On 01/19/2010 06:55 PM, Mark Phippard wrote:
On Tue, Jan 19, 2010 at 7:58 AM, Kamesh Jayachandran<kam...@collab.net> wrote:
Mike has suggestion at [1] to make use 'DAV:apply-to-version' to avoid the
problematic PROPFIND altogether.
This suggestion works very well.
It is just the ra_neon fix i.e no mod_dav_svn fix
How does this work without also changing mod_dav_svn? I thought it
did not already have code to handle this?
<snip from http://www.ietf.org/rfc/rfc3253.txt>
(DAV:create-working-resource-from-checked-in-version): If the
request-URL identified a version-controlled resource, and
DAV:apply-to-version is specified in the request body, the
CHECKOUT is applied to the DAV:checked-in version of the version-
controlled resource, and not the version-controlled resource
itself.
</snip>
I think mod_dav does this for us.
Attached patch implements this fix.
All tests pass with this test(except the known mergeinfo test 8 failure.)
Does this change eliminate the PROPFIND? If it doesn't, given that we
do not have any tests that show this problem it does not seem
unexpected that tests would pass.
It eliminates the PROPFIND. I manually tested this by examining the
concerned neon-debug output.
In next few hours I will commit this as it looks fairly obvious unless there
are some objections.
It sounds great if this fixes the problem. I am still a little
perplexed as to how this works and how it changes the commit process
and the interactions with a proxy.
It just does not do the problematic PROPFIND leaving way to CHECKOUT to
figure it out.
With regards
Kamesh Jayachandran
