Re: Quixote, WSGI, next-gen hypishness, etc

Titus Brown wrote:
> 	Will Quixote, in the future, move only in directions that are
> 	primarily useful to mems-exchange?
> 
> I don't want to be combative, I'm truly curious ;).

 From my perspective, it would be easy for Quixote development to open
up in this way if it uses WSGI in a more core way.  Which is why it
seems like a natural directory for Quixote.

So, taking session2 as an example, that could be usable as WSGI
middleware.  It *would* require some framework hooks at some level, to
make the object available as part of the Quixote request object.  Those
could probably be generalized, or who-knows-what, but there needs to be
something.  But all of the hard work can go in the middleware, which
doesn't need to be part of the Quixote package directly.  And the hook 
isn't hard to include in Quixote; in contrast, actual session 
implementations require a fair amount of maintenance, and reluctance on 
MEMS' part to taking that code in is quite reasonable.  And if session2 
stops being maintained, someone can just use someone else's middleware 
with a Quixote hook.

David pointed out the issue: "Our business has never  been,
and will probably never be, web framework development, or
coordinating open source projects."  And that would be fine if Quixote 
had a more "open" framework, and that doesn't have to mean 
everyone-edits-the-code, it can just mean giving people an architecture 
it is easy to plug in to.  WSGI is one such architecture; notable since 
it's the only one that people are using in this way.  Well, that and 
some stuff in Zope, but I'm guessing people here aren't thinking about 
moving to Zope.



-- 
Ian Bicking  /  [email protected]  /  http://blog.ianbicking.org