[issue287] aborting "hg qrev" leaves tree in inconsistent state

[Jeremy Fitzhardinge]

New submission from Jeremy Fitzhardinge <jeremy at goop.org>:

In a large tree, "hg qrev" is a long operation.  If I interrupt it with ^C, then
it leaves the tree in an inconsistent state.  If I try to qpush/pop or any other
mq operation, I get:
queue top not at dirstate parents. top 3bcd2514bb64 dirstate ec906c339b35
000000000000

In this case, the patch I'm working on is "asm-offset-sizeof_vcpu".  "status"
ends with:
ec906c339b35d444cc85a988699cee955b994353:blkfront
3bcd2514bb64bac8ce62574b2da486be754744cb:asm-offset-sizeof_vcpu
hg id says:
ec906c339b35 tip

So in other words, mq thinks the patch has been applied, but hg itself doesn't.

I can recover by manually editing status and manually reverting the patch and
then reapplying, but I'm not sure if that would work all the time.  And it
certainly wasn't clear how to fix it when I first got into this state; when it
happened before, I ended up just cloning the repo and starting the operation again.

----------
messages: 1508
nosy: jsgf
priority: bug
status: unread
title: aborting "hg qrev" leaves tree in inconsistent state
topic: mq

____________________________________________________
Mercurial issue tracker <mercurial-bugs at selenic.com>
<http://www.selenic.com/mercurial/bts/issue287>
____________________________________________________