[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: cvs2svn - file moving

From: Greg Hudson <ghudson_at_MIT.EDU>
Date: 2001-12-12 15:46:31 CET

On Wed, 2001-12-12 at 05:07, Greg Stein wrote:
> * moving a ,v file makes it appear as if the file was *always* at that
> location. thus, cvs2svn couldn't do anything about it.

Agreed.

> * same for copying a ,v to a new location

Well... in some cases, we *can* recognize the move, though we might not
want to bother.

For one, we have a pretty good heuristic test: if file B has exactly the
same revisions as file A (except for the last, dead revision of A) plus
some more, then there's a damn good chance that it was copied at the
divergence point. No absolute guarantee, but it's more likely than
not. We'd have to think about common screw cases where this heuristic
triggers falsely.

And two, when NetBSD moves a file, it copies the ,v file and changes the
state of all of the existing revisions to "Dead", and then creates a new
live revision with the current date (and no change to the file
contents). That should give us enough information. I assume NetBSD
isn't the only project to use this practice, since it makes checkouts by
date and tag work. (I think they also remove all the existing tags, but
that's not necessary since all of the tags would point to revs which are
all dead.)

However, applying either of these tests sounds really tough. Not fodder
for an initial cut of cvs2svn.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:36:52 2006

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.