Joe Linoff
Resolved On: 2001/09/21
Id: 0058
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Reporting unascribed as the default package author and reporting version as unknown is not so useful in some projects, since we do it on package level. Perhaps -[no]rptdnpa and -[no]rptdnpd, where np is for non-package entity?
Solution Description:
Ccdoc provides a mechanism for turning off the default author and version for any package by allowing the client to specify them explicitly. The technique for doing this is to create a text file with entries for each property. That file could look something like this:
// Create empty entries for packages to avoid using the defaults.
/**
 * &nbsp;
 * @author &nbsp;
 * @version &nbsp;
 * @pkgdoc PKG1
 */
/**
 * &nbsp;
 * @author &nbsp;
 * @version &nbsp;
 * @pkgdoc PKG1::SUBPKG1
 */
/**
 * &nbsp;
 * @author &nbsp;
 * @version &nbsp;
 * @pkgdoc PKG1::SUBPKG2
 */
This file would be added to the phase 1 flow as follows:
ccdoc -db $CCDOC_DATABASE pkgdoc.txt

Title: Validate ccdoc generated HTML.
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/09/20
Resolved By: Joe Linoff
Resolved On: 2001/09/20
Id: 0059
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
You can use the verifier at http://validator.w3.org to find errors in ccdoc generated output.
Thank you for a great suggestion! I never knew that an HTML verifier existed.
Solution Description:
Modified the ccdoc output to conform to the HTML 4.01 Transitional specification.

Title: Group entries by type.
Status: OPEN
Reported By: Dékány Dániel
Reported On: 2001/09/20
Resolved By:
Resolved On:
Id: 0060
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In many situations it's very bad, that all entities are droped into one big list, you know, under the "Contents" tite, like this:
CONTENTS

Entry    Type Scope Shor Description
b    function ...
c    macro ...
d    function ...
e    package ...
f    macro ...
This would be much better in many cases:
CONTENTS


Packages

Entry    Type Scope Short Description
e    package ...


Functions

Entry    Type Scope Short Description
b    function ...
d    function ...


Macros

Entry    Type Scope Short Description
c    macro ...
f    macro ...

A switch like -[no]grpbytype? I don't hink so, cos it's not enough generic (E.g. maybe an alphabetical index is required too). I think the real solution is to allow user to specify any number of tables, with custom grouping and title. Something like this:

$ ccdoc ...  -table(type) "Contents" -table() "Index" ...
(Note that I have skipped unimportant parts with ellipsis)
It would generate two tables: one with title "Contents" and with grouping by entity type, and below another table with title "Index" without grouping (just an alphabetiacal index, like the table in the current version). I put the grouping condition into parenthesis to allow multi-level grouping for future developmens, like -contents(type,functionality).
This is an interesting suggestion. The previous version of ccdoc did use to group by entity. I will revisit that decision.
Solution Description:

Title: Typedef struct statement does not include struct comments.
Status: OPEN
Reported By: Dékány Dániel
Reported On: 2001/09/23
Resolved By:
Resolved On:
Id: 0061
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The fix for issue 0054 only recognizes ccdoc style comments for the typedef, not for the structure.
This occurs because ccdoc only associates comments with statements.

In this case there is only one statement but it is clear that it could be broken into the struct definition and the typedef. It might be possible to modify ccdoc with a heuristic that recognizes this construct and breaks it into two.

This is a low priority enhancement because this syntax is C artifact that is not necessary in C++ because the struct [name] declares [name] as a type.

Solution Description:

Title: Long entry name deforms HTML table.
Status: OPEN
Reported By: Dékány Dániel
Reported On: 2001/09/23
Resolved By:
Resolved On:
Id: 0062
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Just an idea: Long entry name stretches the first column too much. In the worst case it makes the "Contents" table unusable, since it stretches the first column so much that there will be no sufficient space for the short description.

A possible solution is to give a switch to specify the maximum entry name length, then break long names into two (or three, etc...) with an "\\<br>\\" in the HTML (to be more precise, with '\\</a><br>' + indent*'&nbsp;' + '<a href="' + relpath + '">\\' (two backslash in the HTML output, not C escaping of one backslash...))

Solution Description:

Title: Allow user defined defaults for author, short description and version.
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/09/23
Resolved By: Joe Linoff
Resolved On: 2001/09/24
Id: 0063
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Reporting unascribed as the default package author and reporting version as unknown is not so useful in some projects, since we do it on package level. Perhaps -[no]rptdnpa and -[no]rptdnpd, where np is for non-package entity?
This is what issues 0058 and 0059 were originally supposed to address. I misinterpreted the request and addressed the wrong problem.

I believe that the correct way to address the problem is to allow the user to specify their own default strings for author, version and short description types using new switches: -rptdefa, -rptdefsd and rptdefv.

Solution Description:
Fixed in r20.

Added:


Title: Friend classes in namespaces are not recognized correctly.
Status: FIXED
Reported By: Dmitry A.Steklenev
Reported On: 2001/09/25
Resolved By: Joe Linoff
Resolved On: 2001/09/26
Id: 0064
[Top]
[Summary]
[Next]
[Previous]
Problem Description:

Code Sample:

namespace nms  {

    class some_class
    {
       public:
         void some_class();
    };

    class some_other_class
    {
       /** Red, but not linked */
       friend class some_class;
       public:
         void some_other_class();
    };
}

Ccdoc does not resolve scoped identifiers (because it does not process included header files that may contain "using" statements). This means that the user must be explicitly specify the full scope of an identifier.

In this particular case, I can modify ccdoc to try looking in the parents namespace it the id is not found.

Solution Description:
Fixed in r21.

Modified ccdoc:phase3::html::write_friends_info to look in the namespace of the parent to resolve unresolved names. Unfortunately this will not work for all cases. The best solution is to always fully qualify scoped names.


Title: Links from documentation in namespaces are not recognized correctly.
Status: CLOSED
Reported By: Dmitry A.Steklenev
Reported On: 2001/09/25
Resolved By: Joe Linoff
Resolved On: 2001/09/25
Id: 0065
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Code Sample:
namespace nms  {

  class some_class
  {
     public:
       void some_class();
  };

  /** 
   * Some description
   * 
   * Red, but not linked:
   * @$ some_class
   * and "see also" have same problem.
   * @see some_class
   */
  
  class some_other_class
  {
     public:
       void some_other_class();
  };
}
Solution Description:
Ccdoc does not resolve scoped identifiers (because it does not process included header files that may contain "using" statements). This means that the user must be explicitly specify the full scope of an identifier as shown below.
namespace nms  {

  class some_class
  {
     public:
       void some_class();
  };

  /** 
   * Some description
   * 
   * Red, but not linked:
   * @$ nms::some_class
   * and "see also" have same problem.
   * @see nms::some_class
   */
  
  class some_other_class
  {
     public:
       void some_other_class();
  };
}

I added fix that will resolve partially scoped references in most cases. See issue 0118 for details.


Title: The <sstream> header is not needed in phase3_html.cc.
Status: FIXED
Reported By: Dmitry A.Steklenev
Reported On: 2001/09/25
Resolved By: Joe Linoff
Resolved On: 2001/09/27
Id: 0066
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The <sstream> header is not needed in phase3_html.cc.
Solution Description:
Fixed in r21.r

You are absolutely correct. The <sstream> header is not needed.

Earlier I was using stringstreams but had a porting problem and had to remove them. When I did that I neglected to remove the header file.

Thanks for pointing it out.


Title: Ccdoc does not correctly reset the access specifiers for methods in base classes that are derived as protected or private.
Status: OPEN
Reported By: Dmitry A.Steklenev
Reported On: 2001/10/01
Resolved By:
Resolved On:
Id: 0067
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Code Sample:
  class some_base_class
  {
     public:
       /** Must be reported as "protected" in some_class_01 and
         * as "private" in some_class_02, but reported as "public" */
       void some_public_method();
     protected:
       /** Must be reported as "protected" in some_class_01 and
         * as "private" in some_class_02, but reported as "protected" */
       void some_protected_method();
  };

  class some_class_01 : protected some_base_class {};
  class some_class_02 : private some_base_class   {};

Ccdoc always reports methods as they appear in the original class specification and expects the user to understand how the access-specification changes if protected or private inheritance is used.

This may be difficult for many users. I will look into changing this behavior or making it optional via a switch. Meanwhile, if you don't want the users to see the methods, you can work around this as follows:

  class some_base_class
  {
     public:
       /** Must be reported as "protected" in some_class_01 and
         * as "private" in some_class_02, but reported as "public" */
       void some_public_method();
     protected:
       /** Must be reported as "protected" in some_class_01 and
         * as "private" in some_class_02, but reported as "protected" */
       void some_protected_method();
  };

  class some_class_01
#if !defined(__ccdoc__)
   : protected some_base_class
#endif
    {};
  class some_class_02
#if !defined(__ccdoc__)
   : private some_base_class
#endif
    {};
Solution Description:

Title: Core dump on solaris in the index phase.
Status: FIXED
Reported By: John Rostykus
Reported On: 2001/10/01
Resolved By: Joe Linoff
Resolved On: 2001/10/01
Id: 0068
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Under certain conditions a core dump is generated during the index phase for ccdoc r20.
Solution Description:
Fixed in r21.

This problem occurred because phase2 was deleting statements without removing them from the load map in the database object. This caused a bad pointer to be referenced under certain conditions. This bad pointer caused the core dump on solaris.


Title: Cannot handle construct: "Float operator/(const Float & rval)"
Status: FIXED
Reported By: Gustavsson Mĺrten
Reported On: 2001/10/04
Resolved By: Joe Linoff
Resolved On: 2001/10/08
Id: 0069
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc cannot handle the construct:
Float operator/(const Float & rval)
You can work around this problem by inserting a space after the forward slash as follows:
Float operator/ (const Float & rval)
Solution Description:
Fixed in r22.

There was a bug in ccdoc::phase1::scanner::scan_token() for processing forward slash tokens. When the scanner looked ahead and failed, it did not reset the next character properly. The problem was easily fixed by resetting the next character properly. This occurred around line 506 in phase1_scanner.cc 1.8. Test 29 was added to verify the fix.


Title: Users guide documentation error.
Status: FIXED
Reported By: Gustavsson Mĺrten
Reported On: 2001/10/04
Resolved By: Joe Linoff
Resolved On: 2001/10/08
Id: 0070
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
By the way, the following section in the user documentation contains an error:

Directive: /**@#-*/, /**@#+*/

Turns ccdoc token processing off and on. They are useful for turn off processing for tokens on a line. For whole lines, use the __ccdoc__ compiler pragma as shown in the example below:
// Turn off sections of code using __ccdoc__ 
#if defined(__ccdoc__) 
  // Ignore stuff until the end 
#endif 
It should be:
#if !defined(__ccdoc__) 
Solution Description:
Fixed in r22.

Title: Operator keyword duplicated for method operators.
Status: FIXED
Reported By: Lou Sanchez-Chopitea
Reported On: 2001/10/08
Resolved By: Joe Linoff
Resolved On: 2001/10/08
Id: 0071
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Operator keyword duplicated for method operators.
Solution Description:
.

There was a bug in ccdoc::phase3::html::write_section_header for recognizing method operators for creating titles. It was not checking for statement::base::STMT_METHOD_OPERATOR when creating the title.

This is verified by test19.


Title: New switch to turn off sorting of class entities.
Status: FIXED
Reported By: Michael Wang
Reported On: 2001/10/11
Resolved By: Joe Linoff
Resolved On: 2001/10/12
Id: 0072
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I would like to see a switch that allows the user to turn off output ordering/sorting so that the listing of methods/functions can be dictated by the programmer.
After some additional communication the specific functionality was refined to: add a new switch that turns off the sorting of class entities. Specifically it refers to the methods, variables, enumerations and typedefs inside of a class.

In keeping with the current naming convention this switch should be called something like -rptdsfce (disable sorting for class entities).

Solution Description:
Fixed in r22.

Added two new switches: -rptsci and -norptsci to allow the user to control the sorting of class information. The default is -rptsci. If -norptcsi is specified the class contents and details are not sorted.

The idea is that programmers may wish to logically organize class information in a way that is more understandable than a simple sort.


Title: Can't tell ccdoc to ignore a privately declared default constructor.
Status: FIXED
Reported By: Lou Sanchez-Chopitea
Reported On: 2001/10/11
Resolved By: Joe Linoff
Resolved On: 2001/10/12
Id: 0073
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
When I create a class with a private default constructor ccdoc still reports it. Here is an example:
class A {
private:
  A(); // ccdoc always reports this
public:
  A(int);
};

I would like to see a directive that allows the user to turn off output generation for a specific entity. This would allow me to tell ccdoc to ignore the reporting of private default constructors on a per class basis.


I don't think that a new directive is needed to solve this problem. In fact, I think it is a simple bug.

Ccdoc must be fixed to not report private entities (even those generated by default) unless the -rptpri switch is specified.

Solution Description:
Fixed in r22.

Ccdoc will not report the methods and operators generated by default if they are privately declared.


Title: Ccdoc cannot handle international character sets.
Status: FIXED
Reported By: Dmitry A.Steklenev
Reported On: 2001/10/12
Resolved By: Joe Linoff
Resolved On: 2001/10/12
Id: 0074
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
CCDoc inserts a generated html meta variable:
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
Please, provide any command line arguments for switching to user specified charset.
Thank you for pointing this out. The charset was added for compliance with the HTML 4.01 Transitional specification (see issue 0059).

I will add a switch that allows you to define your own character set in the next release (probably on Monday).

Solution Description:
Fixed in r22.

Added a new switch called -rptctcs that allows users to specify the character set that they wish to use.


Title: Ccdoc cannot handle function typedef of the form: typedef void fct(int arg1);
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/10/13
Resolved By: Joe Linoff
Resolved On: 2001/10/14
Id: 0075
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc cannot handle function typedef of the form: typedef void fct(int arg1);. It reports arg1 as the typedef rather than fct.
Solution Description:
Fixed in r23. Verified by test33.

The phase1_parser did not understand this syntax correctly.


Title: Conflict between -meta switch and ccdoc defined meta variables.
Status: FIXED
Reported By: Dékány Dániel
Reported On: 2001/10/14
Resolved By: Joe Linoff
Resolved On: 2001/10/16
Id: 0076
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
There is a conflict between the -meta switch and the fact that ccdoc automatically creates some meta-variables for HTML 4.01 compliance because it does not allow users to override or change the ccdoc generated meta variables.

If user gives a -meta switch, ccdoc must *not* generate standard meta elements, like http-equiv-s or keyword-s. Generating ccdoc_... metas is OK.


I completely agree. This is a well thought out criticism of the current behavior. I will modify r23 to not generate standard meta elements if the -meta switch is specified.
Solution Description:
Fixed in r23. Verified by test34.

Also added a warning if both -meta and -rptctcs are specified.


Title: Ccdoc incorrectly identifies a class attribute name.
Status: FIXED
Reported By: Clint Bauer
Reported On: 2001/10/18
Resolved By: Joe Linoff
Resolved On: 2001/10/19
Id: 0077
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc identifies the attribute name as BAR instead of m_foo:
class A {
public:
  char m_foo[BAR];
};
Solution Description:
Fixed in r23. Verified by test35.

The variable id recognizer in ccdoc::phase1::parser::parse_var_or_fct (phase1_parser.c) was confused by the square brackets.


Title: Ccdoc incorrectly identifies a variable as a function.
Status: FIXED
Reported By: Clint Bauer
Reported On: 2001/10/18
Resolved By: Joe Linoff
Resolved On: 2001/10/19
Id: 0078
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc incorrectly identifies a variable as a function in the following case:
const long is_not_a_function = (1+1);
If the parentheses are removed, ccdoc correctly identifies it as a variable.
Solution Description:
Fixed in r23. Verified by test36.

The function type recognizer in ccdoc::phase1::parser::parse_var_or_fct (phase1_parser.c) was confused by the presence of parentheses after an equals sign.


Title: platform.pl cannot handle embedded parentheses.
Status: FIXED
Reported By: Ben Zoe
Reported On: 2001/10/23
Resolved By: Ben Zoe
Resolved On: 2001/10/24
Id: 0079
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I use cygwin and it attaches a trailer build number in parentheses which is an illegal path name.
Solution Description:
Fixed in r24.

Ben added a line to filter out parentheses to platform.pl.


Title: Trim trailing spaces in path name for get_stmt_no_pkgs().
Status: FIXED
Reported By: Ben Zoe
Reported On: 2001/10/23
Resolved By: Ben Zoe
Resolved On: 2001/10/24
Id: 0080
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
If the path name has trailing spaces, the get_stmt_no_pkgs() may fail unexpectedly.
Solution Description:
Fixed in r24. Verified by test #38.

Ben added code that will trim trailing white space. I modified it slightly to include the \r white space character for PCs.

It looks like this problem can be tickled by @links with trailing spaces.


Title: Multi-space blank lines between SHORT and LONG descriptions cause them to merge.
Status: CLOSED
Reported By: Ben Zoe
Reported On: 2001/10/23
Resolved By: Joe Linoff
Resolved On: 2001/10/24
Id: 0081
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Multi-space blank lines between SHORT and LONG descriptions cause them to merge.
Solution Description:
The current ccdoc behavior is correct. Blanks are used to indicates lines in HTML <pre> clauses. The fix that was described (testing for lines that contain w/s) would fail to properly parse the following case:
/**
 * This is a short description with some code in it:
 * 
 * // Whitespace test
 * 
 * 
* * This is the start of the long description. */

I am not sure that this resolution is correct because the ccdoc short description does not conform to the javadoc specification. See issue 0089 for details.


Title: Add the @since directive per the javadoc 1.2 specification.
Status: FIXED
Reported By: Ben Zoe
Reported On: 2001/10/23
Resolved By: Ben Zoe & Joe Linoff
Resolved On: 2001/10/25
Id: 0082
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Add the @since directive per the javadoc specification.
Solution Description:
Fixed in r24. Verified by test #39.

This change invalidates old ccdoc db files.


Title: Modify ccdoc to create a separate links page (as was done in v0.7a).
Status: FIXED
Reported By: Ben Zoe
Reported On: 2001/10/23
Resolved By: Joe Linoff
Resolved On: 2001/10/26
Id: 0083
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
When ccdoc see's multiple links to the same entities it creates a list. In ccdoc v0.7a it created a separate page with the links. I like that approach better.
That is an intriguing idea. I will look into it in the near future.
Solution Description:
Fixed in r24. Verified by test #43.

Modified ccdoc to handle multiple links on a separate page. It is much easier to read than the previous version. The main work was done in the ccdoc::phase3::html::new write_links() method.


Title: Alias @exception to @throws per javadoc 1.2.
Status: FIXED
Reported By: Ben Zoe
Reported On: 2001/10/23
Resolved By: Ben Zoe & Joe Linoff
Resolved On: 2001/10/25
Id: 0084
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Alias @exception to @throws per javadoc 1.2.
Solution Description:
Fixed in r24. Verified by test #39.

Title: Ccdoc not recognize default constructors with member initialization.
Status: FIXED
Reported By: Ben Zoe
Reported On: 2001/10/23
Resolved By: Joe Linoff
Resolved On: 2001/10/25
Id: 0085
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc not recognize default constructors with member initialization as shown in the following example:
class A {
public:
  A() : x(0), y(0) {}
private:
  int x,y;
};
Solution Description:
Fixed in r24. Verified by test #42.

Title: Comment form to handle single line suffix comments.
Status: FIXED
Reported By: Ben Zoe
Reported On: 2001/10/24
Resolved By: Ben Zoe
Resolved On: 2001/11/25
Id: 0086
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
One another topic, I was looking for a way to do a one line ccdoc comment for this type of suffix documentation:
class A
{
public:
  /** An int variable */
  int intVar;
 
  /** A long variable */
  long longVar;
};
as
class A
{
public:
  int   intVar;  //@- An int variable
  long  longVar; //@- A long Var
};
This is not a priority item. It probably has limited application. But I personally have many files where a single short comment line to the right of the documented item makes the file easier to read. I tried finding a place to do this but stopped when it took too much time. I'll keep at it when I some spare time.
This is an interesting idea. Your idea of a //@- syntax makes sense.
Solution Description:
Fixed in r26. Verified by test 53.

Ben did all of the work to implement the changes to phase1_scanner.cc.


Title: Exceptions were not reported as links.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/10/25
Resolved By: Joe Linoff
Resolved On: 2001/10/25
Id: 0087
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc was not reporting links to exceptions in @exception and @throws directives.
Solution Description:
Fixed in r24. Verified by test #39.

The code for resolving the exception class was not implemented in ccdoc::phase3::html::write_ccdoc_exception_directive_info() (phase3_html.c).


Title: An extra invalid link reported for @link references to classes.
Status: FIXED
Reported By: Ben Zoe
Reported On: 2001/10/23
Resolved By: Joe Linoff
Resolved On: 2001/10/25
Id: 0088
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
An extra invalid link reported for @link references to classes. The following code fragment demonstrates the problem:
class A {
};
class B {
public:
  /**
   * Comment.
   * @link A
   */
  void fct();
};
The link will be reported as "link: A {A}" where the second link is invalid.
Solution Description:
Fixed in r24. Verified by test #39.

The STMT_CLASS_END statement (a dummy) was being reported as an invalid link. This was corrected by adding ccdoc::phase3::html::find_and_write_links() to detect and ignore end of class statements as well other send statements for structs, unions and namespaces.


Title: Ccdoc short description syntax is not compliant with javadoc.
Status: FIXED
Reported By: Mĺrten Gustavsson
Reported On: 2001/10/25
Resolved By: Joe Linoff
Resolved On: 2001/10/25
Id: 0089
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
From the javadoc manual http://java.sun.com/j2se/javadoc/writingdoccomments/index.html#generalform:
First Sentence
The first sentence of each doc comment should be a summary sentence, containing a concise but complete description of the API item. This means the first sentence of each member, class, interface or package description. The Javadoc tool copies this first sentence to the appropriate member, class/interface or package summary. This makes it important to write crisp and informative initial sentences that can stand on their own.

This sentence ends at the first period that is followed by a blank, tab, or line terminator, or at the first tag (as defined below). For example, this first sentence ends at "Prof.":

/**
  * This is a simulation of Prof. Knuth's MIX computer.
  */
Ccdoc doesn´t handle first sentences according to javadoc standard. I believe the summary is created from the first paragraph.

I think that there is a point in following the javadoc specification as close as possible, the documentation is for free.


This is an excellent point and I agree with your conclusion. My only concern is that changing the existing behavior may adversely affect users that depend on the current incorrect behavior.

I feel that the best solution is to modify ccdoc to conform to the javadoc standard but provide a switch that allows the current behavior.

Solution Description:
Fixed in r24. Verified by test #40.

Changed the default behavior of ccdoc to conform to javadoc and added two phase 1 switches: -[no]jdsds. The -jdsds (JavaDoc Short Description Syntax) switch enables the javadoc behavior and is the default. The -nojdsds switch enables the old behavior for backward compatibility.

Beware! This change will affect your output. Use the -nojdsds switch to enable the old behavior.


Title: Support the {@link ...} javadoc syntax.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/10/25
Resolved By: Joe Linoff
Resolved On: 2001/10/26
Id: 0090
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc does not support inline link references. It should be enhanced to support the {@link ...} directive.

As an interesting aside, the @link directive in ccdoc was developed independently of javadoc.

Solution Description:
Fixed in r24. Verified in test #44.

Title: Constant values are not hyperlinked in the documentation for function arguments.
Status: FIXED
Reported By: David McBride
Reported On: 2001/10/25
Resolved By: Joe Linoff
Resolved On: 2001/10/25
Id: 0091
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I have a suggestion. I've noticed that if an enum is a parameter in a funtion, then a link gets created back to the enum definition. That's perfect. However, it doesn't seem that the same is true for constants defined either in the global or the class namespace (unless I'm missing something). It would be handy if ccdoc treated constants the same as enums in this regard.

Example:

enum myEnum {..........};
const double myConstant = 30.0;
 
class myClass
{
public:
  void myFunct(myEnum fred, double ethel = myConstant);
}
ccdoc creates a link in myFunct at "myEnum" back to the definition of "myEnum". However, no link is created in myFunct at "myConstant".
Solution Description:
Fixed in r24. Verified by test #41.

The ccdoc::phase3::html::write_code_subsection_token() method was not looking for variables or functions.


Title: The @link syntax does not support the # scoping operator.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/10/26
Resolved By: Joe Linoff
Resolved On: 2001/10/26
Id: 0092
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc does not support the '#' scoping operator in @link directives. It works correctly for @see directives. The example below illustrates the problem.
class A {
public:
  void fct();
  /**
   * {@link #fct foo}
   * @link #fct
   * @see #fct
   */
  void fct1();
};
Solution Description:
Fixed in r24. Verified in test #44.

Title: Add javadoc style doc comment inheritance.
Status: OPEN
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/01
Resolved By:
Resolved On:
Id: 0093
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Javadoc has quite an advanced feature for doc comment inheritance, see http://java.sun.com/j2se/1.4/docs/tooldocs/win32/javadoc.html#inheritingcomments.

I would very much appreciate doc comment inheritance, prefferably according to javadoc 1.4. Javadoc 1.3 (http://java.sun.com/j2se/1.3/docs/tooldocs/win32/javadoc.html#inheritingcomments) is not so advanced but would be quite useful also.


This is an excellent idea. I will look into how feasible it is for ccdoc. The only issue is how well ccdoc can recognize these relationships.
Solution Description:

Title: Comment reuse for overloaded functions.
Status: OPEN
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/01
Resolved By:
Resolved On:
Id: 0094
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Previously I suggested that you implement doc comment inheritance. If you decide to do so I have another suggestion.

I believe that the javadoc rules says that a method must override a method in the ancestor class to reuse a doc comment. Doc comments cannot be reused for constructors, destructors etc.

I find it quite tedious to write doc comments for constructors and destructors. It would save a lot of typing if one could reuse doc comments beween them.

I can see that the implementation can be difficult, but there is another enhancement that could simplify a lot. Lets call it doc comment reuse of overloaded functions. There is no such thing in javadoc, but maybe there should be. I have read the javadoc spec over and over since I really believed there was such a rule, but there is no rule.

Doc comment reuse of overloaded functions

If an overloaded member function does not have a doc comment, and placed after a function it overloads, it will inherit the doc comment of that function. The names of the parameters of the overloaded functions must be same, only types may vary.

The result is that one only needs to document the first function in a group of overloaded functions.

Doc comment reuse of constuctors etc

Hope that you will excuse me doing specification by example: If a member function (constructor, destructor, etc) uses the class name itself in the parameter list, a second "invisible" function is generated, replacing the current class name with a superficial "this". Using the superficial type "this" is good because it is a reserved word in C++ preventing any real class to be defined.

If the original function has a doc comment the generated function will be placed after the original one, inheriting its doc comment using the rule for overloaded functions.

If the original function doesn't have a doc comment, the generated one will be placed before the original one, allowing the original one to inherit from the generated one using the rule for overloaded functions.

A code example:

class Base 
{ 
        /** Default construcor */ 
        Base(); 
        /** 
         * Copy constructor. 
         * 
         * @param src   is the object to copy from 
         */ 
        Base(const Base &src); 
        /** Destructor */ 
        ~Base(); 
        /** 
         * Assignment operator. 
         * 
         * @param src   is the object to copy from 
         * @return              a const reference to this object 
         */ 
        const Base& operator=(const Base& src); 
} // class Base 

class Derived : public Base 
{ 
        Derived(); // inherit doc comment 
        Derived(const Derived &src); // inherit doc comment 
        ~Derived(); // inherit doc comment 
        const Derived& operator=(const Derived & src); // inherit doc comment 
        /** 
         * Strange copy constructor that slices the object. 
         * 
         * @param src   is the base class object to copy from 
         */ 
        Derived(const Base &src); 
} // class Derived 
Theses classes will internally be represented as something like this:
class Base 
{ 
        /** Default construcor */ 
        Base(); 
        this(); // inherit doc comment from previous using overload rule 
        /** 
         * Copy constructor. 
         * 
         * @param src   is the object to copy from 
         */ 
        Base(const Base &src); 
        this(const this &src);// inherit doc comment from previous using overload rule 
        /** Destructor */ 
        ~Base(); 
        ~this();// inherit doc comment from previous using overload rule 
        /** 
         * Assignment operator. 
         * 
         * @param src   is the object to copy from 
         * @return              a const reference to this object 
         */ 
        const Base& operator=(const Base& src); 
        const this& operator=(const this& src);// inherit doc comment from previous using overload rule 
} // class Base 

class Derived : public Base 
{ 
        this(const this &src); // this one will pick up doc comment from base class 
        Derived(); // inherit doc comment from previous using overload rule 
        this(const this &src); // this one will pick up doc comment from base class 
        Derived(const Derived &src); // inherit doc comment from previous using overload rule 
        ~this(); // this one will pick up doc comment from base class 
        ~Derived(); // inherit doc comment from previous using overload rule 
        const this& operator=(const this& src); // this one will pick up doc comment from base class 
        const Derived& operator=(const Derived & src); // inherit doc comment from previous using overload rule 
        /** 
         * Strange copy constructor that slices the object. 
         * 
         * @param src   is the base class object to copy from 
         */ 
        Derived(const Base &src); 
} // class Derived 

I really like this idea. I will look into how feasible it is for ccdoc. The only issue is how well ccdoc can recognize these relationships.
Solution Description:

Title: Comma's inserted for each line in @returns and @deprecated comments.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/11/02
Resolved By: Joe Linoff
Resolved On: 2001/11/02
Id: 0095
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc inserts comma's for each new line in an @returns comment. The example below illustrates the problem:
/**
 * Returns bug?
 * @returns Are extra
 * comma's inserted
 * at each
 * new line?
 */
void fct();
The output looks like this:
Are extra , comma's inserted , at each , new line?

The workaround is to wrap the comment so there are no newlines. This will be fixed in r25.
Solution Description:
Fixed in r25. Verified by test #45.

Title: Provide an @repeat operator to avoid having to re-type comments.
Status: OPEN
Reported By: David McBride
Reported On: 2001/11/02
Resolved By:
Resolved On:
Id: 0096
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
As you know, if you have a set of overloaded functions, they often perform essentially the same function, and could be described by the same ccdoc comment. Therefore, it would be handy to have a tag that would tell ccdoc to apply the comment attached to the previous method to the current method. For example:
class MyClass {
public:
     /**
     * Performs some useful function. blah blah blah.
     * Some long, involved explanation that I don't
     * want to repeat over and over again.
     * @param someVar Some user defined variable.
     * @returns Returns 0 if successful or not 0 otherwise.
     */
     void myOverloadedFunct(A* someVar);

     /**
     @repeat
     */
     void myOverloadedFunct(B* someVar);

     /**
     @repeat
     */
     void myOverloadedFunct(C* someVar);

     /**
     @repeat
     */
     void myOverloadedFunct(D* someVar);

     /**
     @repeat
     */
     void myOverloadedFunct(E* someVar);
}

This is an interesting suggestion. I believe that it relates to issues 0093 and 0094. They are all trying to address the problem of having to type redundant comments. I am actively looking into this.
Solution Description:

Title: Private method is reported incorrectly for overloaded methods with different scopes.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/11/02
Resolved By: Joe Linoff
Resolved On: 2001/11/12
Id: 0097
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Private method is reported incorrectly for overloaded methods with different scopes. The following example illustrates the case.
class A {
public:
  void fct();
private:
  void fct();
};
The private void fct shows up in the list.
The workaround is to tell ccdoc to ignore the private function using the __ccdoc__ pragma. This will be fixed in r25.
Solution Description:
Fixed in r25. Verified by test #46.

Title: It would be a help if you made the reference red if the class cannot be found for @link.
Status: FIXED
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/09
Resolved By: Joe Linoff
Resolved On: 2003/02/18
Id: 0098
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
It would be a help if you made the reference red if the class cannot be found for @link.
Solution Description:
Fixed in r34, verified by test 73.

Title: Ccdoc does not correctly follow the standard for default constructors.
Status: FIXED
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/09
Resolved By: Joe Linoff
Resolved On: 2001/11/28
Id: 0099
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
A default constructor is implicitly defined if there are no constructors defined for a class. If a copy constructor is the only constructor defined there will be no default (implicitly defined) constructor and the class will be unusable. Trying to instasiate such a class will cause a compilation error. I believe that the implicitly defined default constructor is there to achieve compatibility with C structs.

ccdoc generates a default constructor entry if there is no default constructor defined. This is not according to C++ standard. I suggest that you change ccdoc to only generate documentation for implicitly defined default constructors if there are no constructors defined in the class.

Also if you call generated member functions "implicitly defined" this will improve understanding since it will adhere to C++ lingo.

Solution Description:
Fixed in r26. Verified by test 55.

Modified special member processing to recognize when any constructor is declared so that it can turn off the generation of the default constructor correctly.


Title: If using switch -files with a wildcard expression the first file is not processed.
Status: CLOSED
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/09
Resolved By: Joe Linoff
Resolved On: 2001/11/12
Id: 0100
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I saw in the issue list that someone has a problem with gnuwin32. If using switch -files with a wildcard expression the first file is not processed. No problem if removing the -files switch.
Solution Description:
The first file is not really ignored because the ccdoc expects it to contain a list of files to process. When it doesn't find any files it goes on to the other entries on the command line.

Title: Template problem.
Status: FIXED
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/09
Resolved By: Joe Linoff
Resolved On: 2001/11/27
Id: 0101
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The following code
    template<class VALUE_TYPE> class Prompt : public Component  
    { 
    public: 
        /** 
         * Reads from the instream to an object and returns a reference to it. 
         * First, a text defined in the constructor is written to the ostream defined 
         * in the constructor. Then the instream defined in the constructor is 
         * read to the internal object, using operator>>(). A reference to the 
         * internal object is returned. 
         */ 
        Input< ReturnPrimitiveInterface<const VALUE_TYPE&, const Event&> > in; 
generates this output:
attribute const ? ^ < >        
Undocumented 

   Source: 

      metaf/src/components/io/Prompt.h:52 

   Code: 

      public Input < ReturnPrimitiveInterface < const VALUE_TYPE const Event & > > in 
Solution Description:
Fixed in r26. Verified in test 54.

The variable recognition heuristic in phase1::parser::parse_var_or_fct did not correctly recognize template arguments.


Title: Returned nested classes and types in namespaces are not recognized correctly.
Status: FIXED
Reported By: Dmitry A.Steklenev
Reported On: 2001/11/10
Resolved By: Joe Linoff
Resolved On: 2001/11/26
Id: 0102
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Returned nested classes and types in namespaces are not recognized correctly.

Code Sample:

namespace nms {
class some_class
{
   public:
     typedef int nested_type;
     class nested_class {};
};

class another_class { public: /** Linked correctly */ some_class get_some_class(); /** Not linked correctly */ some_class::nested_type get_nested_type(); /** Not linked correctly */ some_class::nested_class get_nested_type(); }; }
Solution Description:
Fixed in r26. Verified by test 52.

See issue 0118 for more details.


Title: Invalid processing of operator~().
Status: FIXED
Reported By: Dmitry A.Steklenev
Reported On: 2001/11/10
Resolved By: Joe Linoff
Resolved On: 2001/11/12
Id: 0103
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Sample code:
class some_class
{
  public:
    some_class operator~();
};

CCDoc output:

~    operator public undocumented
Solution Description:
Fixed in r25. Verified by test #47.

The ccdoc::phase1::parser::get_fct_id() method was being called for operators.


Title: Invalid processing of operator&().
Status: FIXED
Reported By: Dmitry A.Steklenev
Reported On: 2001/11/10
Resolved By: Joe Linoff
Resolved On: 2001/11/12
Id: 0104
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Sample code:
class some_class
{
  public:
    some_class operator&( const some_class& );
};
CCDoc output:
some_classoperator & operator public undocumented
Solution Description:
Fixed in r25. Verified by test #47.

The ccdoc::phase1::parser::get_fct_id() method was being called for operators.


Title: The javadoc @see tag defines the label part as optional.
Status: OPEN
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/08
Resolved By:
Resolved On:
Id: 0105
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The javadoc @see tag defines the label part as optional: http://java.sun.com/j2se/1.4/docs/tooldocs/win32/javadoc.html#@see

If you decide to do the label optional, javadoc specifies how to generate the link: http://java.sun.com/j2se/1.4/docs/tooldocs/win32/javadoc.html#shortened

I don´t feel that this improvement is very important, but if you like the idea put it on the issue list and maybe implement it if there is time and no more important things to do.

Solution Description:

Title: The javadoc spec for @see allows unqualified names as well as qualified names.
Status: OPEN
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/08
Resolved By:
Resolved On:
Id: 0106
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The javadoc spec for @see allows unqualified names as well as qualified names: http://java.sun.com/j2se/1.4/docs/tooldocs/win32/javadoc.html#seesearchorder

I believe that ccdoc requires a fully qualified c++ class name if the reference is to another class. The javadoc spec relies on rules including the java compiler and I don´t believe that following their search patterns will add that much.

However, it would be nice if the @see tag accepts valid c++ class names. Leaving out the namespace part if the referencing class belongs to the same namespace as the referenced class would be an improvement.

I don´t think this improvement is very important since namespaces are quite short compared to java packages.

Solution Description:

Title: javadoc # syntax is not supported in the {@link...} directive.
Status: FIXED
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/08
Resolved By: Joe Linoff
Resolved On: 2003/02/18
Id: 0107
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
It seems like it isn´t possible to link to a member function in another class. Example from my code: * {@link metaf::ComAbstractGetAdoRecord#get_one_record ComAbstractGetAdoRecord::get_one_record()}

Javadoc allows this construct: http://java.sun.com/j2se/1.4/docs/tooldocs/win32/javadoc.html#@see

The syntax would be quite useful.


Thanks for all of your suggestions. I am still processing them but I wanted to respond to this issue quickly because it might help you.

It turns out that ccdoc does allow you to reference member functions but it doesn't use the javadoc syntax, you must use a '::' separator as shown in the following example.

     class A {
     public:
       void fct();
     };
     class B {
     public:
       /**
        * See the {@link A::fct fct} for additional details.
        * @see A::fct
        */
       void bar();
     };
I completely agree that ccdoc should support the javadoc syntax and will schedule this fix for the next release.
Solution Description:
Fixed in r34.

Title: -rptdefasd, -rptdefsd and -rptdefv problems and suggestions
Status: OPEN
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/08
Resolved By:
Resolved On:
Id: 0108
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I am playing around with the switches -rptdefasd, -rptdefsd and -rptdefv, the intention to mark the text red "Undocumented" in order to find missing documentation and implicitly defined member funcions.

For the implicitly defined member funcions I want the text "Implicitly defined", and for things that I forgot to document "Undocumented", the switches used:

-rptdefasd "Implicitly defined" -rptdefsd "Undocumented" This causes the class documentation look just like I want. However, I have found the following errors:

The switches I am quite happy with: -rptdefa "Unascribed" -rptdefasd "Implicitly defined" -rptdefsd "Undocumented" -rptdefv "Unknown"

And some suggestions:

Solution Description:

Title: Print warnings to stderr.
Status: FIXED
Reported By: Mĺrten Gustavsson
Reported On: 2001/11/08
Resolved By: Joe Linoff
Resolved On: 2001/02/18
Id: 0109
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Marking stuff with problems with red is great. You can mark even more, like missing default constructors, copy constructors and other things automatically generated. Another thing to warn for is destructors not being virtual. Im my experience a missing copy constructor, for a class with pointers in it, combined with a missing * or & in a function call will corrupt memory. I have spent days looking for such errors in my code. When starting doing C I found lint to be an exellent tool to identify problematic code.

If the programmers intention is for C++ to automatically generate a constructor, there should be something in the documentation telling this, prefferably as a doc comment. If so there should be no red warning text.

Since I have so many classes, I find it tedious to click through all genereated html to find problematic links. I suggest that you also print warning messages to stderr. Finding the problems will be much faster.

If you decide to implement warnings for dangerous C++ constructs I guess that your user community will come up with even more ideas if asked.

Solution Description:
Fixed in r34, verified by test 73.

Added warning messages for all conditions that trigger red in the HTML code.


Title: Invalid processing of operator[].
Status: FIXED
Reported By: Lou Sanchez-Chopitea
Reported On: 2001/11/16
Resolved By: Joe Linoff
Resolved On: 2001/11/12
Id: 0110
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Invalid processing of operator~[]. This is exactly the same problem as issues 0103 and 0104.
Solution Description:
Fixed in r25.

Title: Multiple link file overflow problem.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/11/16
Resolved By: Joe Linoff
Resolved On: 2001/11/17
Id: 0111
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I ran across a problem where multiple links caused too many files to be generated.
Solution Description:
Fixed in r25.

The problem was fixed by re-using common link files.


Title: Phase 3 exception for operator|.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/11/19
Resolved By: Joe Linoff
Resolved On: 2001/11/19
Id: 0112
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Operator | clauses cause an exception to be generated in phase3.
Solution Description:
Fixed in r25. Verified by test48.

Title: Merge separate ccdoc databases.
Status: OPEN
Reported By: John Rostykus
Reported On: 2001/11/20
Resolved By:
Resolved On:
Id: 0113
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
On a related note, I would like to talk to you about possibly supporting a feature in ccdoc to merge databases. We could really parallelize the process if we were not limited to having every ccdoc process write to the same database. For example, run the header processing on each of the 50 or so subsystems individually in parallel, run a (new) step to merge the resulting databases into 1, and then do the index/html steps. Does ccdoc have any capability like this? If not, what do you think of this idea? In general, long processes that are single threaded are what kill our build times, and ccdoc (the way we use it) at ~1 hour is in that camp.
I think that this is very interesting idea that shouldn't be too difficult. I will look into it for the next release.
Solution Description:

Title: Error processing of a operator unsigned long().
Status: FIXED
Reported By: Dmitry A.Steklenev
Reported On: 2001/11/22
Resolved By: Joe Linoff
Resolved On: 2001/11/26
Id: 0114
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Error processing of a operator unsigned long().

Sample code:

class some_class
{
  public:
    some_class operator unsigned long();
}
CCDoc output:
operator unsignedlong operator public undocumented
Solution Description:
Fixed in r26. Verified by test49.

There was a name munging heuristic in the phase1::parser that collapsed operators like [] and (). It wasn't checking for cast operator entries like "unsigned long".


Title: Mangled links for .\doc\ references on the command line.
Status: FIXED
Reported By: Richard Cooper
Reported On: 2001/11/22
Resolved By: Joe Linoff
Resolved On: 2001/11/27
Id: 0115
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Running ccdoc with the following command line produces documentation with mangled links. The documents are in .\doc but the hyperlinks link to .\doc\doc
ccdoc -db docs.db -html .\doc\ ..\Headers\*.h
The system works fine if I do the following but as this will be part of an automatic process I don't want to.
cd .\doc
ccdoc -db docs.db -html .\ ..\..\Headers\*.h
It also works if I move the doc directory to the same level as the Headers directory
ccdoc -db docs.db -html ..\doc\ ..\Headers\*.h
but not if I move it further up
ccdoc -db docs.db -html ..\..\doc\ ..\Headers\*.h

The problem that you encountered with the -html .\ specification arose because ccdoc uses the -html specification to generate URLs and the backslash is not supported as an HTTP path specifier.

You can fix the problems that you were having by replacing the backslashes with forward slashes. Ccdoc will figure out the correct DOS path and the URL will work.

I apologize for not making this clear in the users guide. I will update the information and close out this issue if this works for you. If I don't hear from you by tommorrow (Nov 27) I will assume that things are working for you.

Solution Description:
Updated the documentation to reflect the requirement that -html path specifications must use forward slashes on all platforms (even DOS).

Title: The string constant "\\" isn't parsed correctly.
Status: FIXED
Reported By: Chris Martin
Reported On: 2001/11/25
Resolved By: Chris Martin
Resolved On: 2001/11/25
Id: 0116
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The string constant "\\" isn't parsed correctly; resulting in a spurious "unterminated string literal" warning.

Fix:

void ccdoc::phase1::scanner::get_string_literal(char* token,int max)
{
   int ch = '"';
   int pch = ch;
   while( max>0 && (ch = get_char()) ) {
     max--;
     *token++ = ch;
     if( ch == '"' && pch != '\\' ) {
       *token = 0;
       return;
     }
#ifdef CHRIS
     if (ch == '\\' && pch == '\\')      // escaped escape
         ch = 0;
#endif
     pch = ch;
   }
   .
   .

Thank you for taking the time to track down and fix this problem. I will incorporate it in the next release.
Solution Description:
Fixed in r26. Verified by test 50.

Incorporated the fix by Chris and added a verification test.


Title: Class method reported as global function.
Status: FIXED
Reported By: Chris Martin
Reported On: 2001/11/25
Resolved By: Chris Martin
Resolved On: 2001/11/25
Id: 0117
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The following code results in the second "Set" method appearing incorrectly as a standalone (global) function:
class Foo {
public:
     void Set(int a);
     void Set(char *cp, int a);
};
void Foo::Set(int a)
{
     /** Set one. Red */
}
void Foo::Set(char *cp, int a)
{
     /** Set two. Yellow */
}
I think the problem is in ccdoc::database::erase_from_path_map, where the entire list of statements associated with a path map entry are removed from the path map, when only the particular statement should be.

My suggested fix is as follows:

void ccdoc::database::erase_from_path_map(ccdoc::statement::base* stmt)
{
   if( stmt ) {
     if( stmt->get_type() != statement::base::STMT_COMMENT_PKGDOC &&
         stmt->get_type() != statement::base::STMT_COMMENT_PKGDOC_URL &&
         stmt->get_type() != statement::base::STMT_COMMENT_PREFIX &&
         stmt->get_type() != statement::base::STMT_COMMENT_SUFFIX ) {
       // Issue 0041
       // Ignore comments, they don't have meaningful names.
       const char* pid = stmt->get_id();
       if( *pid ) {
         string id;
         stmt->get_hier_id_no_pkgs(id);
         if( id.size() ) {
           path_map_itr_type itr = m_path_map.find(id);
           if( itr != m_path_map.end() ) {
#ifdef CHRIS
               statement::base::stmts_t &vec = (*itr).second;
               statement::base::stmts_itr_t vitr = vec.begin();
               for(;vitr != vec.end(); vitr++) {
                   if ((*vitr) == stmt) {
                       vec.erase(vitr);
                       break;
                   }
               }
               if (vec.size() == 0)
                   m_path_map.erase(itr);
#endif
         }
       }
     }
     statement::base::stmts_t& children = stmt->get_children();
     statement::base::stmts_itr_t itr = children.begin();
     for(;itr!=children.end();++itr) {
       erase_from_path_map(*itr);
     }
   }
   .
   .

Thank you for taking the time to track down and fix this problem. I will incorporate it in the next release.
Solution Description:
Fixed in r26. Verified by test 51.

Incorporated the fix by Chris and added a verification test.


Title: Resolve partially scoped references.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2001/11/26
Resolved By: Joe Linoff
Resolved On: 2001/11/26
Id: 0118
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
This is similar to issue 0065.

Ccdoc does not manage partially scoped names as shown in the following example:

namespace test52
{
  class A
  {
  public:
    typedef int nested_type;
    class nested_class {};
    void fct();
  };
  
  class B
  {
  public:
    /** Linked correctly */
    A get_some_class();
    /** Not linked correctly */
    A::nested_type  get_nested_type();
    /** Not linked correctly */
    A::nested_class get_nested_class();
    /** Not linked correctly */
    A::fct get_nested_fct();
  };
}
Solution Description:
Fixed in r26. Verified by test 52.

Added a heuristic that attempts to resolve partially scoped names. It works better than before but it is not perfect because ccdoc doesn't understand the using namespace clause.


Title: Inconsistent command line syntax for -html paths under DOS.
Status: OPEN
Reported By: Richard Cooper
Reported On: 2001/11/27
Resolved By:
Resolved On:
Id: 0119
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In regards to issue 0115, replacing the backslashes with forward slashes in the -html switch does fix the problem. i.e. this works
ccdoc -db docs.db -html ./doc ..\Headers\*.h
But in DOS you can't use backslashes in the headers path. i.e. this doesn't work
ccdoc -db docs.db -html ./doc ../Headers/*.h
This is not entirely satisfactory because now the two path descriptions in the command line need to be in different formats (the html path in UNIX form, the header path in DOS form). IMO the best fix would be to accept any paths in either format on any platform, but I have no idea how tricky that would be to implement.
I agree.

Unfortunately this is somewhat difficult because ccdoc can't reliably determine whether it is running under cmd.exe or under a unix shell that is running on top of cmd.exe at run time.

If it is running under cmd.exe, it can safely replace backslashes with forward slashes but, if it is running under a unix shell that is running on top of cmd.exe it cannot.

I will look into whether ccdoc can be enhanced to allow you to specify UNIX style paths (with wildcards) under DOS.

Solution Description:

Title: Add support for the proposed todo tag in javadoc.
Status: FIXED
Reported By: Gustavsson Mĺrten
Reported On: 2001/11/28
Resolved By: Joe Linoff
Resolved On: 2002/11/11
Id: 0120
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I think that a todo tag would be very useful. It is a javadoc proposed tag: http://java.sun.com/j2se/javadoc/proposed-tags.html.
I agree. I will schedule it for the relatively near future.
Solution Description:
Fixed in r27. Tested in test59.

Title: Reuse namespace comments between packages.
Status: OPEN
Reported By: Gustavsson Mĺrten
Reported On: 2001/11/29
Resolved By:
Resolved On:
Id: 0121
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I use package names that reflect the directory structure. Each package is run through ccdoc in a separate ccdoc invocation.

I use the same namespace between packages.

The namespace comment is valid only for the package in which it is defined. This is alright. The problem is that the documentation is a bit unclear about the scope for namespace doc comment reuse. I thought the namespace comment could be reused between packages.


This makes good sense. In fact, this is how it should work. I suspect that there is a bug in there somewhere.
Solution Description:

Title: Enhance template reporting.
Status: FIXED
Reported By: Gustavsson Mĺrten
Reported On: 2001/11/29
Resolved By: Joe Linoff
Resolved On: 2003/02/18
Id: 0122
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Template classes are reported ordinary classes in html output. It would be great if the html showed that a class is a template class.

It would also be nice if the template definition was shown just like a function declaration.


I like this idea.
Solution Description:
Fixed in r34.

Title: {@link ...} not reported correctly in the @deprecated directive
Status: FIXED
Reported By: Joe Linoff
Reported On: 2002/02/25
Resolved By: Joe Linoff
Resolved On: 2002/11/11
Id: 0123
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The following comments does not report the link correctly:
struct A {
  /**
   * Do stuff.
   */
  void do_stuff();
};
struct B {
  /**
   * Do stuff.
   * @deprecated See {@link A::do_stuff}.
   */
  void fct();
};
This was actually reported by a number of folks.
Solution Description:
Fixed in r27. Verified by test58.

Title: Scanner condition evaluates to a constant value.
Status: FIXED
Reported By: Michel Joly de Lotbiniere
Reported On: 2002/02/25
Resolved By: Joe Linoff
Resolved On: 2002/03/21
Id: 0124
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Just compiled "right out of the box" ccdoc v0.8 r26 2001/11/28 with IBM's VisualAge C++ 4.0 on Win32. There was only one message:

CPPC1104I:The condition evaluates to a constant value. ("phase1_scanner.cc", line 179.9: )

The code around that is:

// ================================================
// Eliminate the "\\\n" white space. This cannot
// be done in skip_ws() because of the backslash
// trigraph sequence "??/".
// ================================================
if( '\\' == ch ) {
  char ch1 = get_char();
  if( '\n' ) {           <===== The warning is here
    return get_token();
  }
  put_char(ch1);
}
Did you intend to write: if( '\n' == ch1 ) { etc.?
I did indeed mean that. This is fixed in r27.
Solution Description:
Fixed in r27.

This was rolled in with the fixed for issue 129.


Title: Hash mark displayed for default {@link...} reference.
Status: FIXED
Reported By: Ethan Nagel
Reported On: 2002/02/25
Resolved By: Joe Linoff
Resolved On: 2002/11/11
Id: 0125
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I noticed another minor issue that would be nice if it was cleared up. Inline @links seem to work pretty well (actually they work better than the documentation, but when using the "#" syntax to reference a member of the current class the default text leaves the "#" sign attached! I'm frankly not sure this is even really a bug and there are easy workarounds, but it is annoying. Perhaps I'm just using ccdoc wrong somehow. Here's an example:
class Sample
{
public:
   /**
    * Calls {@link #b}.
    */
   int A();
   /**
    * Calls {@link #A A}.
    */
} ;
The documentation for A would have the "#" sign displayed, while the documentation for B would give me the results I think you should get automatically.
You are not using ccdoc the wrong way. This is an annoying bug and will be fixed in the next release.
Solution Description:
Fixed in r27. Verified by test60.

Title: CcDoc does not recognize ccdoc c++ short comment in a "typedef struct" definition.
Status: OPEN
Reported By: Luis Menina
Reported On: 2002/02/26
Resolved By:
Resolved On:
Id: 0126
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
CcDoc does not recognize ccdoc c++ short comment in a "typedef struct" definition (comment does not appear).
typedef struct 
{ 
    int foo1; //@- useful stuff 1
    int foo2;   //@- useful stuff 2
} MY_STRUCT;
Solution Description:

Title: CcDoc does not recognize recognize template derivations.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2002/03/01
Resolved By: Joe Linoff
Resolved On: 2002/11/11
Id: 0127
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
CcDoc does not recognize recognize template derivations. The following example does not work properly:
template <class T>
struct A {
   void fct1();
};

struct B {
   void fct2();
};

struct C : public A<B>
{
   void fct3();
};
You can work around this bug as follows:
template 
struct A {
   void fct1();
};

struct B {
   void fct2();
};

struct C : public A /**@#-*/<B>/**@#+*/
{
   void fct3();
};
But, as you can see, it is ugly. I will fix this in r27.
Solution Description:
Fixed in r27. Added test57 to verify it.

Title: Order class entities the same way that were ordered in v07.
Status: OPEN
Reported By: Luis Menina
Reported On: 2002/03/01
Resolved By:
Resolved On:
Id: 0128
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
May the class content be sorted not just alphabetically, but separating method and attributes ? I think it would be more clear and efficient (as it was in ccdoc v0.7).
Some folks like it this way, others like it the way that you have described. I think that a switch should be added that allows this.
Solution Description:

Title: All subdirectories are picked up on NT when the * wildcard is specified.
Status: FIXED
Reported By: Clint Bauer
Reported On: 2002/03/14
Resolved By: Joe Linoff
Resolved On: 2002/03/21
Id: 0129
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
When I use this line to ccdoc my files: ccdoc -v *.h -db medias.db, ccdoc happens on every .h file in the current directory, and also on every .h file in every sub-directory to n-deep sub-directories. This I have found by experimentation.
This is a known problem on windows and I will try to fix it in the next release. Meanwhile you can work around it by specifying the files explicitly.
Solution Description:
Fixed in r27.

In DOS mode ccdoc used the "DIR /B /S /O:N" command to deduce wild card file names. I changed this to "DIR /B /O:N" to eliminate the spurious subdirectory searches.


Title: Code section wraps in a strange way for small browser windows.
Status: FIXED
Reported By: Doug Wieland
Reported On: 2002/03/15
Resolved By: Joe Linoff
Resolved On: 2002/11/12
Id: 0130
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The ccdoc output for code segment can cause unreadable output if the browser window is too small.
This is a good point. I will add the HTML <TD> attribute tag NOWRAP in the next release for Code sections.
Solution Description:
Fixed in r27.

Title: Support comma separated declarations.
Status: OPEN
Reported By: Esmond Pitt
Reported On: 2002/09/24
Resolved By:
Resolved On:
Id: 0131
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I can't report this verbatim because my e-mail system crashed but here is my best recollection.

Modify ccdoc to allow different comments for each item in a comma separated declaration. For example:

   const string
      /** This is one string */
      string1 = "aaa",
      /** This is another string */
      string2 = "bbb";
Note that you can workaround this in the near term as follows:
   // Workaround
   /** This is one string */
   const string string1 = "aaa";
   //** This is another string */
   const string string2 = "bbb";
Solution Description:

Title: support single-line prefix comments via //@+
Status: OPEN
Reported By: Esmond Pitt
Reported On: 2002/09/24
Resolved By:
Resolved On:
Id: 0132
[Top]
[Summary]
[Next]
[Previous]
Problem Description:

Presently we have

        string  m_memberString; //@- the member documentation for this item
How about adding this:
        //@+ the member documentation for this item
        string  m_memberString;

This makes sense as a shorthand but since you can workaround it by using the following syntax. This modification will be scheduled with a lower priority.
        /** the member documentation for this item */
        string  m_memberString;
Solution Description:

Title: Anonymous namespaces are not handled correctly.
Status: FIXED
Reported By: Esmond Pitt
Reported On: 2002/09/24
Resolved By: Joe Linoff
Resolved On: 2002/11/12
Id: 0133
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
We use anonymous namespaces extensively. Each source file has one or more of these:
  namespace
  {
    // local static stuff, classes, constants, ...
  }
These are supposed to be merged within a single source file but distinct across source files.

Problems which occur:

  1. the namespace name is reported as 'namespace'
  2. a single 'Source:' file is reported
  3. all but the latest namespace contents are lost

You need to do what the compiler does: auto-generate a namespace name based on the file name, so you can keep all the contents separate, and recognize and suppress an auto name when reporting (i.e. report each name as <anonymous>), something like:

  <anonymous>     namespace       // documentation or 'undocumented'
    // contents of this anonymous namespace
  <anonymous>     namespace       // documentation or 'undocumented'
    // contents of this anonymous namespace
  <anonymous>     namespace       // documentation or 'undocumented'
    // contents of this anonymous namespace
  <anonymous>     namespace       // documentation or 'undocumented'
    // contents of this anonymous namespace

I agree. This fix has been scheduled for r27.
Solution Description:
Fixed in r27. Verified by tests 61 and 62.

This was an involved fix. The current implementation keeps all of the comments associated with a namespace and chains them in the documentation for that namespace. All of the members of the namespace are grouped (by package).

Two new switches were added: -[no]rptcfuns to provide users with some level of control over the namespace output.


Title: Need a forward link to the contents section for classes and packages.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2002/11/11
Resolved By: Joe Linoff
Resolved On: 2002/11/12
Id: 0134
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
When the class description is large, it is hard to find the contents section. The change will allow users to click to get to the table of contents.
Solution Description:
Fixed in r27.

Title: Fix the copyright notice in the code.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2002/11/13
Resolved By: Joe Linoff
Resolved On: 2002/11/13
Id: 0135
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Recently I discovered a mail thread in the linux.debian.legal newsgroup discussing the ccdoc copyright notice. The author: "Branden Robinson", made some excellent points which I will directly quote here.
This license's intentions are clearly DFSG-free, but I think it leaves a couple of things unclear.
I think that the answer is that the author doesn't really understand all of this legal stuff and really appreciates any help that he can get:-)
Solution Description:
Fixed in r28.

Title: Ccdoc gets confused by a nested struct.
Status: FIXED
Reported By: Dennis Marsa
Reported On: 2002/11/13
Resolved By: Joe Linoff
Resolved On: 2002/11/14
Id: 0136
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The following code fragment causes ccdoc problems:
class A {
protected:
   A();
public:
   virtual ~A() = 0;
private:
   A(const A& rhs);
   A& operator=(const A& rhs);
public:
   struct B {
      int type;
   };
};
The problems are:
  1. The A default constructor was reported as automatically generated. It was defined as a protected member.
  2. The A copy constructor was reported as automatically generated. It was defined as a private member.
  3. The A copy operator was reported as automatically generated. It was defined as a private member.
  4. The A::B default constructor was not defined.
  5. The A::B destructor was not defined.

This bug has been scheduled to be fixed in r28.
Solution Description:
Fixed in r28 and verified by tests 63 and 64.

Title: The {@link} directive does not work @param clauses.
Status: OPEN
Reported By: Joe Linoff
Reported On: 2002/12/06
Resolved By:
Resolved On:
Id: 0137
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The {@link} directive is not correctly recognized in @param clauses. The code example below demonstrates the problem:
namespace testXX {
   class A {
   public:
      A();
      ~A();
      /**
       * Run the test using the specified arguments.
       * @param argc The number of arguments.
       * @param argv The collection of arguments.
       * @returns The run status.
       */
      virtual int Run(int argc,char** argv);
   };
   class test : public A {
   public:
      test();
      ~test();
      /**
       * Run the test using the {@link test99::A::Run} command.
       * @param argc The number of arguments {@link test99::A::Run}.
       * @param argv The collection of arguments {@link test99::A::Run}.
       * @returns The run status {@link test99::A::Run}.
       */
      virtual int Run(int argc,char** argv);
   };
};

2002/12/06 jdl
After a bit more investigation, I found that there was a workaround as shown below which makes the fix less urgent.
namespace testXX {
   class A {
   public:
      A();
      ~A();
      /**
       * Run the test using the specified arguments.
       * @param argc The number of arguments.
       * @param argv The collection of arguments.
       * @returns The run status.
       */
      virtual int Run(int argc,char** argv);
   };
   class test : public A {
   public:
      test();
      ~test();
      /**
       * Run the test using the {@link test99::A::Run} command.
       * @param argc The number of arguments 
       *             {@link test99::A::Run}.
       * @param argv The collection of arguments
       *             {@link test99::A::Run}.
       * @returns The run status {@link test99::A::Run}.
       */
      virtual int Run(int argc,char** argv);
   };
};
Solution Description:

Title: The @see and {@link} directives do not correctly reference entities in different packages.
Status: FIXED
Reported By: Markus Stulle
Reported On: 2003/01/09
Resolved By: Joe Linoff
Resolved On: 2003/01/13
Id: 0138
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
We are facing the following problem while documenting source code produced using programming language ansi-c: if a function 'foo()' is documented in a named packaged, e.g. 'A.B.C', we fail in referencing that function using the '@link' directive:
// file 'foo.c':
/**
 * @pkg A.B.C
 */

/**
 * this function does nothing,
 * so it is error-free by design.
 */
void foo( void );
// file 'bar.c':
/**
 * @pkg D.E
 */

/**
 * again an important procedure.
 * functionality is identical to {@link A.B.C#foo foo}.
 */
void bar( void );
We tried the following entity strings for referencing 'foo', but none of them produced a link in the resulting hypertext document:
0) A.B.C.foo
1) A.B.C#foo
2) #A.B.C.foo
3) A::B::C#foo
4) A::B::C.foo
5) A::B::C::foo
6) #A::B::C::foo
The motivation for our investigations is the fact, that, unfortunaltely, 'foo()' is defined multiply in our project (i.e. in different packages) and we want to have a direct link to it from 'bar()' generated by the 'ccdoc' processor. if we employ the directive '{@link #foo foo}' ccdoc works fine but delivers a link to a page where the user can select from one of the multiply defined foo()s.
Solution Description:
Fixed in r29 and verified by test 65.

The problem was that these references were not considering the path info when looking up a name in the ccdoc::database::get_stmt_no_pkgs() method. This method was updated to try harder when a name was not found by looking up the name using the path information.


Title: Core dump when an invalid package name of the form "A::B:" appears.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/01/13
Resolved By: Joe Linoff
Resolved On: 2003/01/13
Id: 0139
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Solution Description:
Fixed in r29 and verified by test 66.

There was a missing break that caused the iterator in void ccdoc::database::parse_path() to run off of the end.


Title: Core dump when a friend declaration of the form: foo&operator<< appears.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/01/14
Resolved By: Joe Linoff
Resolved On: 2003/01/14
Id: 0140
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Solution Description:
Fixed in r29 and verified manually. This caused the release date to bump up to 2003/01/15.

There was a missing break that caused the iterator in void ccdoc::database::parse_path() to run off of the end for angle brackets.


Title: Architecture is not reflected in the bin directory name.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/01/21
Resolved By: Joe Linoff
Resolved On: 2003/01/21
Id: 0141
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The problem is that I can't differentiate between linux on an i386 platform, linux on MacOS X and linux on a sparc station.
Solution Description:
Fixed by using $Config{'archname'} instead of $Config{'osname'} in the platform script. This change doesn't affect users to it will be released as part of r30 when the StrongARM port is done.

Title: No man page for linux ports.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/01/21
Resolved By: Joe Linoff
Resolved On: 2003/01/21
Id: 0142
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The Debian linux port had to create its own man page for ccdoc.
Solution Description:
I added a man page that is automatically derived from the help text. This change doesn't affect users to it will be released as part of r30 when the StrongARM port is done.

Title: Support javadocs frames format.
Status: OPEN
Reported By: Thomas Bednarz
Reported On: 2003/01/24
Resolved By:
Resolved On:
Id: 0143
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
"Javadoc documentation is great and very handy to use".
I agree. I will look into this in the near future. A couple of additional thoughts: 1) the new format should be switch controllable to avoid disrupting existing clients, 2) the new format may be higher priority than the XML release (this needs to be investigated).
Solution Description:

Title: Mark static members and methods in the class overview.
Status: FIXED
Reported By: Thomas Bednarz
Reported On: 2003/01/24
Resolved By: Joe Linoff
Resolved On: 2003/02/17
Id: 0144
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
It would be helpful, if static members and methods would be specially marked in the class overview.
This makes sense but I am not sure how to do it cleanly. Do you have any suggestions? Also, do you think that volatile and inline methods should also be marked?
Solution Description:
Fixed in r33 and verified by test 70.

Since static indicates class scope for attributes and methods on classes it is in a category of its own. Class scoped attributes are now marked in the contents table.


Title: It is impossible to reference an operator in the @see and {@link..} directives.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/01/30
Resolved By: Joe Linoff
Resolved On: 2003/01/30
Id: 0145
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I cannot reference an operator== in the @see and {@link..} directives.
Solution Description:
Fixed in r31 and verified by test 67.

The problem was that the operator keyword was not understood by the lookup ccdoc::database::get_stmt_no_pkgs() method. In ccdoc, all operator names consist of two parts: the operator keyword and the operator tokens. They are separated by a space. The link syntax for @see and {@link} do not allow spaces so special dispensation was made to allow users to specify links without spaces. The system then figures them out during the stmt lookup. This example below shows how to specify links for operators:

namespace test67
{
  class Test
  {
  public:
     /**
      * Is equal, this is the same as {@link #is_equal is_equal}.
      * @see #is_equal
      */
     bool operator==(const Test&) const;
     /**
      * Test method, this is the same as {@link #operator== ==}.
      * @see #operator==
      */
     bool is_equal(const Test&) const;
  };
}

Title: Namespace processing doesn't work properly in certain cases.
Status: FIXED
Reported By: Sakari Forsberg
Reported On: 2003/02/07
Resolved By: Joe Linoff
Resolved On: 2003/02/10
Id: 0146
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The current version of ccdoc fails to correctly associated entities between two entities in the same namespace in the following example:
// file1.h
namespace A {
   int B;
}
// file2.h
namespace A {
   /**
    * @see #B
    */
   int C;
}
Solution Description:
Fixed in r32 and verified by test 68.

The problem was a bug processing undocumented namespaces. In older versions it can be worked around by specifying the -rptcfuns switch or by adding a comment to one of the namespaces.


Title: Lots of compiler warnings when building ccdoc using the Borland C++ compiler.
Status: FIXED
Reported By: Sakari Forsberg
Reported On: 2003/02/07
Resolved By: Joe Linoff
Resolved On: 2003/02/10
Id: 0147
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Saw lots of warnings when compiling ccdoc using the Borland C++ compiler.
Solution Description:
Fixed in r32.

This was occurring because I explicity turned off warning messages in the compilers that I use (g++,sun CC,Mac OSC c++) for some silly reason. I turned on warnings and cleaned up all of the warning messages for this release.


Title: Allow users to customize ccdoc output using a CSS file.
Status: OPEN
Reported By: Sakari Forsberg
Reported On: 2003/02/10
Resolved By:
Resolved On:
Id: 0148
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Have you thought, instead of giving color parameters etc. to ccdoc that it can make custom looking html you could give a custom tailored .css-file to ccdoc and it could generate html which uses a .css-file?
I have not thought about it because I don't know much about style sheets but it sounds like a good idea.
Solution Description:

Title: Template method implementations confuse ccdoc.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/17
Resolved By: Joe Linoff
Resolved On: 2003/02/17
Id: 0149
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Ccdoc still gets confused by the following code:
template <class T>
class Test
{
public:
   Test();
   ~Test();
   T GetValue() const {return m_val;}
   void SetValue(T val);
private:
   T m_val;
};
template <class T>
Test<T>::Test()
{
}
template <class T>
Test<T>::~Test()
{
}
template <class T>
void Test<T>::SetValue(T val)
{
   m_val = val;
}
template <class T>
T real_fct(T val)
{
   Test<T> f;
   f.SetValue(val);
   return f.GetValue();
}
template<> int real_fct<int>(int);
template<> class Test<double>;
Solution Description:
Fixed in r33 and verified by test 69.

I added a phase 1 switch called -[no]tcms to tell ccdoc to ignore template class methods that are implemented outside of the declaration. The default for this switch is -notcms.

This implies that all template method documentation must be done in the declaration.


Title: Make it easier to update the revision number in the source files.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/17
Resolved By: Joe Linoff
Resolved On: 2003/02/18
Id: 0150
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
During the last few releases I noticed that updating the various source files with the new revision numbers is error prone. To make it less error prone, I would like to automate it.
Solution Description:
Fixed in r34.

Added the release.pl script to update various source files with the new revision number.


Title: Translate the generated html to other languages.
Status: OPEN
Reported By: Sakari Forsberg
Reported On: 2003/02/14
Resolved By:
Resolved On:
Id: 0151
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Have you thought adding a change to translate contents of generated html? It would be nice, if ccdoc would read certain words like "class", "contents" etc. from a file which would be kind of resource file. If you change contents a file to f. ex. finnish, then you would have finnish documentation!

The file (simple ex.) could have key-string pairs:

file: ccdoc.en.language
  CLASS=class
  DESTRUCTOR=destructor

file: ccdoc.fi.language
 CLASS=luokka
 DESTRUCTOR=tuhoaja
Then you could generate english documentation: ccdoc -langfile ccdoc.en.language ... and finnish: ccdoc -langfile ccdoc.fi.language ...
Solution Description:

Title: Ignore duplicate macros names.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/17
Resolved By: Joe Linoff
Resolved On: 2003/02/26
Id: 0152
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
This occurs when the -rptmac switch is specified and windows DLL import/export macros are defined in header files as import by default without ccdoc guards.
Solution Description:
Originally fixed in r34 (2003/02/18) and verified by test 71.

Fixed again in r39 and verified by test 86 because I found another problem related to this when the same macro was defined in different packages.


Title: Add a heuristic to ignore include guards and DLLIMPORT macros.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/17
Resolved By: Joe Linoff
Resolved On: 2003/02/18
Id: 0153
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Add a heuristic that ignores certain macros under switch control. Documentation for include guards and windows DLLIMPORT/DLLEXPORT macros do not help large systems.

This heuristic would ignore macros named with the following rules:

  1. Macros with a _h or _H suffix (include guards).
  2. Macros with a _hh or _HH suffix (include guards).
  3. Macros with a include_ or INCLUDE_ prefix (include guards).
  4. Macros with a _include or _INCLUDE suffix (include guards).
  5. Macros with a _dll or _DLL suffix (DLL import/export).
  6. Macros with a dll_ or DLL_ prefix (DLL import/export).

In the following example only the DBG macro would be documented:

#ifndef file_h
#define file_h

#ifdef WINDOWS
#include 
#  ifndef file_dll
#    define file_dll DLLIMPORT
#  else
#    define file_dll DLLEXPORT
#  endif
#endif

/**
 * This macro defines the debug prefix.
 */
#define DBG cerr << "DEBUG:" << __FILE__ << ":" << __LINE__ << " " 

#endif
I propose that the switch be called -rptmac1 (report macros heuristically) in phase 1.
Solution Description:
Fixed in r34 and verified by test 72.

This is my best guess at reasonable macro names. In the future it might be better to make this file driven and use these as the default if the file is not specified.


Title: Eliminate the requirement that the -rptmac and -rptmac1 switches appear in phase 1.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/18
Resolved By: Joe Linoff
Resolved On: 2003/02/19
Id: 0154
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Requiring that the -rptmac and -rptmac1 switches appear in phase 1 and phase 3 processing is inconsistent and could lead to user confusion and unexpected results. These switches should only be required in phase 3.
Solution Description:
Fixed in r35.

Title: Incorrect documentation when there is only one ccdoc namespace comment for namespace declarations in multiple files.
Status: FIXED
Reported By: Sakari Forsberg
Reported On: 2003/02/18
Resolved By: Joe Linoff
Resolved On: 2003/02/19
Id: 0155
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
When a namespace is defined in more than one header file but is only documented in one place, ccdoc auto generates comments and puts the user documentation in a table. The test case below illustrates the problem when the ccdoc is generated.
-----------------------
// file1.h
/**
 * This is the namespace documentation.
 * @author Author
 * @version 1.0
 */
namespace Test {
   int fct1();
}

-----------------------
// file2.h
// No ccdoc documentation here. It should derive
// the documentation from file1.h
namespace Test {
   int fct2();
}
Solution Description:
Fixed in r35, verified by test 75.

Title: More namespace problems in r35.
Status: FIXED
Reported By: Sakari Forsberg
Reported On: 2003/02/19
Resolved By: Joe Linoff
Resolved On: 2003/02/19
Id: 0156
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Namespace processing gets confused in certain cases.

Case #1: Lost N1::Class1, N2::Class2 is reported as N1::Class2

In the case below, the N1::Class1 is completely lost during processing.
namespace N1 {
   class Class1
   {
   public:
      Class1();
      ~Class1();
   };
}
namespace N2 {
   class Class2 : public N1::Class1
   {
   public:
      Class2();
      ~Class2();
   };
}

Case #2: N1::Class1 is reported correctly but N2::Class2 is reported as N1::Class2

The only difference between this case and the one above is the presence of ccdoc comments.
/**
 * This namespace should contain Class1.
 * @author Author
 * @version 1.0
 */
namespace N1 {
   class Class1
   {
   public:
      Class1();
      ~Class1();
   };
}
/**
 * This namespace should contain Class2
 * with a reference to N1::Class1.
 * @author Author
 * @version 1.0
 */
namespace N2 {
   class Class2 : public N1::Class1
   {
   public:
      Class2();
      ~Class2();
   };
}
Solution Description:
Fixed in r36, verified by test 76.

This was a critical bug that warrants a new release.


Title: In some cases operator code is being processed incorrectly.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/19
Resolved By: Joe Linoff
Resolved On: 2003/02/19
Id: 0157
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In some cases the body of an operator is parsed incorrectly because there was a portion of code in the parser that exited prematurely before processing the body.

This has been fixed and will be released in r37 later this week.

Solution Description:
Fixed in r37, verified by test 69.

Title: Class method implementations are reported as global functions.
Status: FIXED
Reported By: Richard Durwin
Reported On: 2003/02/19
Resolved By: Joe Linoff
Resolved On: 2003/02/19
Id: 0158
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
In the following code example ccdoc reports the class method implementations as global functions.
class Test {
public:
   Test();
   Test(const Test& obj);
   ~Test();
   Test& operator=(const Test& obj);
   void Set(int v);
   int Get() const;
private:
   int m_a;
};
namespace {
  const char* rcsid = "$Id: issues.html,v 1.35 2004/09/30 16:09:24 jlinoff Exp $";
};
Test::Test()
  : m_a(0)
{
}
Test::Test(const Test& obj)
  : m_a(obj.m_a)
{
}
Test::~Test()
{
}
Test& Test::operator=(const Test& obj)
{
  m_a = obj.m_a;
  return *this;
}
void Test::Set(int v)
{
  m_a = v;
}
int Test::Get() const
{
  return m_a;
} 
Solution Description:
Fixed in r37, verified by test 77.

Modified the parser to ignore these.


Title: Very long names make the contents table very hard to read.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/20
Resolved By: Joe Linoff
Resolved On: 2003/02/20
Id: 0159
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
When ids are very long it makes the contents table very difficult to read.
Solution Description:
Fixed in r37, verified by tests 78,79,80.

Added a switch called -rptmlcei <num> (maximum length of the contents entity id) that limits the size of the entity id. When the string exceeds that length, only the first <num> characters are printed followed by .. to indicate that it has been truncated. The default length is 32. If no inherited from column exists, the value of the -rptmlcifi is added to make this field bigger.

Also added a switch called -rptmlcifi <num> (maximum length of the contents inherited from id) that limits the size of the inherited from column. When the string exceeds that length, only the first <num> characters are printed followed by .. to indicate that it has been truncated. The default length is 32.


Title: Empty function names appear under certain conditions.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/20
Resolved By: Joe Linoff
Resolved On: 2003/02/20
Id: 0160
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Empty function names appear when the following syntax is encountered:
#ifdef FOO
int Fct(int arg)
#else
long Fct(long arg)
#endif
{
  return arg;
}
Solution Description:
Fixed in r37, verified by test 81.

This occurs because ccdoc keeps around the pre-processing directives. The current solution is to print a warning and let the user fix the problem.


Title: Strange function names like "max)" appear sometimes.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/20
Resolved By: Joe Linoff
Resolved On: 2003/02/20
Id: 0161
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
While parsing some STL header files I came across the following construct which caused ccdoc to become confused about the method name. It reported the name as "max)" instead of "max".
class Test {
public:
  static double (_STLP_CALL max)() _STLP_NOTHROW { _STLP_USING_VENDOR_CSTD return DBL_MAX; }
};
Solution Description:
Fixed in r37, verified by test 81.

This was occurring because the get_fct_name method in phase 1 parsing was getting confused under certain conditions.


Title: Doxygen compatibility - support single line comments using doxygen syntax.
Status: FIXED
Reported By: Richard Durwim
Reported On: 2003/02/21
Resolved By: Joe Linoff
Resolved On: 2003/02/21
Id: 0162
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Single line comments use a different flag. That's a lot of edits to make!
I am all in favor of making ccdoc and doxygen style comments compatible so that users can use both of them. I will add support for this in the next release.

Here are the flavors that ccdoc will support:

int var; /*!< Detailed description after the member */
int var; /**< Detailed description after the member */
int var; //!< Brief description after the member
int var; ///< Brief description after the member
Solution Description:
Fixed in r37, verified by test 82.

Title: Doxygen compatibility - ignore the @endlink directive.
Status: FIXED
Reported By: Richard Durwim
Reported On: 2003/02/21
Resolved By: Joe Linoff
Resolved On: 2003/02/21
Id: 0163
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
@link uses @endlink to delimit it, rather than the end-of-line. The @endlink appears in the ccdoc output.

I am all in favor of making ccdoc and doxygen style comments compatible so that users can use both of them. One possible solution is to modify ccdoc to ignore the @endlink directive. This would allow users to have links in both systems but they would have to exist on the same line for compatibility.
Solution Description:
Fixed in r37, verified by test 82.

Modified ccdoc to ignore the @endlink directive.

In r38 updated ccdoc to officially recognize the @endlink directive as part of normal processing.


Title: Doxygen compatibility - ignore @file comment blocks.
Status: FIXED
Reported By: Richard Durwim
Reported On: 2003/02/21
Resolved By: Joe Linoff
Resolved On: 2003/02/21
Id: 0164
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Doxygen uses @file to document files. You could just ignore this flag. At the moment it appears in ccdoc documentation.
I am all in favor of making ccdoc and doxygen style comments compatible so that users can use both of them. One possible solution is to modify ccdoc to ignore the @file directive.
Solution Description:
Fixed in r37, verified by test 82.

Modified ccdoc to ignore the @file comment blocks.


Title: Doxygen compatibility - support virtual grouping of definitions.
Status: OPEN
Reported By: Richard Durwim
Reported On: 2003/02/21
Resolved By:
Resolved On:
Id: 0165
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Doxygen allows a virtual grouping of definitions that is very useful (@name and @{).
I am all in favor of making ccdoc and doxygen style comments compatible so that users can use both of them. This change is a challenge because the syntax conflicts with a current ccdoc usage. I will look into it when I can.
Solution Description:

Title: Variable values are not reported properly when braces occur in the value.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/23
Resolved By: Joe Linoff
Resolved On: 2003/02/23
Id: 0166
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The variable value is not reported properly in the following case:
static const char* s_rcsid = {"value"};
The resulting variable declaration strips out the {"value"} so it looks like this:
static const char* s_rcs_id =
The good news is that the variable is correctly identified and the comments are correctly associated. The value is the only thing missing.
Solution Description:
Fixed in r38, verified by tests 2 and 83.

Ccdoc was discarding what it thought was the function body for this special case.


Title: In German we want to use embedded dots in the short description.
Status: OPEN
Reported By: Christian Ehrlicher
Reported On: 2003/02/25
Resolved By:
Resolved On:
Id: 0167
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I've found: another little problem. In German we write for the short form of "for example" "z.B.". ccdoc ignores silently the rest of the line after the B:

--> in .h : /** das ist z.B. test */
--> in .html : "das ist z.B" without the last point an the rest of the line


Your example should work. This is a bug in ccdoc. It is supposed to terminate the short description after it encounters a dot that is followed by white space.

For now you can workaround it by embedding trailing spaces before the last dot as shown below:

--> in .h : /** das ist z.B . test */
--> in .html : "das ist z.B. test"

Solution Description:
Fixed in r39, verified by test 84.

The short description string size was not updated properly in ccdoc::phase1::scanner_doc::add_line.


Title: Slow g++ 3.2 compile of help.cc because of the large help string.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/25
Resolved By: Joe Linoff
Resolved On: 2003/02/25
Id: 0168
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Very slow compilation of help.cc using g++ 3.2 because it tries to verify the format for the very large help string.
Solution Description:
Fixed in r39.

Modified the script that generates the help text in help.cc to break it up periodically. This improved compile times by more than 100X on my system.


Title: Man page has formatting problems.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/25
Resolved By: Joe Linoff
Resolved On: 2003/02/25
Id: 0169
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
The automatically generated man page has formatting problems.
Solution Description:
Fixed in r39.

Modified the script that generates the man page to recognize the constructs that were missed in the previous version.


Title: Bug on NT platforms parsing command line with spaces and wildcards.
Status: FIXED
Reported By: Gonzalo Matamala Torres
Reported On: 2003/02/26
Resolved By: Joe Linoff
Resolved On: 2003/02/26
Id: 0170
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Hello (and sorry by my poor english)!

I have found a bug over NT platforms parsing command line with spaces and wildcards:

Unfortunately, the code I'm documenting has white space in paths. (It's a horrible name coding, I know it!). But I need document it. For example, library code in:

smartcard components\scot

when I try, for example:

ccdoc -db example.db "smartcard components\scot\*.h"
ccdoc try resolve wildcard with (switches.cc, line 411):
cmd = "DIR /B /O:N " + file + " > " + tmp + "\n";
that result (in my example):
DIR /B /O:N smartcard components\scot\*.h
and dir command try list "smartcard" directory first and "components\scot\*.h" after Then, fail because these paths aren't found.

A solution is using -files switch of course. But I think the above line can be modified to:

cmd = "DIR /B /O:N \"" + file + "\" > " + tmp + "\n";
that result:
DIR /B /O:N "smartcard components\scot\*.h"
(double quotes are added to path) and then it's a success.
2003/02/26 jdl
Your English is excellent and so is your solution.

Thank you for taking the time to thoroughly analyze the problem and propose a solution. I will add this fix to the v08r39 release so it should be available to you next week.

Solution Description:
Fixed in r39.

Manually tested.


Title: Constructors with constructed default arguments are reported with the wrong method name.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/02/26
Resolved By: Joe Linoff
Resolved On: 2003/02/26
Id: 0171
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Constructors with constructed default arguments are reported with the wrong method name as shown in the example below:
class Test
{
public:
   typedef Allocator<Property*> allocator_type;
   typedef Allocator<Association*> allocator_type1;
   Test( const allocator_type& x = Allocator<Property*>() ) : m_val(0) {}
   Test( const allocator_type1& x = allocator_type1() ) : m_val(0) {}
private:
   int m_val;
};
Solution Description:
Fixed in r39, verified by test 85.

Modified ccdoc::phase1::parser::get_fct_id to correctly recognize the constructor situation.


Title: Template friend declarations are not resolved properly when packages are used.
Status: FIXED
Reported By: Joe Linoff
Reported On: 2003/03/11
Resolved By: Joe Linoff
Resolved On: 2003/03/11
Id: 0172
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Template friends are not resolved properly when packages are used. They work correctly when packages are not used. Here is some source code that demonstrates the problem:
template <class T>
class A {
   friend class B<T>;
   T m_val;
};

template <class T>
class B {
   T m_val1;
   A<T> m_val2;
};
Here is the ccdoc command that triggers the problem:
% ccdoc -db foo.db *.h -pkg test -index -html html/
If the -pkg switch is removed, the reference is found correctly.
Solution Description:
Fixed in r40, verified by test87.

The phase1 parser was not correctly stripping the names to eliminate template arguments.


Title: The md5hash macro is being improperly ignored by the rptmac1 heuristic.
Status: FIXED
Reported By: Alexander Filonov
Reported On: 2003/04/09
Resolved By: Joe Linoff
Resolved On: 2003/05/15
Id: 0173
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I've encountered a problem with ccdoc using -rtpmac1 heuristics:

Macro called "md5hash" will not appear in final html code.

I see two solutions for this:

  1. fix heuristics algorithm
  2. include option for turning on and off command-line options by innline text commands. (or may be such option already exists? If so, then I'm sorry for buzzing)
Solution Description:
Fixed in r40, verified by test72.

The backwards comparison heuristic was not working properly. The option for telling ccdoc to ignore inline text commands already exists. You can use the predefined __ccdoc__ macro for this as follows:

#ifndef _include_guard 
#  ifndef __ccdoc__
#    define _include_guard 
#  endif
#endif

Title: Bug#195282: ccdoc: FTBFS with g++-3.3
Status: FIXED
Reported By: Magnus Ekdahl
Reported On: 2003/05/30
Resolved By: Joe Linoff
Resolved On: 2003/06/05
Id: 0174
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Package: ccdoc
Version: 0.8.38-1
Severity: serious
>From my build log:

...
source='statement.cc' object='statement.o' libtool=no \
depfile='.deps/statement.Po' tmpdepfile='.deps/statement.TPo' \
depmode=gcc3 /bin/sh ./depcomp \
g++ -DHAVE_CONFIG_H -I. -I. -I.  -DCCDOC_CID=\"/usr/bin\"   -g -O2 -c -o statement.o `test -f 'statement.cc' || echo
'./'`statement.cc
statement.cc: In static member function `static bool
   ccdoc::statement::base::is_rptmac1_id(const char*)':
statement.cc:987: error: brace-enclosed initializer used to initialize `const
   char*'
statement.cc:987: error: brace-enclosed initializer used to initialize `const
   char*'
...
make[2]: *** [statement.o] Error 1
make[2]: Leaving directory `/tmp/buildd/ccdoc-0.8.38/src'
make[1]: *** [all] Error 2
make[1]: Leaving directory `/tmp/buildd/ccdoc-0.8.38/src'
make: *** [build-stamp] Error 2

-- System Information:
Debian Release: testing/unstable
Architecture: i386
Kernel: Linux frobnitz 2.4.21-pre5 #1 Sat Mar 1 09:01:10 PST 2003 i686
Locale: LANG=C, LC_CTYPE=C
Solution Description:
Fixed in r41.

The array initialization was incorrect (in r40), g++ 3.3 was correct in reporting it.


Title: Generate an index in HTML and XML.
Status: OPEN
Reported By: Joe Linoff
Reported On: 2003/06/05
Resolved By:
Resolved On:
Id: 0175
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
Several users have requested a search capability in ccdoc. This capability can be provided in two ways: use a standard HTML based search engine on the generated files or provide a database of indexed items that can be used for a custom search engine. Although the standard HTML based search engine approach works well, there are still some users who would like a pre-generated index of searchable items.

My current thinking about how to implement this is to provide a -rptidx switch for phase 3 that generates two HTML indices: one by name and one by entity type by name.

It would also generate an XML based index file that could be used by 3rd party search engines. Each XML record would contain the name of the entry (i.e. GetName), the type (method, function, etc.) and the associated URL. Multiple entries with the same name would be allowed.

It might also be desirable to have it contain information about the scope of the name. This means that each enclosing namespace and class might be specified as list. Consider the case of the X::Y::A::B::GetName() method where X and Y are namespaces, and B is a nested class inside of A. The scope information would contain data like ((nsp X) (nsp Y) (cls A) (cls B)).

Yet another desirable attribute would be to have this index view of the data tie in to the XML output that is being generated in the proto-type version for the next major release. In this release the database format will convert to XML. The phase 2 processing will generate fully cross-indexed entities that are used in phase 3 to generate HTML. Of course the user can stop at phase 2 and work directly with the XML. This is different than today where the database format is NOT XML and it is not fully cross referenced.

Solution Description:

Title: Enhance the table of contents for easier navigation with letter based links at the top.
Status: OPEN
Reported By: Ragav Satish
Reported On: 2003/05/23
Resolved By:
Resolved On:
Id: 0176
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
This suggestion was actually made some time ago but I just got around to adding it.

Ragav suggested that ccdoc be enhanced to support quick links to entries in the table of contents. The idea is that for tables with a large number of classes (or methods), the user could click on a quick link to the region they were interested in.

In practice it would look something like this:

Table Of Contents

Quick Links: A B C D E F G H ... Z

.
.
.
I think that this would be very useful for large systems that have hundreds or even thousands of classes.
Solution Description:

Title: Invalid link from parameter variable name to member variable.
Status: OPEN
Reported By: Chris Rogers
Reported On: 2003/06/18
Resolved By:
Resolved On:
Id: 0177
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I have found something wrong with my output though - not sure if this is a bug or if I am doing something wrong. I have the following (simplified) situation:
class SOMECLASS
{
public:
    void  DoSomething(int x);
    int x;
};
You'll see that I have a member variable 'x' and also a member function which has 'x' in it's argument list. This causes CcDoc to place a link from the argument name 'x' to the documentation for the member variable 'x'.

I could simply change one of these names to ensure that there is no clash, however I thought you'd like this kind of problem submitted as a bug.

Solution Description:

Title: Namespace/anonymous enum interaction bug.
Status: OPEN
Reported By: William A. Rivet
Reported On: 2003/06/26
Resolved By:
Resolved On:
Id: 0178
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I have a file as_config.h:
#ifndef AS_CONFIG_H
#define AS_CONFIG_H

namespace ApplicationInterface
{
   enum 
      {  zero=0,
         one=1,
         two=2
      };
   enum {
      number_of_tasks         = 60,
      task_stack_size         = 4096,
      number_of_queues        = 32,
      number_of_event_groups  = 32,
      number_of_timers        = 32
   };
   
} // namespace ApplicationInterface
#endif

The problem is that I see two "$anonymous$" enum entities in the namespace page, but both link only to a page that lists only the first enum.

I know that in this case the enums could be combined, but this seems less than satisfactory as a general solution. Is this a known issue? (I couldn't find one that seemed to match on the ccdoc website) Is there interest in fixing this? I have no idea how hard it is to change this. Either way I likely will have to do something to at least address this in-house.(perhaps just complain when the second enum is detected...this would at least allow us to detect the problem and perhaps "invent" a name for at least one of them).

Solution Description:

Title: Add a new @entity directive to allow referencing.
Status: OPEN
Reported By: Steve Perry
Reported On: 2003/07/15
Resolved By:
Resolved On:
Id: 0179
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
It would be nice if ccdoc had a directive like "@entity <entityName>".

Here is how I would want to use it. I have a macro which declares and defines a static member function. It is a macro because every class in the hierarchy should have this function (it returns an "interface ID"). The code currently looks like this:

  //@{
  // Returns the ID value for this interface.
  //@}
  Gf_DECL_INTERFACE_ID("{8DC3F1C6-F7CA-42d8-AFD6-7AF950C5C11E}")
# if defined(__ccdoc__)
  static Gf_GUID InterfaceID();
# endif
It would be nicer if I could write:
  //@{
  // Returns the ID value for this interface.
  // @entity static Gf_GUID InterfaceID()
  //@}
  Gf_DECL_INTERFACE_ID("{8DC3F1C6-F7CA-42d8-AFD6-7AF950C5C11E}")

If there is already a way to do this, I'd appreciate hearing about it. If not, please consider adding a feature of this sort sometime in the future.


I can see where it would be useful to have real function name in the table of contents. The #ifdef workaround will certainly work until this issue is addressed.
Solution Description:

Title: Limit the length of generated file names.
Status: OPEN
Reported By: Willie Sakai
Reported On: 2003/08/15
Resolved By:
Resolved On:
Id: 0180
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I've been happily using CCDOC for quite some time, but I'm stumped on how to control the length of the generated HTML filenames. I'm using nested packages and the generated filenames are so long that some filenames get truncated when written to a CD-ROM. This results in broken links when the documentation tree is copied from the CD-ROM.

I've tried using the -maxpathlen option during the (-htm) output generation phase, but this doesn't shorten the filename length.

Solution Description:

Title: Copy constructor documented incorrectly under some conditions.
Status: OPEN
Reported By: Charles Doland
Reported On: 2003/08/15
Resolved By:
Resolved On:
Id: 0181
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I would like to report a minor issue with ccdoc, described below.

The documents created by ccdoc indicate that a constructor will be generated automatically by the compiler even if one is specified. This seems to occur when the constructor for class X is specified as X(X const &), but does not occur when the constructor is specifed as X(const X &), although these are equivalent.

Solution Description:

Title: Add directory tree/outline expansion (+/-) capabilities to better organize the information.
Status: OPEN
Reported By: Brian Alexander
Reported On: 2004/06/04
Resolved By:
Resolved On:
Id: 0182
[Top]
[Summary]
[Next]
[Previous]
Problem Description:
I was going through the ccdoc for [one of my classes] this morning when I had a couple of feature ideas for ccdoc. I was thinking that it would be nice if overloaded methods in a class could be collapsed and expanded so that you only saw one line with the function name. If all of the short descriptions matched for the method, you could see the short description - if the short descriptions didn't match you would see something like "6 overloads". E.g. you might see something like this:
 +   GetClient                       Xdm_ObjectDerived       method protected   16 overloaded methods (click '+' to expand)
When you click the + to expand you would see the display much like it is today with each overload displayed on its line with its short description. At this point the '+' would turn into a '-' and clicking the '-' would collapse it back down.

Extending that idea, I thought it would be nice if we had an extra tag that we could tag methods with to categorize them - E.g. "@category Debug Helper". All methods with matching categories would go into an expandable / collapsable grouping like the overloaded methods. (If there's just one or two methods in a category maybe it defaults to expanded but more than two and the category defaults to collapsed.) Methods with no category would be displayed as they are today without the expand/collapse capability. Methods within a category are alphabetized within the category. We would also probably want links at the top and bottom to 'expand all' and 'collapse all' - maybe throw in some cookies to store the user's preferences (overall and for each class) - hah.

Then I also thought that it was somewhat annoying with the ccdoc for some classes - Xdm_Model is one example - where there is a lot of good documentation for the class but it's a pain to have to scroll through all of that to try and find the method index so I can see if it supports the method I'm looking for. Having all of the class documentation and examples are great for the first time you're using a class but for quick reference which is most of the time after the first time you've used a class, all of that stuff gets in the way. If classes could have extra tags to separate things out like @Examples, @Extending, @UserGuide or something like that then in the ccdoc, you could either have expand / collapse things like above to expand out the examples text - or you could have links that take you to the examples for a specific class etc. I like the expand/collapse symbols just to keep everything on one page but it would take some javascript or something I'm sure and may be more pain than it's worth.

Anyway, these were just some pie in the sky ideas I had this morning. I didn't think them through a whole lot - just wanted to send them off to you to see if you liked them.


These are very interesting ideas. I will definitely look into them for future releases.
Solution Description:

Title: Ccdoc started failing with a segment violation in phase 3 when processing a large system (thousands of classes) after a small change to one of the header files.
Status: FIXED
Reported By: Xilinx
Reported On: 2004/09/16
Resolved By: Joe Linoff
Resolved On: 2004/09/29
Id: 0183
[Top]
[Summary]
[Previous]
Problem Description:
Ccdoc started failing with a segment violation in phase 3 when processing a large system (thousands of classes) after a small change to one of the header files. This system had been running for years without a problem.
This is a critical bug.
Solution Description:
Fixed in r41, verified by test88.

There was a particular order of entities in the class that was causing consternation in the compare method used during the sorting of the table of contents in phase3_html.c. I was able to duplicate the problem (test88) and then fix it by using a std::stable_sort() instead of a std::sort(). It is interesting to note that this failure occurred on solaris and linux but it did not occur on windows.


[Top]
This page is maintained by Joe Linoff.
Last updated: $Date: 2004/09/30 16:09:24 $

This page was automatically generated by issues.pl.