Re: Is there something fishy with branches created in CVS?

Skip Montanaro wrote:
> I converted a project from CVS to Subversion in early May.  For various
> reasons I only got a workable TkCVS/TkSVN program going last week.  When
> browsing the branch tree for a file I only see branches which were
> created from trunk after the conversion.  I don't see branches which had
> been created when he code was still in CVS, nor do I see any branches
> descended from them.  These "invisible" branches are fine to use.  We
> have a couple which are active.  We check changs into them and do
> releases from them.  They just aren't visible in TkCVS/TkSVN.  We don't
> have ViewVC installed on our Subversion repos yet so I can't check to
> see if it has similar issues.  In case it matters, I'm using TkCVS
> 8.0.4.  8.1 and 8.2 have known bugs which have yet to be fixed.
> 
> Is this an issue with TkCVS, with cvs2svn or with my usage of it?  At
> this late date (six-ish months after the conversion) is there something
> I can do to "fix" the repository?

I have never heard of such an issue, and I can't think of what on the
cvs2svn side might cause it.  I don't know how TkSVN determines a file's
history.  (This is not quite trivial because Subversion records only
where a file was copied from, not where it was copied to.)  Perhaps
TkSVN needs some kind of database to be primed before use?

If the cause is really something in the Subversion repository, it is
conceivable that you could fix it by doing an "svnadmin dump", fixing up
the dump stream, then "svnadmin load".  Please let us know what you find
out.

Michael

------------------------------------------------------
http://cvs2svn.tigris.org/ds/viewMessage.do?dsForumId=1670&dsMessageId=2418645

To unsubscribe from this discussion, e-mail: [[email protected]].