On 11/8/05, hugo <[EMAIL PROTECTED]> wrote: > Although that could easily be solved by dumping the current mod_python > handler in favor of a mod_python WSGI handler and only using the WSGI > stuff. That way we could make much better use off WSGI middleware and > other things Ian is working on.
My thought is YAGNI on this one. Until there's a demonstrable benefit in moving the mod_python handler to use WSGI, there's no point in adding an entire layer of abstraction over mod_python. I also like the fact that the mod_python handler is optimized for mod_python, not having to jump through a bunch of hoops to convert to WSGI only to be converted *again* from WSGI into our custom request/response objects. For that reason, if we do end up deciding to route the mod_python handler through WSGI, I'd want to do some benchmarking, and keep a non-WSGI (raw) mod_python handler as the prefered mod_python handler for performance -- assuming there's a measurable performance difference. Adrian -- Adrian Holovaty holovaty.com | djangoproject.com | chicagocrime.org
