[issue1645] non-intuitive internal:merge behaviour / configuration, maybe regression

[Thomas Waldmann]

New submission from Thomas Waldmann <tw-public at gmx.de>:

IIRC, in the past this was enough to just switch off all external merge tools
and just get <<<<< >>>>>> marks in the code:

[ui]
merge = internal:merge

I recently reinstalled my machine and used hg 1.1.2 from ubuntu 9.04.
First I got vimdiff(?) as merge tool, then I added the above line to ~/.hgrc.

It seemed to work, I got the conflict markers.

Then I had to do a merge involving a binary file. It just aborted when
encountering this file and didn't ask me which one I want. It aborted the merge,
it didn't have 2 parents afterwards.

Thanks to the help of pmezard on #mercurial, we found out that one needs to do
the following to get correct internal merge behaviour:
1. REMOVE the merge = internal:merge line (!)
2. REMOVE all merge-tools configuration from /etc/mercurial/hgrc.d/merge-tools.rc

After this, it behaved as it should.

IIRC, this was somehow simpler with my previous hg installations, when just
using that ui line was enough to get sane internal:merge behaviour (including
merge [or rather choosing] of binary files).

----------
messages: 9243
nosy: thomaswaldmann
priority: bug
status: unread
title: non-intuitive internal:merge behaviour / configuration, maybe regression

____________________________________________________
Mercurial issue tracker <mercurial-bugs at selenic.com>
<http://www.selenic.com/mercurial/bts/issue1645>
____________________________________________________