Lele Gaifax <[EMAIL PROTECTED]> writes:

>>>>>> "Bruce" == Bruce Stephens <[EMAIL PROTECTED]> writes:

[...]

>     Bruce> So that's what I did.  I was only using the result in trac,
>     Bruce> so I didn't notice that it doesn't do what I expected.
>     Bruce> Instead, it gets all changes since that date, regardless of
>     Bruce> branch, which is logical I guess, although I'm not sure it
>     Bruce> would make a useful repository.
>
> Surely not, but that's definitely *not* what tailor was born for.

I'm converting a few months of our CVS repository to subversion, to
use in trac (which everyone finds handy: we could use cvstrac, I
guess), which I'll continue to keep up to date using tailor.  

I also intend to do the same thing on the resulting subversion
repository (probably into monotone), which I'll actually use for
developing.

I'll be using tailor one-way: I'll use the monotone (or subversion, or
whatever) SCM to produce patches which I'll commit using CVS.  So
that's not ideal, because there's a danger of conflicts and so on.
However, we use scripts around CVS, so doing anything more two-way
would be more work than I want to bother with (that's also a reason I
don't want to bother with cvstrac: I'd have to teach that something
about the way our scripts work so it knows what changes are in a
changeset).

(One day we'll migrate to something better.  Sadly, not darcs, which
seems to me to match our workflow best; darcs just seems too slow for
the size of our working copies.)

I think I'm using tailor much as it's intended to be used?  (Once upon
a time I had thoughts of actually converting our repository, but I
gave that idea up some time ago.)

[...]

_______________________________________________
Tailor mailing list
[email protected]
http://lists.zooko.com/mailman/listinfo/tailor

Reply via email to