Re: [Zope-dev] Single Sign On

2009-02-18 Thread Wichert Akkerman
Previously Shane Hathaway wrote: Alternatively, I have wondered if we actually need full-blown SSO; perhaps a carefully constructed domain-wide cookie would do the trick. Any experiences with that? auth_tkt based cookies sounds like a good option, possibly combined with something like SQL or

[Zope-dev] Zope Tests: 7 OK, 1 Failed

2009-02-18 Thread Zope Tests Summarizer
Summary of messages to the zope-tests list. Period Tue Feb 17 12:00:00 2009 UTC to Wed Feb 18 12:00:00 2009 UTC. There were 8 messages: 8 from Zope Tests. Test failures - Subject: FAILED (failures=1) : Zope-trunk-alltests Python-2.5.4 : Linux From: Zope Tests Date: Tue Feb 17

Re: [Zope-dev] Single Sign On

2009-02-18 Thread Gary Poster
On Feb 17, 2009, at 7:55 PM, Shane Hathaway wrote: Gary Poster wrote: Launchpad uses OpenID. We don't have that slated for abstraction and open-sourcing immediately. However, most of the Launchpad code (including this bit) is to be open-sourced by this summer, abstracted or not.

Re: [Zope-dev] Single Sign On

2009-02-18 Thread Shane Hathaway
Gary Poster wrote: We use the OpenID 2.0 identifier select URL. This is a special OpenID url that basically means: identity using whatever ID you have on that server. The OpenID response will contain the actual OpenID identifier of the user at the end of the request. So sites that

Re: [Zope-dev] Single Sign On

2009-02-18 Thread Reinout van Rees
Shane Hathaway schreef: * The SSO process should be very similar to an ordinary cookie-based login process. I don't want the user to have to enter their username on one form and their password on another, but that's the standard OpenID process. * This will be implemented in Zope 3.

Re: [Zope-dev] Single Sign On

2009-02-18 Thread Shane Hathaway
Reinout van Rees wrote: So: easiest way is to let some trusted apache plugin handle the hard part and then laugh all the way to the bank with some 100-line authentication plugin. That would usually work, but in this case, customers will be doing their own installation, so we need to keep

[Zope-dev] Proposed Patch for zope.site

2009-02-18 Thread Shane Hathaway
Today I ran into an exception masked by Zope 3. I found the code that was masking the exception and fixed it locally, but since this small bit of code has no docs or tests, I can't be sure I won't break stuff if I check in my change. What do y'all think I should do? Here is the patch:

Re: [Zope-dev] Single Sign On

2009-02-18 Thread Marius Gedminas
On Wed, Feb 18, 2009 at 09:00:10AM -0500, Gary Poster wrote: On Feb 17, 2009, at 7:55 PM, Shane Hathaway wrote: Gary Poster wrote: Launchpad uses OpenID. We don't have that slated for abstraction and open-sourcing immediately. However, most of the Launchpad code (including this

Re: [Zope-dev] Single Sign On

2009-02-18 Thread Shane Hathaway
Marius Gedminas wrote: It's my impression that launchpad.net is an OpenID provider only, while Shane is trying to figure out how to use the OpenID consumer API in AuthKit. No. I am going after the more conventional single sign on use case where many consumers depend on only one centralized