Re: Submodules and rewind

From: Stephen Hemminger <shemminger@vyatta.com>
Date: Feb 20, 2008, 00:52 UTC
Message-ID: <20080219165243.7de6cbf5@extreme>
In-Reply-To: <alpine.LFD.1.00.0802191635010.7833@woody.linux-foundation.org>

On Tue, 19 Feb 2008 16:47:15 -0800 (PST) Linus Torvalds <torvalds@linux-foundation.org> wrote:

Show 9 quoted lines

> 
> 
> On Tue, 19 Feb 2008, Stephen Hemminger wrote:
> > 
> > Don't be stupid, I am not trying be obstreperous, just fix the problem.
> 
> Umm. Stephen - this is the first time you even *describe* the problem.
> 
> Your previous emails were just "hey, submodules don't work, fix it".

No I abused a submodule and broke it, it isn't git's fault directly.

> Why do you then call Dscho stupid for pointing out that you never even 
> bothered to reveal any details of what your problem was?

The cause was a accidental push of the wrong tree up into a public repository. So I manually went back and moved the original tree aside and replaced it with a new tree that was at the point before the errant push.

Show 14 quoted lines

> 
> But it's not clear how you even got into that state to begin with. How did 
> your index get that confused? You said "I had to rewind one project back 
> to a known good state", but considering the output, it looks like you 
> didn't actually rewind it, but left it in some half-way state. 
> 
> A "git reset" should have reset the index, or you could probably have done 
> something like "git add <submodule>" to basically force the index entry 
> for just that submodule to the current state it had.
> 
> At a guess, it *looks* like you reset the submodules themselves, but never 
> reset the superproject.
> 
> 		Linus

The submodule is okay shape but the superproject isn't.