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

svn_wc_conflict_result_choose_repos found dangerous?

From: Daniel Rall <dlr_at_collab.net>
Date: 2007-07-18 01:55:52 CEST

WRT issue #2830, "Improve handling for conflicts encountered during a
merge": http://subversion.tigris.org/issues/show_bug.cgi?id=2830

On IRC yesterday, we were discussing some of the new "automated
conflict resolution" result codes (from svn_wc_conflict_result_t)
available to the excellent new conflict resolution callback
(svn_wc_conflict_resolver_func_t) when the conflicted path is a file,
and its merge sources share a common ancestry.

Specifically, we were talking about the potential dangers of the
svn_wc_conflict_result_choose_repos result code, which causes the
repository version of the file to be copied over the top of the user's
working version, ***without preserving any of the user's unconflicting
local mods***. This last part is going to be a gotcha for users, who
are going to expect only conflicting portions of the merge to be
discarded in favor of the repository's version, and are going to be
pissed when their local cahnges are discarded. Other VC systems I've
looked at work by operating on a chunk-by-chunk basis, rather than on
a whole file.

On further reflection I think our concerns also apply somewhat to the
other result codes (e.g. choosing svn_wc_conflict_result_choose_user
will discard unconflicting hunks frmo the merge).

Thoughts?

  • application/pgp-signature attachment: stored
Received on Wed Jul 18 01:55:00 2007

This is an archived mail posted to the Subversion Dev mailing list.