David Roundy writes:

 > Given our previous troubles, I can't help but wonder if this removal is
 > worthwhile.  It's probably safe, but that's what I thought about your last
 > patch.  And the absolute safest approach is to not change how we parse
 > dates, which is the conservative approach I lean towards, since we have
 > absolutely no way to know for certain what's present in older
 > repositories.

It's always a bad idea to break such a promise; as you say you just
can't know.  With something like dates that humans manipulate a lot,
you might want to keep the code around to help with extended features
in the future.

Why not keep an array of date parsers, and if you like, issue warnings
for the deprecated ones?

A wild thought: It might be interesting to have a general translation
utility available for "fixing" such issues when cloning repositories,
although how you would preserve the name I'm not sure.  (The changed
patches would have to have new names, I'm pretty sure, but you'd want
to be able to refer to the originals and keep them around, I think.)
Another application would be if you have a repo created by

1. copy Darcs workspace from a repo to "just a tree"
2. import that tree to another version control system (aVCS)
3. tailor aVCS to new Darcs repo

you could update the dependencies to be those of the originally copied
repo.  This is a use case just described on the list (or maybe that
was Darcs Users).

_______________________________________________
darcs-devel mailing list
darcs-devel@darcs.net
http://lists.osuosl.org/mailman/listinfo/darcs-devel

Reply via email to