Stefan Sperling wrote:
> On Tue, Jun 23, 2009 at 06:34:04PM +0200, Stefan Küng wrote:
>> Stefan Sperling wrote:
>>> On Tue, Jun 23, 2009 at 06:17:18PM +0200, Stefan Küng wrote:
>>>> Yesterday I started to adjust TSVN to link against the Subversion trunk
>>>> so I can start implementing the new features and the new APIs.
>>>> But I now have to revert back, because the new wc format has serious
>>>> (and I mean serious) performance problems. I doubt that any user would
>>>> ever accept that.
>>>> Am I the only one seeing this?
>>> No you're not.
>>> This is to be expected right now. Things are still in development.
>> Thanks for the info.
>> I'll have to revert my changes for now and go back to linking against
>> svn 1.6.x - trunk is just not usable for me (and the nightly build
>> testers) right now.
> I use trunk every day to work on Subversion itself and it's slow
> but bearable. Please reconsider exposing trunk to more testing,
> it certainly can't hurt. Our trunk code receives way too little
> testing as is.
I can't consider an update time of more than 10 minutes bearable. That's
unusable for me, sorry.
Especially since the CPU is at 100% during that time. That means I can't
really use my computer for anything else during an update.
> Maybe it's exceptionally bad on Windows and there's an easy way
> to fix this?
The stacktrace shows that the problem is in sqlite (or how its used).
That's where the time is spent during an update. I haven't checked
exactly where that is (just used process explorer to get an idea).
Received on 2009-06-23 18:51:15 CEST