Maciej Pasternacki

Discovery is almost finished now. HTML-based needs only tests written, and for Yadis, only XRDS analysis is left (plus tests, of course). For HTML parsing I use CL-HTML-Parse – it's light, doesn't drag any dependencies (only Allegroism it needs is the IF* macro, which is included), and handles HTML tag soup quite well.

Although I'm writing fairly separate bits of functionality, the need for some actual code architecture starts to show up. For every identity, the relying party gathers bits of information during various stages of protocol (and there are already two stages almost done), so I need to choose representation for this state. For now, to stay flexible while I need it, I made functions pass an alist with the information – initial alist is accepted by each function, and updated (possibly destructively) alist is returned. This is enough for now, and when (if) the need occurs, this shouldn't be hard to convert to proper object- or structure-based design.

Work done since last report

• Researched and choose HTML parsing library
• Researched public identity providers for reference identities
• Implemented HTML discovery

Problems

Almost every public provider provides OpenID version 1. This makes compatibility with v1 much more important than I predicted, especially in the relying party. Also, many of those actually use Yadis, so its priority goes up. Yadis protocol, however, is really simple and I have most of it already done (no actual XRDS parsing yet, but this format is fairly simple and shouldn't take long).

Plans for next half week

• Tests for HTML-based discovery and its helper functions (possible refactoring for better testability)
• XRDS parsing & tests

Because of exams (two last finals on Monday and Wednesday, wish me luck – especially on Wednesday), I'll probably send next report on Wednesday. After it, back to regular Tuesday/Friday schedule.