Re: Since there's a resurgence of interest in cvs2svn...

On Tue, Oct 22, 2002 at 03:45:20PM -0500, Eric Gillespie wrote:
> Marko Macek <Marko.Macek@gmx.net> writes:
>
> > It handles two things:
> > - different log message/author (id)
> > - the same file must not be twice in the same commit
> >
> > This should also fix the problem reported by Shun-ichi Goto.
>
> Woohoo! I applied your patch and the resulting repository
> appears correct. 408 revisions vs. 289 :). I'm inspecting svn
> log -v / now to see if i see anything awry, but so far it looks
> good.

That patch will break interleaved commits. I just updated cvs2svn.py...
please give that one a shot.

A while back, I "fixed" a problem where a Commit could span a *lot* longer
than the COMMIT_THRESHOLD. We needed to flush *before* adding items; prior
to my fix, we'd add an item and it would "extend" the duration of the Commit
bucket, *then* check for flushing, but the bucket had been extended to "now"
and wouldn't get flushed. I had a Commit that spanned a day or so :-)

Anyways, when that happened, the scanning loop moved "up" in the function,
and it overwrote the ID used to key into the Commit buckets. I changed the
variable, so it should be hunky dory now...

Try it out :-)

(please... I didn't test my change :-)

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org