Re: Patching and Releasing

[James Youngman]

Kent Henneuse <address@hidden> writes:

> I am currently struggling with some of the "administrative" duties
> of CVS.  I want to be able to determine the changes that have
> happend on a branch between two different dates.  This will then be
> used to make up a patch that can be added to fix defects.  Currently
> I have only been able to do some of this using Perl scripts and the
> CVS/Entries file.  When I have tried to use cvs diff with two -D
> args or -r and -D I get differences between the branch and HEAD.  I
> am sure there is a more logical command set to accomplish this task
> that can be done using cvs.


I have two shell scripts, make-source-release and
make-binary-release.  The former tags the main trunk and then exports
the tagged code.  It then tars this up and stores this in an archive
directory. It also does "cvs rdiff -s" to produce a summary of what
changed in this release (it parses the output with, erm, Awk I
think).  

The make-binary-release script picks up the generated tar file,
configures and builds the code and installs it into a temporary
directory.  It then calculates the MD5 checksup of all the installed
files.  A compressed tar file is made of the installed files. 

This data is merged with the MD5 checksum data from the previous
release.  This is used to produce a list of new and changed files.   A
compressed tar file is generated containing only the changed files. 

A human being then makes the decision as to which file is released
(i.e. is this a "patch release" or a "full release"?).  

Outstanding problems with this process are that on Solaris you can
compile the same thing twice and get a different result.  This only
happens with files compiled with debug info, but I have a policy of
not shipping stripped binaries.  



-- 
James Youngman
Manchester, UK.  +44 161 226 7339
PGP (GPG) key ID for <address@hidden> is 64A95EE5 (F1B83152).