On Tue, Feb 12, 2002 at 08:03:24PM -0500, Kevin Pilch-Bisson wrote:
> On Tue, Feb 12, 2002 at 04:24:43PM -0800, Greg Stein wrote:
> > We have it already. The server only sends svndiff data when the client sends
> > a specific header. In other words, the client said "here is what I have, if
> > you can give me a diff against that, then great. otherwise, just send me a
> > fulltext of X." On the other side, the server sees the header and can choose
> > to send a diff, or it can send the fulltext. If the server doesn't know
> > about the headet, then it will just be sending the fulltext because it
> > doesn't know what to diff against.
> > So the fact that the person's WC got garbled is a bug somewhere. Not a lack
> > of feature negotiation.
> We do it, but only one way. The server looks at the rev, and can generate the
> svndiff, but there is no way for the client to say "I don't know what svndiff
> is, send me the fulltext".
There is no way for the client to send rev info, and *not* understand the
results. If the client didn't understand the svndiff result, then it should
not have sent the header.
> Or do we? and the client just always said that it understood svndiff's when
> before it did?
By virtue of the client sending that header, it is explicitly stating that
it understands an svndiff return format.
We've never sent any other header that says "give me svndiff" or that it
could understand it.
There was a short time period over a weekend, or somesuch, where the client
sent "I understand svndiff" but could not handle the result (I was still
working on the code). svn.collab.net wasn't sending svndiff, so there wasn't
a problem. If you happen to fetch that revision (or two or three; whatever)
and ask for an update, then you could corrupt your WC.
But that is a bug. My email was to point out: we *do* negotiate the svndiff
Greg Stein, http://www.lyra.org/
To unsubscribe, e-mail: firstname.lastname@example.org
For additional commands, e-mail: email@example.com
Received on Sat Oct 21 14:37:07 2006