It is now possible to specify groups of users, as another way of reducing per-user graph clutter. The userdata will have a line displayed per-group rather than per-user.
The -cvs-global-args switch was added so CVS global switches can be set for all CVS commands cvsplot executes.
The per-user data plots were broken since the 1.7.2 release. This has now been fixed.
Fixed minor bug in 1.7.0, which contains the following new features:
* Added per-user CVS statistics, with initial work provided by Jacob
Schroeder <firstname.lastname@example.org> and Tim Wood
<email@example.com>, but extended so that it properly
handles files that are removed, then re-added, and so that it is
possible to see the CVS contributions performed by individual users
plotted using gnuplot. The -userdata option will output per-user
information into the specified file. By default, this will record
data for all CVS users found, however this can be limited by using
the -userlist option, if there are only a limited number of users of
interest.... read more
* Made a number of changes to support CVS repositories running under
Windows, and for running cvsplot against these repositories. The
format of Window CVS diffs are slightly different to cause a number
of minor changes, plus the script had a number of UNIXisms in it
which needed to be removed to make the script cross-platform.
Note the line count may change slightly from previous versions.
Prior to this version, the "wc -l" command was used to count lines
for 1.1 files, however this command counts the number of _newline_
characters found, which is slightly different to the number of lines
present in a file. The correct count is now used.... read more
Made the cvs log parser more flexible. It seems not all cvs logs have "state: Exp;", but sometimes have "state: Rel;" when a new file is added. Accepted patch from Era Eriksson which fixes this. This was initially detected when running cvsplot over spamassassin. Added the -linestyle argument, which generates plots using lines, rather than discrete points. For large-range plots, this can improve readability.
* Fixed typo when handling non-standard version numbers. For normal repositories, this won't make any difference.
Incorporated patches from several contributors. Its possible to generate a single plot containing both the line and file statistics, to handle CVS repositories with non-standard revision numbers (ie, head starting at 8.1) and to use the ShellQuote module for better handling of unusual filenames. Also improved the error handling if a CVS command fails.
Fixed bug with using -rlog and -cvsdir when the directory was remote. The generated statistics weren't correct in this case.
Added the -rlog option so that it is now possible to gather statistics without requiring a checked out copy of the module of interest. This option requires CVS version >= 1.11.1.
Cvsplot will now automatically exclude files that are indicated as being binary to CVS.
Cvsplot has been modified so that it works now for very large CVS histories without running out of stack space. Also accounted for CVS repositories where file versions don't necessarily start with revision 1.1.