Re: Using alternate working directory
- From
Jeff King <peff@peff.net>
- Date
- Sep 24, 2013, 05:00 UTC
- Message-ID
- <20130924050011.GE2766@sigill.intra.peff.net>
- In-Reply-To
- <523E09D8.8090808@rachum.com>
On Sun, Sep 22, 2013 at 12:04:24AM +0300, Ram Rachum wrote:
Show 6 quoted lines
> I'm making a script `gm` which lets me merge one branch into another > without having either checked out. It works for some cases but not > all. I'm trying to make it work for more cases. > > I concluded that the best way to do it would be by using an > alternate, temporary working directory instead of the repo itself.
Yes, otherwise you will be stomping all over the working tree of whatever branch _is_ checked out.
Show 10 quoted lines
> This is my script: > > https://gist.github.com/cool-RR/6575042 > > Now, the problem is that when I try it, it gives these errors: > > git checkout-index: my_file is not in the cache and then error: > my_file: cannot add to the index - missing --add option? > > Anyone has any idea what to do?
Your script is quite similar to the one that is used server-side at GitHub to generate the "this can be merged" button for each pull request. So it should work in principle.
Just a guess, but using a relative path for the temporary index file might be a problem. read-tree will operate from $GIT_DIR as its working directory, for example, but I think the git-merge-one-file script will be at the top-level of $GIT_WORK_TREE. Meaning that all of the sub-commands it runs will see an empty index.
-Peff