[Python-Dev] pymigr: Ask for hgeol-checking hook.

Tim Delaney timothy.c.delaney at gmail.com
Sat Feb 26 21:43:36 CET 2011


On 27 February 2011 05:23, "Martin v. Löwis" <martin at v.loewis.de> wrote:

> It actually happened to me, so please trust me that it's not a legend.
> Yes, I could fix it with hg commands, and a lot of text editing.
> It took me a day, I considered the repository corrupted so that I
> actually had to branch from the last ok revision, and redo all checkins
> since (I also discarded changes which I didn't chose to redo). It was
> a real catastrophe to me.
>
> Since the changes actually changed all lines, "hg blame" became useless,
> which was unacceptable.
>

I'd disagree that that is catastrophic repository corruption - it's fixable
by creating a new clone from before the "corruption", discarding the old one
and redoing the changes (possibly via cherry-picking).

Catastrophic corruption of a mercurial repository happens when the history
itself becomes corrupted. This should never happen under normal usage, but
it is possible to happen if you commit using an older version of hg to a
repo that's been created (or modified) by a newer version.

You can pull from a newer version repo using the older version, but you
shouldn't ever commit to it (including pushing) except through the "remote"
interfaces (where the remote hg is the one doing the actual commits).

Tim Delaney
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-dev/attachments/20110227/9d1ab932/attachment.html>


More information about the Python-Dev mailing list