question on service commands in http-backend.c

Sitaram Chamarty· Sep 5, 2010, 15:29 UTC · lore

Hi,

Short version: can I assume that, of all the service commands [I assume that's what they're called] in the services[] array in http-backend.c (approx line 530), only 'git-receive-pack' represents a *write* attempt?

Long version: I just managed to fixup gitolite so it can be used with smart http also, and after all that was done I got to wondering about this. It matters to gitolite because it needs to recognise whether the user is trying to perform a write or a read so it can make the correct access control decision.

With ssh it was very simple: you only had 3 command (git-upload-archive, git-upload-pack, git-receive-pack) and the first 2 are "read"s.

And yes, I do notice that these new ones are all "GET" so it is hardly likely they'd be writes -- just wanted to make sure...

Thanks,

-- 
Sitaram

Shawn Pearce· Sep 5, 2010, 21:20 UTC · re: Sitaram Chamarty · lore

Re: question on service commands in http-backend.c

On Sun, Sep 5, 2010 at 8:29 AM, Sitaram Chamarty <sitaramc@gmail.com> wrote:

> Short version: can I assume that, of all the service commands [I
> assume that's what they're called] in the services[] array in
> http-backend.c (approx line 530), only 'git-receive-pack' represents a
> *write* attempt?

Correct.

> And yes, I do notice that these new ones are all "GET" so it is hardly
> likely they'd be writes -- just wanted to make sure...

Right.  :-)

-- 
Shawn.