From: Max Kirillov <max@max630.net>
Date: Sat, 16 Jun 2018 08:22:43 GMT
Subject: Re: [PATCH] unpack-trees: do not fail reset because of unmerged skipped entry
Message-ID: <CAF7_NFR8dKZtbRr+p_HGECdZhyqSWAsMuZ2JthXefGyFL+57zA@mail.gmail.com>
In-Reply-To: <xmqqh8m3zurz.fsf@gitster-ct.c.googlers.com>

> I do not know offhand if "reset --merge" should force succeeding in
such a case, but I agree that it is criminal to stop "reset --hard"
with "not uptodate", as the whole point of "hard reset" is to get
rid of the 'not up-to-date' modification.

I originally had a fix just for "reset --hard". It was in
verify_uptodate_1(), to move check for "->reset" earlier. But then I
found that "merge --abort" does not use "reset --hard", but rather
--merge, so I fixed that. Because --merge should work also, shouldn't
it?

Actually, I think that fix in verify_uptodate_1() was right, I just
did not find what it affects, after the other fix