Re: Does smart-http need git-daemon-export-ok?

From: Shawn O. Pearce <spearce@spearce.org>
Date: Dec 27, 2009, 21:06 UTC
Message-ID: <20091227210653.GA609@spearce.org>
In-Reply-To: <7vk4w963np.fsf@alter.siamese.dyndns.org>

Junio C Hamano <gitster@pobox.com> wrote:

Show 6 quoted lines

> Tarmigan <tarmigan+git@gmail.com> writes:
> > Should the git-http-backend check something like git-daemon-export-ok
> > before serving a repository?
> 
> I'd agree that it would make sense to have a way to mark individual
> repository for (or not for) export.

Just for some background... early drafts of git-http-backend actually did check for, and require, this file before it exported a repository.

I took the check out because I was relying on the HTTP server's document root translation to provide the mapping into the local filesystem. That meant the HTTP repository was already exported via dumb-http, and the git-daemon-export-ok flag wasn't being checked.

Later in the series development we got the patch to allow a
different filesystem root via an environment variable, which means
its possible to hide repositories and make them available only
through git-http-backend.  In that configuration, checking the
git-daemon-export-ok flag makes sense again.

Show 8 quoted lines

> In "native" case, the chain of events are: client talks to the daemon, the
> daemon checks and decides to (or not to) export, and it runs upload-pack.
> 
> In "smart http" case, http-backend is one half of what corresponds to the
> daemon (the other half being your http server configuration), and it is
> more flexible and git specific half, so I'd say it would make sense to
> implement the check that honors the same git-daemon-export-ok flag file in
> it.

Yea, I'd agree.

-- 
Shawn.