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

Re: Flushing the entries cache

From: Greg Stein <gstein_at_gmail.com>
Date: Fri, 18 Sep 2009 06:53:23 -0400

Well... third party code could be calling svn_wc_entried_read.

We can flush it at any time. The real problem is the growth of the access
pool.

On Sep 18, 2009 4:25 AM, "Bert Huijben" <B.Huijben_at_competence.biz> wrote:

 21:34 <@gstein> Bert, hwright : I don't think we can remove the entries
cache

I was not woried about keeping the entries cache, which we probably never
can, just about keeping it up to date during processing of commit, update,
merge, etc.

If we remove all svn_wc_entries_read() calls no outside code can have a
reference to the actual entries hash. Then we should be able to allow
flushing it in more places. (And nobody can update the hash by adding or
removing entries behind our back).

The update completion code removed entries from the hash, assuming they
would be written later...

(This specific example was resolved months ago, because the removal was
duplicated in another layer, but there were are probably more cases)

                Bert

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2396355
Received on 2009-09-18 12:53:39 CEST

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