Re: update_tests.py 60 FAIL over ra_serf

Philip Martin <philip.martin_at_wandisco.com> writes:

> Philip Martin <philip.martin_at_wandisco.com> writes:
>
>> update_tests.py 60 attempts to update to HEAD+1 and expects it to fail.
>> This tests is failing over ra_serf, the update hangs in some way. This
>> started with r1412651, which is just a wc change.

>> #6 0x00007f6a51728f19 in svn_ra_get_inherited_props
>> (session=0x7f6a5208b0f0,
>> iprops=0x7fff0eb5b988, path=0x7f6a51c76d65 "", revision=2,
>> result_pool=0x7f6a52060028, scratch_pool=0x7f6a52060028)
>> at ../src/subversion/libsvn_ra/ra_loader.c:1315
>> #7 0x00007f6a51c3dfcf in svn_client__get_inheritable_props (
>
> The r1412650 client doesn't set needs_cached_iprops in
> svn_client__get_inheritable_props and so doesn't call
> svn_ra_get_inherited_props.
>
> The r1412651 client sets needs_cached_iprops, makes the call, and hangs.

So this is a deliberate change:

* subversion/libsvn_client/iprops.c
  (need_to_cache_iprops): Use svn_wc_check_root() to work properly on files
    and just return TRUE for the repository root as fetching from the wc is
    just as expensive as testing for the wcroot.

I don't understand the iprops code yet. This change is causing an extra
RA call to occur and I don't know whether this extra call is required
for correct behaviour and so ommitting it is a bug, or whether omitting
it is merely an optimisation.

-- 
Certified & Supported Apache Subversion Downloads:
http://www.wandisco.com/subversion/download