On Mon, May 24, 2004 at 11:35:51AM -0700, Titus Brown wrote:
> The critical issue is the authentication cookie (I'm assuming that's
> how Gallery works...).  If you can write a Python system that sets
> that cookie in a fashion that Gallery will accept, then the rest
> should "just work".
>
> If you can make the Quixote system accept the Gallery auth cookie,
> then you could use either, or both, systems for logging in...

I assume PHP/Gallery works in similar fashion to Quixote session
management: there is a PHPSESSID session ID cookie, which simply
identifies the session. This is all that is "stored" on the client
side, server side there is then data associated with the session,
namely whether the user has logged in, and what his username is. The
result is that it is probably not possible to solve this problem by
simply setting cookies, you would have to get into PHP's session
datastructures from Quixote, or into Quixote's from PHP. Hence the
speculation that I may need to move some of this session information
into a shared data area, like the DB, or filesystem.

This is all purely speculation on my part though, I see a PHPSESSID
cookie, and I know a bit about Quixote web programming. Tie that
together with a bit of logic and programming experience, and that is
what I conclude.

My thought experiments thus still lead to: - auto fill-in the login
form from my Quixote code, adjusting the cookie as necessary and
passing it on to the client - hoping PHP does not mind the login
coming from localhost, and the rest of the interactions from the
client, else mess with Gallery's session code, make it obtain its
info from a DB, which can then be accessed/modified from Quixote as
well. Problem with the latter approach might be that I might still
not be able to get at the PHP session ID cookie from my Quixote app.
It does look like it is global though (path is set to /) - so maybe
it is possible?

I hope someone with a little more experience can shed light, or confirm
any of my suspicions here. If it sounds right, or sounds wrong, to
anyone, feel free to inform me as such.

Thanks,
Hugo van der Merwe

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFAsntFSmzXbWBxVpgRAqNtAJ0Yu4uOfYOPSur/kDpYI+xeBlZrIwCePbKL
uRKGTLJOO337pBicMcqBfII=
=5M8W
-----END PGP SIGNATURE-----