On Fri, Apr 23, 2010 at 06:48, Philip Martin <philip.mar...@wandisco.com> wrote: > gst...@apache.org writes: > >> Author: gstein >> Date: Fri Apr 23 01:39:05 2010 >> New Revision: 937128 >> >> URL: http://svn.apache.org/viewvc?rev=937128&view=rev >> Log: >> Change some basics around the svn_wc__merge_props before digging in for >> some further work. >> >> Note: this restricts svn_wc_merge_props3() to ONLY "normal" props. The >> prior behavior of allowing wc/entry props would result in installing these >> as real properties (oops!). >> >> * subversion/libsvn_wc/props.h: >> (svn_wc__merge_props): pass the node's KIND information (or the intended >> KIND if it has not (yet) been created) >> >> * subversion/libsvn_wc/props.c: >> (svn_wc_merge_props3): disallow non-normal propchanges. fetch the KIND >> earlier and pass to merge_props. fetch the pristine/actual props and >> pass them along, too. >> (svn_wc__merge_props): accept the KIND rather than fetching it (and then >> compensating for "unknown"). assert that we always get the set of >> BASE/ACTUAL properties rather than attempting to fetch these or fill >> in some empty defaults. remove all the IS_NORMAL stuff. we only ever >> get normal properties. > > This causes > > FAIL: merge_tests.py 54: set no mergeinfo when merging from foreign repos > FAIL: update_tests.py 33: update wc containing a replaced-with-history file > FAIL: update_tests.py 34: update handles obstructing paths scheduled for add > > all in svn_wc__merge_props with > > ../src/subversion/libsvn_subr/io.c:2935: (apr_err=235000) > svn: In file '../src/subversion/libsvn_wc/props.c' line 1548: assertion > failed (base_props != NULL)
Yup, thanks. I just discovered that after my full test run completed. update-34 is not due to the assertion, however, so this isn't as straightforward... Cheers, -g
