Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
On Fri, 27 Feb 2004 12:24:21 +0100, Oliver Giesen <ogware at gmx.net> wrote: >Tony Hoyle wrote: > >> It was added in 2.0.21, after it was pointed out you could completely >> trash your repository by typing the wrong import command. Repository >> integrity overrides any convenience of using the same tag. > >Have you read the post by Phil Richards on this thread? It actually >looks as if it is just the other way round: Using *different* vendor >branches on subsequent imports is what messes up the repository (unless >combined with the -b option, I guess)! Currently this is exactly what >you effectively _enforce_ by forbidding to use the same branch twice... It can only work if a) You use the same branch name, and b) you use the same branch number Anything else will cause things to go wrong to a lesser or greater extent. That's easy to check for (2.0.32 does this rather than rejecting out of hand). > >> One day if there is more info stored about tags then CVS might be able >> to tell the difference between a vendor tag and a branch tag, > >Why does there have to be a difference at all? So far I have been >working happily in the belief that the only thing that's special about >the vendor _branch_ created by Import was its revision number and the >fact that it was implicitly declared as the default branch. It's more that you don't want to be able to do 'import repos my_development_branch' and have the whole repository screwed up. A way of detecting vendor branches would allow CVS to avoid deleting existing branches whilst allowing you to update vendor branch tags to any value you like. (You could in fact detect a vendor branch tag because it's not a true branch tag... it's in the wrong format as it has 3 digits (1.1.1) instead of a magic branch tag (1.1.0.1)) >Of course I could probably get the same functionality by importing my >third-party sources to the trunk with the -n option (or doesn't that >work anymore too? - after all it's effectively also always the same >branch that I'm importing to...) and doing my own modifications on a >branch but then the other way round is how I started so it would be nice >if I could just continue to do so as there never was any problem with >that process AFAICS. That's effectively what you're doing anyway. Import just wraps it. Tony