In article <[EMAIL PROTECTED]>, Schwenk, Jeanie wrote: >I received this from one of our engineers here (I added a few details for >clarity). I think the merge behaved this way because of the order of the >tags. Is that correct? > >=============================================== >I tried several times to do the following: > >cvs update -j systema_v1_2 EQC_RobotArm.java > >Where the version in my current working directory and the systema_v1_2 (a >branch) version were both beautified (format). This merge never worked, >with the conflict file saying that nearly the entire files were different.
Since the files are being pumped through some beautifier, it's not surprising that they are considered different. Merging is done in a line oriented way. The slightest difference between two lines, even whitespace, makes them completely different. This experience will hopefully teach your so-called engineer to avoid making unproductive changes to source code. >So I tried something different. I checked in my current, beautified >version, so that my version and the branch were both beautified and both >committed. Then I did this: > >cvs update -j HEAD -j systema_v1_2 EQC_RobotArm.java > >The results of this were bizarre. These files were merged. Where there >were conflicts, the contents of systema_v1_2 were taken only. No conflict >file was created. ViewCVS's diff clearly showed the conflicts ... I know >they are there. You can't expect any help when you have entangled some beautification tool into the semantics of the merge. If you can reproduce the problem without the tool, then you have an issue. If, on the branch, you did a few changes and then beautified the source, and on the trunk if you beautified only, it's possible that this will get resolved with few conflicts, or maybe even with none. Beautification could hide conflicts, such as ones introduced by whitespace changes that are undone by the formatting. _______________________________________________ Info-cvs mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/info-cvs
