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

Re: WC merge info elision and paths with empty revision ranges

From: Daniel Rall <dlr_at_collab.net>
Date: 2007-06-12 02:01:52 CEST

On Thu, 07 Jun 2007, Paul Burba wrote:

> > -----Original Message-----
> > From: Daniel Rall [mailto:dlr@collab.net]
...
> > > NEW TYPE OF 'FULL' ELISION:
> > >
> > > If the merge info on PATH_CHILD consists *only* of paths
> > > that map to
> > > empty revision ranges, and *none* of these paths exist in
> > > PATH_PARENT's merge info, then PATH_CHILD's merge info
> > > elides to PATH_PARENT.
> >
> > I agree.
> >
> > Furthermore, if PATH_PARENT has merge info which
> > contains a path with an empty revision range, CHILD_PATH
> > should still elide to PATH_PARENT. (You probably aren't
> > considering this to be new type of full elision.)
>
> Agreed, well if what you are saying is the following:
>
> If the merge info on PATH_CHILD is equivalent to the merge info on
> PATH_PARENT, *except* for paths which exist *only* in PATH_PARENT and
> map to empty rev ranges, then PATH_CHILD's merge info elides fully.
>
> And I'm pretty sure that is what you are saying!
>
> Though to be honest I can't quite see how we would ever end up with such
> a situation...regardless I added a test of this (and every other case
> mentioned here) in r25318.

Users will screw up the merge info in their WC. Best to be
accomodating, when reasonably possible.

...
> > > The solution to this problem seems straightforward: check the
> > > repository for the nearest ancestor with merge info if one
> > > in the WC
> > > cannot be found. svn_ra_get_mergeinfo() would need to be
> > > changed to
> > > indicate if the merge info it obtained was set explicitly
> > > on a path or
> > > was inherited, analogous to what merge.c:get_wc_mergeinfo() does.
>
> What I ended up doing instead was adding a boolean arg to
> svn_ra_get_mergeinfo() allowing us to specifically request inherited
> mergeinfo only. This actually seemed a lot simpler. Simpler to use
> anyhow, never having added a arg to svn_ra_* function I had no idea the
> horrific amount of trickle down it causes :-\

Oh yes! I commented on the revised patch later in this email thread.

  • application/pgp-signature attachment: stored
Received on Tue Jun 12 02:02:01 2007

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.