From: Carl Fürstenberg <azatoth@gmail.com>
Date: Fri, 18 Mar 2011 14:13:24 GMT
Subject: Re: gut submodule update and merge
Message-ID: <AANLkTinK_ftV7rCHM2Q0PNRZQXLwCRzfoqogb=dkzXZM@mail.gmail.com>
In-Reply-To: <AANLkTimneYSkr11yDt1eL5d-r6pJrt2EMLGEOBfOxOoD@mail.gmail.com>

2011/3/17 Carl Fürstenberg <azatoth@gmail.com>:
> Hello,
>
> I've tried to setup in a n makefile a way to easy initialize and
> update submodules and at the same time keep eventual changes in the
> submodules. I though an "git submodule update --init --recursive
> --merge" would work,
> but  this i notices created strange result if the submodules hadn't
> been checked out in the first place. the result then was that it
> marked in index all files to be deleted.
>
> This behavior wasn't what I had in mind, and it resulted in I did
> following "fix" to resolve the issue (ugly):
> https://github.com/azatoth/jquery/commit/9977732a26820dd7f3ced2ce4554edadd62ae52c
>
> Is this the intended behavior? logically I think an clean "git
> submodule update --init --merge" should not try to merge if there
> wasn't anything to begin with.
>
> /Carl
>

If my question wasn't clear I'm sorry, All I'm wondering is if using
--merge when cloning an new submodule should behave this way or not. I
don't think most peoples intentions when using merge on an
non-existent repo should be to delete all file :)

/Carl