Philip Martin <philip@codematters.co.uk> writes:
> Ben Collins-Sussman <sussman@collab.net> writes:
>
> > We have three different code paths for 'svn diff':
> >
> > * working compared to text-base (no network)
> >
> > * working compared to URL
> >
> > * URL compared to URL
> >
> >
> > The latter two cases are fancy diff editors -- one lives in libsvn_wc,
> > the other in libsvn_client, I believe. They both involve calling
> > RA->get_file() into a temporary file, which is then diffed against
> > either your working file, or another temp file.
>
> The "working compared to URL" diff (diff -rN) works like update, it
> describes the working copy to the server and gets a REPORT(?) to
> convert it to the given revision. I don't think it calls
> RA->get_file directly, I think the diffs are in the REPORT.
>
> The "URL to URL" diff (diff -rN:M) does call RA->get_file, twice for
> each file.
Ahhhh, right, I forgot. Then this really is odd; if the 2nd diff
use-case is sending a standard update REPORT to the server, why isn't
it using cached auth data? This might be pretty simple to fix.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Oct 18 16:38:12 2002