Re: DAV is complicated and slow? (was Re: [svndiff1] Accept-Encoding in mod_dav_svn)

On 21 Nov 2005 17:59:23 -0500, C. Michael Pilato <cmpilato@collab.net> wrote:

> The hard-to-maintain code may in fact be in the custom reports, but is
> only hard-to-maintain because it tries to remain DAVvish. There's
> nothing inherantly hard to maintain about an XML report and XML
> response.

I don't think there's even anything particularly wrong with the
formats of the reports, DAVish or not, the main problem I've seen with
ra_dav is just that there code that's there is trying to do many
things at once. For example, compare the update-report in
mod_dav_svn/update.c to the corresponding code for update, switch,
diff, status, etc in svnserve, it's probably about the same amount of
raw LOC, but because the ra_svn code does one thing at a time you can
actually understand what it's doing.

The fact that it's written in terms of SVN concepts as opposed to DAV
concepts doesn't hurt, mind you, nor does the fact that the svnserve
side of things is actually documented...

> That said, if we did choose to use something else that wasn't XML,
> we'd be freed from some of the naming constraints we've had to deal
> with there (propnames with colons, junk like that). I'd be fine, for
> example, with tunneling the SVN protocol over HTTP.

It would, if nothing else, let us get rid of the base64 encoding of
svndiff data, that's gotta be a pretty big win in and of itself for
space reasons.

-garrett

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Nov 22 00:45:47 2005