Max Bowsher wrote: > Michael Haggerty wrote: >>> Max Bowsher wrote: >>>> Ross Golder wrote: >>>>>> On 2005-11-09 at 21:53 +0100, Michael Haggerty wrote: >>>>>>> By the way, a cowardly (but probably relatively robust) way to deal with >>>>>>> such situations would be to rename the file in the attic to something >>>>>>> like repo/a/Attic/Attic-file-f.txt. That would allow cvs2svn to run to >>>>>>> completion and would at least leave some kind of evidence in the svn >>>>>>> repository of the situation. >>>>>> That's the conclusion I came to. So I hacked my copy of cvs2svn to do >>>>>> something like this. >>>> This is probably worth adopting upstream - although, with a non-random >>>> filename, I think. >>> >>> Do you have something in mind for "non-random"? Should the user have to >>> choose a name for each duplicated file? (Maybe be a mapping file?) Or >>> a user-specified regexp substitution? Or maybe there should simply be >>> an "Attic" subdirectory in subversion containing the conflicting Attic >>> files? > > I had been thinking of prepending "Attic-" to the filename. > > Handle the unlikely situation there really being an "Attic-foo,v" file > already in existance by checking for both "Attic/Attic-foo,v" and > "Attic-foo,v", and retrying with a prefix of "Attic-2-", etc. > > But, the "Attic" subdirectory idea could work well too.
This is implemented (using an "Attic" subdirectory) in trunk r2971. Michael _______________________________________________ Gnome-infrastructure mailing list [email protected] http://mail.gnome.org/mailman/listinfo/gnome-infrastructure
