Re: Fixing issue 644

Ben Collins-Sussman <sussman@collab.net> writes:

> Philip Martin <philip@codematters.co.uk> writes:
>
> > Aside from the implementation, a design problem with this approach is
> > that it delays the conflict detection until the deltas are sent. I
> > thought subversion was meant to detect conflicts before sending all
> > the deltas, to avoid the case where large amounts of data get sent
> > only for a conflict to be detected near the end. This means that
> > delaying the conflict detection until the PUT is too late, isn't it?
> > In which case we are back to calling PROPFIND for added files.
>
> Hmm, I'm confused. Why aren't we detecting the conflict when the
> client first tries to CHECKOUT the resource into the activity? That
> should happen before the PUT, no?

Well, commit_add_file in libsvn_ra_dav/commit.c does a checkout on the
parent directory, but there is no reason for that to fail. It doesn't
attempt to checkout the file, probably because it is assumed the file
does not exist.

I suppose that rather than a PROPFIND we could use a CHECKOUT to
detect if the file already exists. That looks quite simple to me (but
I might be talking rubbish, I only started looking at HTTP/WebDAV 2
days ago :-)

-- 
Philip
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org