Re: git rebase regression: cannot pass a shell expression directly to --exec

From: Jonathan Nieder <jrnieder@gmail.com>
Date: May 16, 2017, 17:37 UTC
Message-ID: <20170516173707.GP27400@aiede.svl.corp.google.com>
In-Reply-To: <20170516164750.3tw6xlbcbyuu5t72@sigill.intra.peff.net>

Jeff King wrote:

>>> On Mon, May 15, 2017 at 11:25:03PM -0400, Jeff King wrote:

>>>> One hack would be to look for BASH_FUNC_* in the environment and disable
>>>> the optimization in that case. I think that would make your case Just
>>>> Work. It doesn't help other oddball cases, like:

[...]

Show 5 quoted lines

>> Hm.  execvp explicitly does this when it hits ENOEXEC, but not for
>> ENOENT.  Do you know why that is?
>
> When execvp(foo) falls back on ENOEXEC, it is not running "sh -c foo".
> It is actually running "sh foo" to run the script contents.

Oh --- now I get it. This is about the optimization to bypass the shell in prepare_shell_command. It has nothing to do with execvp --- our execvp emulation already does the right thing, and Brandon's patches did not make this problem any worse or better.

Unconditionally falling back to sh -c when we get ENOENT in the RUN_USING_SHELL case makes sense to me.

Sorry for the confusion, Jonathan