Open jjatria opened 7 years ago
I'll re-iterate my statement from issue #16 and say that I'm a fan of the pluggable strategy way of doing things, still not entirely sure on the details of it but that's the general direction I'm leaning towards :)
As part of a discussion on #mojo
, the following ideas came up ref plugins:
under
should be the mechanism for authentication, and over
(and the condition
"authenticate" in this module) should not be used@jberger also suggests https://metacpan.org/pod/Net::SAML2 is a good implementation.
@mohawk2 Are there any links / logs to the discussion on #mojo
? I don't think I've seen anything, and I'm curious about the under
/ over
issues (not because I'm particularly partial to either, just curious).
Logging is forbidden on #mojo
, so no. While I think of it, could you release M::P::A? The async code has been merged (thanks!) but not yet released?
My apologies.
I pushed it out as a TRIAL release, and there were some test failures reported by CPAN testers that I didn't have the time to track down at the time: http://matrix.cpantesters.org/?dist=Mojolicious-Plugin-Authentication%201.34-TRIAL
Looking at them now, it seems there's only two failures out of 194 tests, so it's not as bad as I had thought. Then again, one of them seems to have died with
Can't use string ("Mojo::Promise") as a HASH ref
I seem to remember trying to reach out on IRC when I got this report, but that message must have got lost in the mists of time. I'll see if I can see anything suspicious with the code, and hopefully push out a proper release.
This issue is to track the inclusion of other authentication methods, as suggested by @carragom in #16.
From that issue:
Currently, the only way to allow for these alternative authentication methods is to bypass the authentication plugin and setting the current user manually. The feature discussed here would solve this in a more controlled manner (while still giving the user the freedom to crash and burn if they want).
Contributions and discussion are welcome!