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

RE: [PATCH] Speedup the python tests (now with log message)

From: Lieven Govaerts <lgo_at_mobsol.be>
Date: 2006-06-20 20:44:57 CEST

Philip Martin wrote:
> > Julian Foad <julianfoad@btopenworld.com> writes:
> >
> >>OK, I don't dispute your evidence but I'm interested to know what the
> >>definition of "live" is that applies to this greek tree reference
> >>repository.
> >
> > The usual: a repository is live when any process is accessing it.
> >
> > In this case the problem was that the testsuite imported into, and
> > then immediately copied, the repository. The svnserve process that
> > was forked to handle the import could still be running when the copy
> > was made.
>
I've updated the patch to use svnadmin hotcopy instead of a normal tree
copy. This is now the default for both FSFS and BDB repositories, as long as
the test doesn't require a repos with a new UUID.

Performance is still similar as the treecopy version.

thanks for your input.

Lieven.

[[[
Improve the performance of the python tests:
 1. Create the test repos with svnadmin hotcopy instead of
    svnadmin dump|svnadmin load.
 2. Check out a pristine working copy once, and copy+relocate it to
    the test working copy.

* subversion/tests/cmdline/svntest/actions.py
  (guarantee_greek_repository): checkout a pristine working copy.
  (make_repo_and_wc): instead of checking out the pristine repos,
   make a copy of the pristine working copy and relocate it.

* subversion/tests/cmdline/svntest/main.py
  (pristine_wc_dir): new global
  (copy_repos): when a test doesn't want a repos with a unique UUID,
   do a hotcopy of the pristine repos.
]]]

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Received on Tue Jun 20 20:50:18 2006

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.