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

Re: svn commit: r38790 - trunk/subversion/libsvn_wc

From: Greg Stein <gstein_at_gmail.com>
Date: Tue, 18 Aug 2009 10:44:41 +0200

On Tue, Aug 18, 2009 at 10:03, Bert Huijben<rhuijben_at_sharpsvn.net> wrote:
>> -----Original Message-----
>> From: Hyrum K. Wright [mailto:hyrum_at_hyrumwright.org]
>> Sent: maandag 17 augustus 2009 23:37
>> To: svn_at_subversion.tigris.org
>> Subject: svn commit: r38790 - trunk/subversion/libsvn_wc
>>
>> Author: hwright
>> Date: Mon Aug 17 14:37:01 2009
>> New Revision: 38790
>>
>> Log:
>> Don't require an access baton for entries_write().  Although we still
>> retrieve
>> one internally, that's not reason the API has to have the access baton.
>
> Note that the caller should have obtained some kind of working copy lock, before calling this function!

There is still a call to svn_wc__adm_write_check() in there.

> This leaves the door open to just writing to the wc-database without any usefull locking, which will break when you have more than one subversion client at the same time.
>
> A very very VERY (is 3 times enough?) common situation outside our test suite.

I bet multiple writers is not that common.

> Other note: I see about 120 test failures on the Windows buildbot after this commit, while the buildbot was green before. (Testing locally to see if this makes a difference)

Not sure why yet. I don't see anything obvious in the change, but I
need to look at more context...

Cheers,
-g

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2384664
Received on 2009-08-18 10:45:17 CEST

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