Re: New module - (session/authentication) seeking a name

2004-05-15 Thread David Nicol

[EMAIL PROTECTED] wrote:
I hadn't heard of AIS before. Sounds like it would make a nice additional
authentication method. Part of my TODO is to abstract both the
authentication and storage methods. I should be able to add this auth
method after that (though, it doesn't offer any group based permissions,
so that would still need handled locally).
AIS provides an authenticated e-mail address, in one line near the
start of a CGI program.
I have built several access control lists based on it, setting up
group based permissions implies dividing everyone into groups --
something that makes no sense until you know who you have.

I looked over AIS::client... there are a lot of exit statements in there,
and a lot of hardcoded HTML. It'd be nice if allowed the user of the
module to handle those parts, and just supplied the information it would
need (the full URL of the AIS server when needed, etc), just a thought.
And if it returned some failure condition, it wouldn't have to exit(), and
could allow the client to handle those errors.
the exits are because the AIS client has to run several times during
the initial handshake.  It redirects, then exits, both to establish a
session cookie and to redirect to the AIS server.  When the session
is established, it just looks up the identity and returns.  The only
thing the user sees from the AIS client module is the redirection pages.
The AIS server provides the log-in interface.
It might be a while before I get the authentication and storage methods
abstracted (shouldn't be all that difficult, but it's not on the top of my
TODO list right now). When I do, I'll definately look into AIS, and see if
I can find similar resources out there.
Wonderful.
The dot-gnu authentication committee has abstracted authentication
methods into a common interface, so if you want to pick up some
pre-built wheels instead of reinventing all of them yourself there
might be something useful in http://sourceforge.net/projects/macs
--
[EMAIL PROTECTED]
There's a fine line between participation and mockery -- Scott Adams


Re: New module - (session/authentication) seeking a name

2004-05-15 Thread Lincoln A. Baxter
And what does AIS stand for?  Or did I miss that.

Lincoln

On Fri, 2004-05-14 at 19:50, David Nicol wrote:
 [EMAIL PROTECTED] wrote:
  
  I hadn't heard of AIS before. Sounds like it would make a nice additional
  authentication method. Part of my TODO is to abstract both the
  authentication and storage methods. I should be able to add this auth
  method after that (though, it doesn't offer any group based permissions,
  so that would still need handled locally).
 
 AIS provides an authenticated e-mail address, in one line near the
 start of a CGI program.
 
 I have built several access control lists based on it, setting up
 group based permissions implies dividing everyone into groups --
 something that makes no sense until you know who you have.
 
 
  I looked over AIS::client... there are a lot of exit statements in there,
  and a lot of hardcoded HTML. It'd be nice if allowed the user of the
  module to handle those parts, and just supplied the information it would
  need (the full URL of the AIS server when needed, etc), just a thought.
  And if it returned some failure condition, it wouldn't have to exit(), and
  could allow the client to handle those errors.
 
 the exits are because the AIS client has to run several times during
 the initial handshake.  It redirects, then exits, both to establish a
 session cookie and to redirect to the AIS server.  When the session
 is established, it just looks up the identity and returns.  The only
 thing the user sees from the AIS client module is the redirection pages.
 
 The AIS server provides the log-in interface.
 
  It might be a while before I get the authentication and storage methods
  abstracted (shouldn't be all that difficult, but it's not on the top of my
  TODO list right now). When I do, I'll definately look into AIS, and see if
  I can find similar resources out there.
 
 Wonderful.
 
 The dot-gnu authentication committee has abstracted authentication
 methods into a common interface, so if you want to pick up some
 pre-built wheels instead of reinventing all of them yourself there
 might be something useful in http://sourceforge.net/projects/macs