Posix call (execve) breaks mercurial?

[Wayne Werner]

So... this is certainly the deepest I've got to dig into any source code.

I'm experimenting with Review Board for code reviews, and trying to get it
set up/working here at work. When using post-review, however, I started
getting issues with untrusted users - even though they were set to trusted
in my ~/.hgrc and things worked fine otherwise.

So here's where things got weird. I could call
`subprocess.check_output(['hg', 'root'])`, and things worked just fine.
But when I added the env parameter, I got the untrusted issues. So if I
did:

import os, subprocess

# Works just fine
subprocess.check_output(['hg', 'root'])

# Gives untrusted issues
subprocess.check_output(['hg', 'root'], env=os.environ)


Long story short, I dug around the source code and ended up at the POSIX
execve function. I've been reading the manpages, but nothing seems to pop
out at me as "hey, this should/shouldn't work!".

Does anyone know what's going on here, or where I should go for more help?

Thanks,
Wayne

[Wayne Werner]

On Thu, 11 Oct 2012, Wayne Werner wrote:

> So here's where things got weird. I could call
> `subprocess.check_output(['hg', 'root'])`, and things worked just fine. But
> when I added the env parameter, I got the untrusted issues. So if I did:
>
> import os, subprocess
>
> # Works just fine
> subprocess.check_output(['hg', 'root'])
>
> # Gives untrusted issues
> subprocess.check_output(['hg', 'root'], env=os.environ)

So... curiouser and curiouser - it looks like it's not *actually* execve's
fault after all. I just compiled the code from the man page, tweaked it to
run 'hg root', and passed it a new environment. No problems. Well, then I
manually called the posix one from Python and thing worked fine. *Then* I
actually tried the above code, and *it* worked fine.

However I *still* get problems with the post-review code. So it looks like
when I get back to work on Monday I'll be looking to see what the
difference in environment is there.

-Wayne

[Antoine Pitrou]

Hello,

Wayne Werner <wayne <at> waynewerner.com> writes:
>
> So... curiouser and curiouser - it looks like it's not *actually* execve's
> fault after all. I just compiled the code from the man page, tweaked it to
> run 'hg root', and passed it a new environment. No problems. Well, then I
> manually called the posix one from Python and thing worked fine. *Then* I
> actually tried the above code, and *it* worked fine.
>
> However I *still* get problems with the post-review code. So it looks like
> when I get back to work on Monday I'll be looking to see what the
> difference in environment is there.

Your problem is a user question for the Mercurial mailing-list, not a Python
problem. See http://selenic.com/mailman/listinfo/mercurial

Regards

Antoine.