Re: cg-seek messed me up

From: Petr Baudis <pasky@ucw.cz>
Date: May 31, 2005, 18:39 UTC
Message-ID: <20050531183907.GE7013@pasky.ji.cz>
In-Reply-To: <20050531143318.GB27047@tumblerings.org>

Dear diary, on Tue, May 31, 2005 at 04:33:18PM CEST, I got a letter where Zack Brown <zbrown@tumblerings.org> told me that...

> Hi folks,

Hi,

Show 13 quoted lines

> I'm not positive, but I *think* I shot myself in the foot by doing a cg-seek
> when there were uncommitted changes in my working file set. The patching process
> created duplicate areas in the files, while losing other areas and giving
> patch rejects. Before I figured out the problem, I wasn't even able to seek to a
> known state because of this, so I ended up having to recreate the files as best
> as I could from the mess I was left with in the working directory. I realize I
> could have gotten a clean tree by cloning, but this didn't occur to me at the
> time. It seemed like the whole repository was just corrupt.
> 
> I think I recovered everything, and I know it was my own mistake, but it
> seems like this will be a common blunder by users. Maybe cg-seek should
> first do a comparison between the working tree and the most recent known
> state. If the two differ, it should exit with an error.

I've rewritten cg-seek to use the magical tree_timewarp function. So what it will do now is that it will save the local changes, then just overwrite the tree with the seeked-to target and then it will reapply the local changes. This should give much better (although still far from perfect) results and reduce the inconsistencies popping up.

Anyway, anytime you are in trouble, cg-cancel is the way out! ;-)

-- 
				Petr "Pasky" Baudis
Stuff: http://pasky.or.cz/
C++: an octopus made by nailing extra legs onto a dog. -- Steve Taylor