On Wed, Dec 7, 2011 at 6:30 AM, Julian Foad <julianf...@btopenworld.com> wrote: > Hi Paul. > >> Author: pburba >> Date: Thu Dec 1 18:38:38 2011 >> New Revision: 1209176 >> >> URL: http://svn.apache.org/viewvc?rev=1209176&view=rev >> Log: >> Fix issue #4056 "don't record non-inheritable mergeinfo if missing >> subtrees are not touched by the full-depth diff". >> >> * subversion/libsvn_client/mergeinfo.h >> >> (svn_client__merge_path_t): Add a few new members. >> >> * subversion/libsvn_client/merge.c >> >> (calculate_merge_inheritance): Stop adjusting a rangelist in place and >> instead simply communicate back to the caller they inheritance type >> required to describe the merge. >> >> (insert_parent_and_sibs_of_sw_absent_del_subtree): Set one of the new >> svn_client__merge_path_t members. >> >> (flag_subtrees_needing_mergeinfo): New. Takes an initial walk over > > > Is there a description missing there?
Yes, I adjusted the log message. > The other parts of the log message detail don't seem to mention a functional > change so I guess it's in this function is it? It is. > - Julian > > >> (record_mergeinfo_for_dir_merge): Factor out the bulk of the code which >> determines if a subtree of interest *needs* new mergeinfo to describe a >> merge -- see flag_subtrees_needing_mergeinfo for its new home. Leave >> behind the code which actually *calculates* what the actual mergeinfo is >> and records it. >> >> (do_directory_merge): Account for new differentiation between missing and >> switched children. >> >> * subversion/tests/cmdline/merge_authz_tests.py >> >> (mergeinfo_and_skipped_paths): Expand test for greater coverage of >> issue #4056 and adjust existing expectations to account for the more >> concise mergeinfo produced by this fix. >> >> * subversion/tests/cmdline/merge_tests.py >> >> (merge_to_path_with_switched_children): Adjust existing expectations to >> account for the more concise mergeinfo produced by this fix. >> >> (noninheritable_mergeinfo_test_set_up): Adjust test expectations a bit. >> We still expect no spurious subtree mergeinfo, but how this is >> accomplished (set it, then elide it) was not what the test >> originally expected. >> >> (unnecessary_noninheritable_mergeinfo_missing_subtrees): Remove XFail >> decorator and comment.
