Graft should signal a conflict but silently drops one side:

[Arnau Castello]

I have come across a situation where I'm trying to graft  a revision which updates one .hgsubstate on top of a another branch, where all the substates are different from either the graft or the parent of the graft.

Given that both sides of the 3-way merge differ from the base(which I understand for grafts is the parent of the grafted revision) on one line, I'd expect mercurial to prompt me to resolve the conflict, but instead it silently resolves it by replacing .hgsubstate completely with the version from the graft, even where it doesn't differ from its parent.

I'd like to understand why it is doing this merge, and why it's not generating a conflict(as I think it should, since both sides of the merge have changed). I tried running the graft with the debug flag, but there is a lot of information there and I don't know what I'm looking for.

These are the offending .hgsubstate files(just changed the subrepo path but otherwise they're the same). If there is any other bit of useful info I can add it, I haven't added -debug output as it's a lot of spam.

Changeset to graft:
7fe8fcdd7648e14bee5889a4b3155fde49f09de4 external/a
5b90cf021b0006c8681247a73e9940f835a959f4 external/b
69e97bdab56155fee6ab4d0d21bbf36b34b040f8 external/c

Parent of changeset to graft:
7fe8fcdd7648e14bee5889a4b3155fde49f09de4 external/a
5b90cf021b0006c8681247a73e9940f835a959f4 external/b
8479ff0a18dc684db7f0771ace700915c51e92e6 external/c

Changeset to graft on top of:
f60c22f43d335e95a95301aee58a092b63800e4b external/a
26e615cf033cffa9fba77d2369e3802cc2b9a95e external/b
ca46ca7e5243439de09a2d14ffa60432c6c56d74 external/c

Result of graft:
7fe8fcdd7648e14bee5889a4b3155fde49f09de4 external/a
5b90cf021b0006c8681247a73e9940f835a959f4 external/b
69e97bdab56155fee6ab4d0d21bbf36b34b040f8 external/c

Thanks in advance!

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mercurial-scm.org/pipermail/mercurial/attachments/20140609/07c5a350/attachment-0002.html>