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

On Thu, Oct 24, 2002 at 07:39:23AM +0200, Marko Macek wrote:
> Greg Stein wrote:
> >>Yes. The way we currently lose revisions during conversion is bad (plus
> >>it breaks my tagging/branching code).
> >
> > I don't doubt it :-) In any case, that problem has been fixed. A stupid
> > error on my part a while back.
>
> I am not sure that it has been fixed. The commits on the same file by
> the same person within the COMMITS_THRESHOLD look like they are still
> getting combined.

Ah, yes. I was unclear in my comment. The problem of combined commits
(regardless of author/message) was solved. The multiple file thing was not
solved by my fix.

Your change will prevent multiple file changes from occurring in a single
commit, but it is also over-aggressive in splitting the changes.

Hmm. Thinking more on it, we have two (broken) behaviors here:

1) two changes can be combined in a single commit (current code)
2) two interleaved changes could be split into N changes (your patch)

It seems that option (2) is better than (1), even though both are still
wrong :-) Option (2) means that the commits will actually go through, yet
in some cases we end up with more than we'd like. The first option can just
flat out stop the conversion process.

So let's say that we go ahead and patch in (2) with a comment indicating its
less-than-ideal behavior and the desire for a future fix. That will get us
*better* behavior, if not (yet) ideal.

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