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

Re: [PATCH] Optimize away HEAD requests during 'svn import' and certain commits in ra_serf

From: C. Michael Pilato <cmpilato_at_collab.net>
Date: 2007-06-13 15:40:30 CEST

Justin Erenkrantz wrote:
> On 6/12/07, C. Michael Pilato <cmpilato@collab.net> wrote:
>> Justin, here's the patch I think you were suggesting in IRC today. It
>> consistently shaves about 15% off the cost of a remote 'svn import' of
>> the
>> ViewVC tree in my testing. I've not run it through the test suite yet,
>> though (would do so before committing, but need to go AFK for now).
>
> Yup - that looks like what I had in mind! Cool... -- justin

Unfortunately, I think there are a pair of problems with it.

In the open_directory() case, the directory's base_revision value comes from
the editor driver. That driver is permitted to specify SVN_INVALID_REVNUM,
which would then make open_directory(somedir, SVN_INVALID_REVNUM)
indistinguishable from add_directory(somedir). No problem, really -- I'll
just toss an 'added' boolean into the dir_context_t and be done with it.

But secondly, we still need to do the HEAD check when the parent directory
was added *with history*, because that doesn't have the "new-dir-is-empty"
guarantees.

-- 
C. Michael Pilato <cmpilato@collab.net>
CollabNet   <>   www.collab.net   <>   Distributed Development On Demand

Received on Wed Jun 13 15:40:56 2007

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.