Managing release numbers and merges in CVS requires quite a bit of repetetive, fault-prone work. Sometime back, I scripted some of the more predictable things I ran into for a release strategy which included releases to "QA" in a branch for patches and bug fixes while new development continued in another branch (or more than one branch for multiple lines of development). The script also calls a build script and scrapes the results for failed compiles, scans CVS output for files that need to merge and various other things I've managed to screw-up manually at one point or another. I'm posting it for comments, so try it out and I will appreciate any feedback you might have if you see something that should be fixed or improved. Here's the script (perl) and a walk-through of some of the merge and release process. (These are the examples I mentioned in the presentation in Austin Feb. 3). If there is enough interest, I might post this to Freshmeat for the amusement of others.
Posted by bill at February 24, 2004 12:02 PM