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

Re: svn_ra.h

From: Greg Stein <gstein_at_lyra.org>
Date: 2000-11-29 20:38:09 CET

On Wed, Nov 29, 2000 at 02:09:32PM -0500, Greg Hudson wrote:
> > Dynamic linking. Definitely.
>
> Any rationale other than wanting to avoid the XML marshalling?

Performance, resource consumption (e.g. working set), avoiding complexity
(logistically, robustness, maintainability), and not wanting to design Yet
Another Protocol (for the two processes to talk to each other). We've got
DAV for an interprocess protocol; I would hate to introduce another one.

[ Ben said the XML wasn't a protocol; I agree: it is a patch format; there
  is more that will happen across the RA boundary than shoving patches back
  and forth ]

> I feel
> like it's easier in the long run to maintain support for a text format
> than for an ABI.

Depends on the stance that you'd like to take. Personally, I don't care
about ABI compatibility across releases. This is Open Source... people can
recompile their private modules *if* they want to update to the latest
version of SVN.

> > Seems reasonable to do, but I'd be wary about leading people down
> > this path. I doubt that it will be as fast as the DAV model (if it
> > is just as fast or faster, then I've done my job wrong).
>
> Really? I understand that with DAV, you can take advantage of
> existing HTTP proxies, but apart from that, I don't see where the
> bottleneck would be running over ssh.

The additional marshalling of the XML format over SSH.

I'm also going to guess that the Apache server has been tuned for network
performance much better than SSH. The server load will also be much smaller
if Apache is used, rather than a bunch of SSH daemon plus SVN
clients-running-on-the-server.

> (Assuming you have equivalent
> security measures on the HTTP connection(s) you're comparing things
> to, of course.)

Right. SSH and SSL are actually pretty fast once you get the initial key
exchange done. The exchange is at high security, the data transfer uses a
symmetric encryption with a much lower key bit-size.

> I thought the goal for running over DAV was compatibility with HTTP
> infrastructure (caching proxies, mainly) and other HTTP versioning
> applications, not speed.

Yes, all of that, *and* speed.

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/
Received on Sat Oct 21 14:36:15 2006

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.