Here's a summary of the open space, as much as I remember. Ben discussed the state of Pylons, Chris BFG, Jorge TurboGears, Robert CherryPy, Ian WebOb/Paste, one guy with a brand-new framework called We, and another guy with his own web code. There were several Pylons users, three BFG, and two TurboGears.
Pylons, TurboGears, and BFG all seem to be in a similar development phase. Feature development has slowed down significantly and seems to be "finished", but internal cleanups still have a way to go. Ben said most of the changes in Pylons 1.0 were not in Pylons but in Beaker, Routes, and WebHelpers. I think all the frameworks deserve a congratulations for maturing to the point that their features are pretty set and they can focus on optimization. There was a discussion about what to do with WSGI. Ian and (Robert?) are going to finish up a 1.1 and 2.0 spec, implement it in WebOb, and send it to the PEP editors. We can let the PEP editors worry about whether tto pursue it as an official change to Python. The rest of us can just use WebOb and forget about it. WSGI 1.1 is just a clarification of some obscure passages, and changing a few details to what the established implementations actually do (e.g., requiring a size arg on the read method). Probable 2.0 plans: - ``mod_wsgi`` has, and ``CherryPy`` will have, options to handle the headers in multiple ways (bytestrings vs Unicode), so that they'll work with whatever the final spec is (or whatever the user wants). - wsgiref decodes all headers as Latin-1, following the HTTP spec. This may be OK because it produces a reversible translation of all possible byte values without raising an exception. The non-ASCII characters may not be correct, but it's difficult or impossible to tell the intended encoding anyway, so they're already hosed. - SCRIPT_NAME and PATH_INFO may be handled differently because they're supposed to be UTF-8. Anyway, Ian & co. will just decide "something" for Unicode and put it in the spec and WebOb. - The much-hated ``start_response`` is going away in 2.0. It doesn't seem to have any defenders. There a discussion about something async but I don't remember what. It wasn't about the NOT_READY token. Maybe something WebOb would do? Ian is all excited about TopCloud, a cloud server configuration for Ubuntu. PylonsHQ is using it, and Ben says it makes it real easy to propagate updates, just two commands. Almost everybody is using WebOb. Ian wants to get it into the stdlib. Who knows if it will. Ben's next project after 1.0 is to refactor PylonsApp. Three particular goals: - Make it easier to subclass, particularly to override the routing and dispatch and other things people want to do. - Add events. This would be similar to Apache modules, providing additional behavior at certain poinds. So they would have callbacks for server_starting, server_quitting, begin_request, end_request, etc. - Get rid of the magic globals. ``request`` would become a controller attribute (passed to the constructor, which most applications do not define). ``response`` and ``session`` would become attributes of the request. (I think Zope puts the response on the request, and WebWare did too?) I'm not sure what would happen to ``app_globals`` and ``url``, maybe the same? There's no timeline for when this might happen, or whether it will be exactly as described. -- Mike Orr <[email protected]> -- You received this message because you are subscribed to the Google Groups "TurboGears" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/turbogears?hl=en.
