The Ape (runnable, source) has a new committer: David Calavera, who’s from Spain. He wrote me out of the blue saying “Here are some patches that are the beginning of an RFC2617 framework.” David needed the Ape to talk WSSE for his own work.
This is exactly the area where the Ape needs work. I think that as far as exploring the workings of an Atom Protocol implementation, most of the useful work has been done. Once you get past that stuff, I’m pretty sure that the main source of interop pain in the Atom space is going to be authentication. the Web offers lots of options; fortunately the RFC2617 framework may be useful in sorting things out.
Among other things, David’s patches provide a place to put the code for GData authentication, once Google provides it.
Contributions
Comment feed for ongoing:
From: DeWitt Clinton (Aug 10 2007, at 16:56)
Hi Tim,
This is great! I actually have "Add GData auth support to APE" on my own personal TODO list, and this change will make it considerably easier.
Unfortunately, that TODO is preceded by the one that reads: "Learn Ruby"
But given what I've heard, Ruby should be easy for someone to pick up if they already know Python.
Hopefully I get a chance to poke at this over the weekend.
-DeWitt
[link]
From: John Cowan (Aug 14 2007, at 12:53)
I'm now on the Google GData API team. How can I help?
[link]
From: Tim Bray (Aug 14 2007, at 13:26)
John: GData *might* be a conforming APP server implementation, such that you could point a generic APP client at it and have hopes of getting a useful result.
The trouble is that we don't know yet, because the google-authent voodoo is kind of a high hoop to expect J.Random Coder to jump through. A good answer would be for Google to provide code in Ruby and Python and Java and C# and Perl and so on implementing Google authent in a 2617-compatible way, so that authors don't have to write their own, they can just wire it in.
Now that David has provided the 2617 core in the Ape code, it might be a real easy place for someone to start on the problem of building such libraries. It'd have the side benefit that then you'd be able to run the Ape against GData; the Ape isn't perfect but has been run against literally dozens of APP server implementations. Chances are that if get an error-free run, you have an excellent chance the next random client will Just Work. Holler at me offline if you want to talk this over.
[link]