Tracking branch points makes it easier to fold changes from branches back into ancestor branches. By identifying a common ancestor (the branch point) in addition to a contributor (e.g. a maintenance branch) and a target (e.g. the HEAD), a 3-way merge is done by "cvs update -jbranchPoint -jbranch". This usually turns out to be easier and better than what you get by supplying a single -j option with the branch.
>--- Forwarded mail from [EMAIL PROTECTED] >From reading some of the docs about NetBSD's and FreeBSD's >release processes, I have noticed that they both use what >FreeBSD calls "branch point" tags. Not having a lot of >experience with CVS, I don't quite understand what problem >is solved by doing this. The FreeBSD release docs claim >that it is impossible to diff against the start of a branch >w/o a branch point tag in place. >The process appears to look something like this: > 1. Create branch point tag > cvs rtag <tag>_BP src > 2. Create the actual branch > cvs rtag -b -r<tag>_BP <tag> src >Is this an old CVS quirk that has been fixed, or does this >problem still exist? What _exactly_ is the problem? An >example or ideas on how I could reproduce the problem in my >own CVS environment would be welcome. >--- End of forwarded message from [EMAIL PROTECTED] _______________________________________________ Info-cvs mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/info-cvs
