You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Nathan Ward <nw...@esphion.com> on 2002/07/29 06:03:18 UTC
Speed
Hi, I'm using svn on an intel p3 system, and its going sloooow.
tcpdump tells me its not a DNS issue, and there's no other packets going around.
I am connecting to an apache2 CVS server on localhost..
I havn't tried file based access as yet, its next on my list...
Anyone else seen this?
Nathan
Re: Speed
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Karl Fogel <kf...@newton.ch.collab.net> writes:
> I hope this isn't bottlenecked *too* much on our time. I'm not
> expecting that we'll just sit by and do nothing :-), of course, but
> you and Philip Martin seemed to have these changes well in hand.
> What's the current state of the /issue-749-dev branch?
>
> The wc access batons rewrite, and the entries caching in particular,
> are both sufficiently large wins that we should have them in before
> 1.0...
I forgot to say: _incrementally_ please :-). Let's first get the
baton stuff done, then start adding entry caching slowly, operation by
operation.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Beta timeframe
Posted by Justin Erenkrantz <je...@apache.org>.
On Mon, Jul 29, 2002 at 11:14:41AM -0500, Karl Fogel wrote:
> Yeah, I know what you mean. Even the smallest change cascades through
> so many layers...
>
> Well, if you can give us an idea of your schedule, we'll be better
> able to plan. We *would* like to get it in 1.0, though exactly how
> much coding time we can shift to it is still an open question.
Since I can't reproduce the Darwin SVN hang anymore (I think due to
the inclusion of -O2 in my current builds), I'm now trying to work on
my libtool replacement so that I can get builds of SVN to take
something less than a hour on Darwin. Once I can get the builds to
become faster, then I'll play with svn-config trying to get the right
logic in-place to handle in-tree dependencies. Then, I'll start
working on the libsvn_wc changes.
I'm slowly making progress on my libtool clone, so I'm hoping by the
end of this week, I'm starting to tackle the libsvn_wc problem. I'd
hazard a guess that it would take about another week to get the
libsvn_wc changes into the branch. Then, it needs to be merged back
into the trunk and vetted.
All of this presupposes that I don't get nailed by something over
in httpd-land. Hopefully, that gives a better picture of my todo
list... -- justin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Beta timeframe
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Justin Erenkrantz <je...@apache.org> writes:
> What I meant is that you guys have a time frame for releasing beta.
> My schedule may or may not coincide with that. If I can have it in
> by the time you guys want a beta, cool. If not, oh well.
>
> As we've mentioned before, I know what needs to happen, but this
> change is massive. I've tried several times to sit down and take
> a stab at it and just converting one function to use the baton is
> really painful. -- justin
Yeah, I know what you mean. Even the smallest change cascades through
so many layers...
Well, if you can give us an idea of your schedule, we'll be better
able to plan. We *would* like to get it in 1.0, though exactly how
much coding time we can shift to it is still an open question.
Thanks,
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Beta timeframe
Posted by Justin Erenkrantz <je...@apache.org>.
On Mon, Jul 29, 2002 at 09:35:49AM -0500, Karl Fogel wrote:
> I hope this isn't bottlenecked *too* much on our time. I'm not
> expecting that we'll just sit by and do nothing :-), of course, but
> you and Philip Martin seemed to have these changes well in hand.
> What's the current state of the /issue-749-dev branch?
What I meant is that you guys have a time frame for releasing beta.
My schedule may or may not coincide with that. If I can have it in
by the time you guys want a beta, cool. If not, oh well.
As we've mentioned before, I know what needs to happen, but this
change is massive. I've tried several times to sit down and take
a stab at it and just converting one function to use the baton is
really painful. -- justin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Speed
Posted by Philip Martin <ph...@codematters.co.uk>.
Vladimir Prus <gh...@cs.msu.su> writes:
> svn ci -m ""
>
> #Sending
> #Sending a.cpp
> #Transmitting file data .
> #subversion/libsvn_client/commit.c:670
> #svn_error: #21002 : <A general problem occured>
> # Commit succeeded, but other errors follow:
>
> #subversion/libsvn_client/commit.c:688
> #svn_error: #21045 : <Working copy not locked>
> # Error bumping revisions post-commit (details follow):
>
> #subversion/libsvn_wc/lock.c:280
> #svn_error: #21045 : <Working copy not locked>
> # directory not locked (/tmp/wc/)
I'm investigating. It appears that
$ svn ci wc
works, but
$ cd wc
$ svn ci
fails.
--
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Speed
Posted by Vladimir Prus <gh...@cs.msu.su>.
Karl Fogel wrote:
> Philip Martin <ph...@codematters.co.uk> writes:
> > Rev 2625 fixed the problems where the client tried to lock a directory
> > twice. That was the know problem before the baton stuff was reverted,
> > so I think now it works as well as the trunk does.
> >
> > Since alpha has happened, I'd like to merge back to the trunk.
>
> If you think it's safe to do so, +1. There's no need to keep this
> development on a branch now that the release has been made.
I've noticed that the branch was merged to trunk and tried an example which
caused attempt to lock a directory twice before. Unfortunately, it does not
work, althought with a different error messages. The exact command sequence
is below with svn output in comments.
#!/bin/bash
rm -rf /tmp/repo
rm -rf /tmp/wc
cd /tmp
svnadmin create repo
svn co file://localhost/tmp/repo wc
cd wc
touch a.cpp
svn add a.cpp
svn ci -m ""
echo "x" > a.cpp
svn propset foo bar
svn ci -m ""
#subversion/libsvn_client/commit.c:662
#svn_error: #21052 : <Working copy is not up-to-date>
# Commit failed (details follow):
#subversion/libsvn_client/commit.c:874
#svn_error: #21052 : <Working copy is not up-to-date>
# Cannot commit propchanges for directory '/tmp/wc/'
svn upSending
#At revision 1.
svn ci -m ""
#Sending
#Sending a.cpp
#Transmitting file data .
#subversion/libsvn_client/commit.c:670
#svn_error: #21002 : <A general problem occured>
# Commit succeeded, but other errors follow:
#subversion/libsvn_client/commit.c:688
#svn_error: #21045 : <Working copy not locked>
# Error bumping revisions post-commit (details follow):
#subversion/libsvn_wc/lock.c:280
#svn_error: #21045 : <Working copy not locked>
# directory not locked (/tmp/wc/)
- Volodya
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Speed
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Philip Martin <ph...@codematters.co.uk> writes:
> Rev 2625 fixed the problems where the client tried to lock a directory
> twice. That was the know problem before the baton stuff was reverted,
> so I think now it works as well as the trunk does.
>
> Since alpha has happened, I'd like to merge back to the trunk.
If you think it's safe to do so, +1. There's no need to keep this
development on a branch now that the release has been made.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Speed
Posted by Philip Martin <ph...@codematters.co.uk>.
Karl Fogel <kf...@newton.ch.collab.net> writes:
> What's the current state of the /issue-749-dev branch?
Rev 2625 fixed the problems where the client tried to lock a directory
twice. That was the know problem before the baton stuff was reverted,
so I think now it works as well as the trunk does.
Since alpha has happened, I'd like to merge back to the trunk.
--
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Speed
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Justin Erenkrantz <je...@apache.org> writes:
> However, it is on my to-do list, but it hasn't bubbled to the top
> of it yet. Depending upon the CollabNet team's time, I may have
> it done before beta, but maybe not. If not, then it is a post-1.0
> item. -- justin
I hope this isn't bottlenecked *too* much on our time. I'm not
expecting that we'll just sit by and do nothing :-), of course, but
you and Philip Martin seemed to have these changes well in hand.
What's the current state of the /issue-749-dev branch?
The wc access batons rewrite, and the entries caching in particular,
are both sufficiently large wins that we should have them in before
1.0...
-K
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Speed
Posted by Justin Erenkrantz <je...@apache.org>.
On Mon, Jul 29, 2002 at 06:03:18PM +1200, Nathan Ward wrote:
> Hi, I'm using svn on an intel p3 system, and its going sloooow.
> tcpdump tells me its not a DNS issue, and there's no other packets going around.
> I am connecting to an apache2 CVS server on localhost..
> I havn't tried file based access as yet, its next on my list...
What exactly are you doing? You need to qualify your complaints.
There are currently some patches flying around to make the entries
file cached - this significantly reduces the time for certain
operations (diff, checkout, update, etc). But, it requires almost
all libsvn_wc funcs to take a baton which is not trivial. The
patch only adds the baton to the bare minimum functions.
However, it is on my to-do list, but it hasn't bubbled to the top
of it yet. Depending upon the CollabNet team's time, I may have
it done before beta, but maybe not. If not, then it is a post-1.0
item. -- justin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org