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

Re: Performance bug? Very large rev files

From: Rick Yorgason <rick_at_ldagames.com>
Date: 2007-03-30 22:11:55 CEST

Daniel Berlin wrote:
> Nope.
> That is why the server and client tell each other they support
> svndiff1 before we use it.

Okay, I think I misunderstood you before. Turns out you're saying
exactly what I expected in the first place >:)

By the way, when I say that you can convert a 1000k window into 10 100k
windows, I'm talking about *de*coded data, since I assume that, under
the current system, a single window will always decode into 100kB
(except at the end of the stream, where it will be smaller). So to
convert svndiff2 to svndiff1:

- Decode first 1000k window (source and target).
- Re-encode that data in ten 100k windows.
- Send to client.
- Decode next 1000k window (source and target)...

The other way around is obviously:

- Decode first ten 100k windows (source and target).
- Re-encode that data in a 1000k window.
- Save on server.
- Decode next ten 100k windows (source and target)...

It would also only be *slightly* more allow non-100k multiples (looking
at the code the actual number is 102400 bytes, which is a little
awkward, as you could neither specify 1MB nor 1MiB).

Like I said, I think we're on the same page, and I just misunderstood.


To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Mar 30 22:11:44 2007

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.