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

Re: [PROPOSAL] Addition of rsync algorithm for saving bandwidth in 'svn takeover'

From: Ben Collins-Sussman <sussman_at_collab.net>
Date: 2005-07-12 19:50:16 CEST

On Jul 11, 2005, at 4:27 PM, Jonathan Gilbert wrote:
>
> I'm sure there are other things that people will want to discuss.
> One of
> them might be: "Is it worthwhile to do this at all?"

I don't want to discourage you too much here, because this is a great
feature that people have wanted for a long time. I'm glad you're
working on it!

Over and over, I see people get frustrated that after a large import
(or after a cvs2svn run) they're forced to run 'svn checkout', when
the original dataset is already sitting in front of them. And the
main source of annoyance isn't about having to 'mv' things out of the
way, it's about sitting around and waiting for the entire tree to re-
download. So it's definitely a Good Thing that you're trying to
avoid re-downloading.

In almost every case, the tree being "taken over" will be identical
to the one already in ther repository: there will be no files to
download. 'svn import' doesn't munge file data. And history
conversion errors are usually related to branch history being
incorrect, not the main trunk data being incorrect.

Sure, once in a while somebody may have a slightly modified local
tree (perhaps a patch in progress), in which case a *few* files might
need to be freshly downloaded. In the context of the whole project,
no big deal at all.

But I feel like we're disagreeing on the scope of the problem here.

 From where I sit, the problem has always been:

    "Oh no, I don't want to re-download the whole project!"

    Solution: no worries, we'll only download just the few files
that differ.

But you've focused on a problem that I don't think is really a problem:

     "Oh no, I don't want to re-download the whole file!"

    Solution: no worries, rsync will only download small bits that
differ.

The problem is that we're no longer KISS. Adding rsync-like code has
a cost: we have to maintain it, debug it, and support it forever
more. And I don't think the problem it's solving is actually a
problem we need to care about. The benefit is negligible.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Jul 12 19:52:01 2005

This is an archived mail posted to the Subversion Dev mailing list.