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

Re: svn_client_upgrade and speed

From: Greg Stein <gstein_at_gmail.com>
Date: Sat, 12 Sep 2009 03:27:39 -0400

On Sat, Sep 12, 2009 at 02:30, Listman <listman_at_burble.net> wrote:
>...
> Everyone here appreciates the work you're doing, thats a given. The issue is
> that we're trying to plan life/work/dinner etc and the 1.7 schedule is just
> getting pushed out on a never-ending basis.

Well aware of that. We've already cut a number of plans from 1.7.
There is a lot more that I'd like to do with libsvn_wc, but won't get
done for 1.7.

We're trying to get this out in a reasonable timeframe.

> We talked about this in February of this year.
>
> http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=1183455
>
> In this discussion we'd asked the dev-team to get some performance
> improvements into 1.6 rather than pushing them out to 1.7. You told us to
> wait until 1.7 and life will be peachy

And I still maintain that. The real client improvements will come from
reducing I/O, and *that* is being done by moving to a single .svn
subdir per wcroot. 1.6 did not have a good platform for further
improvements, and it still doesn't.

> and then you said this:
>
> http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2373285

Yup. I said "until at least end of October", and based on markphip's
burndown charts, it is looking much more like December (at least).
He's going to repost a chart early next week. We'll see how things are
going.

Bert has jumped into the fray and is doing a bang-up job. We have
additional patches coming in from a couple new contributors; we're
slower to apply those, but they are quite helpful.

> and now todays discussion..
>
> Is the 1.7 release going to be even worse than the nightmare that was 1.5?

Don't be insulting.

We're working hard. It is taking more time than expected. For myself,
my wedding and a move from EU back to the States has distracted me for
several months. So sue me that I haven't had as much time as I may
have thought.

> My prediction from February '09 holds, if we get a *useable* 1.7 release by
> February 2010 I'll be (pleasantly) surprised

And I said it back then: I have money to put on that. Do you? I will
*still* bet on that. Put up.

> Again, I'm not trying to beat anyone up here, this is an open source project
> and we appreciate the work. But we're all big boys/girls here and if you
> make *very* strong statements as you did back in Feb and then bail someone
> is going to mention it.

Are you saying that I'm bailing?

If I'm off my a few months, then I'll accept that. But don't even
start to assume that I'm "bailing".

None of us want a lifecycle like 1.5, and we won't let that happen. If
you think that will happen, then place a bet against me. Go ahead.

And I don't recall a single patch from you to help speed up this
development. If you start to dig into libsvn_wc and produce some work,
then I will believe that you understand what is happening here. That
you understand the difficulties we are working with. The scope of the
problem. I've seen others pony up, and be brave.

AND the variability and unknowns that we have run into since February,
both in terms of code and in life. Shit happens.

Yes, I'm cranky. You're insulting my integrity. I can take blame for
bad estimations, and be off by a few months. Fine. But don't push it
further than that.

-g

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2393808
Received on 2009-09-12 09:28:05 CEST

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