In article <[EMAIL PROTECTED]>, Greg A Woods (gaw) writes: gaw> It also depends a _GREAT_ deal on what the file contains. Jar files do gaw> not contain easily diff-able and patch-able text and therefore are not gaw> appropriate for storage in CVS.
Bollocks. The core task of CVS is version control. The abiliy to get a diff is nice but way, way down the list. I want to be able to build the system as it was last week. That means version control on all files used in the build where that is at all possible. I'd shove system libraries and the kernel in there if I could. Certainly I want third party libraries if I can and test data always. >> As for distributing things by some other mechanism, why set up another >> system when CVS versioning is usually exactly what you need (so you >> can always go back and build an older version). gaw> Becasue CVS is _NOT_ appropriate to use for just distributing gaw> non-text files. Versioning versioning versioning. It's the concurrent versioning system, not the concurrent patch generating system. If you distribute non text things by, say, rsync you then need to set up a parallel versioning system just for those files, maintain it, make sure people use it and keep it in sync with CVS, all because of some superstition that you want to be able to generate diffs for everything in the CVS repository. -- Mail me as [EMAIL PROTECTED] _O_ |< _______________________________________________ Info-cvs mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/info-cvs