Re: twisted

On 03/29/05 19:10, Blake Winton wrote:
> Steven Armstrong wrote:
>> CGI        >> /pyblosxom/web/pyblosxom.cgi
>> WSGI_CGI   >> /pyblosxom/web/pyblosxom_wsgi.cgi
>> twisted    >> /pyblosxom/web/pyblosxom.tac
>> mod_python >> /pyblosxom/web/wsgi_app.py
> 
> I'm a little confused as to how this last one is set up.  I can see that 
> the WSGI_CGI uses the application from wsgi_app, but I sort of expected 
> some code like:
>          from wsgiref.handlers import ModPythonHandler
>          from wsgi_app import application
> 
>          ModPythonHandler().run(application)
> in the mod_python case...
> 

The mod_python WSGI wrapper is not part of the wsgiref library. I wrote
it based on a example from the web-sig mailing list. You can find it at [1].

[1] http://www.c-area.ch/code/mp_wsgi_handler.py

> (Keeping in mind that most of my confusion stems from my inexperience in 
> this area, and not because your code is necessarily confusing.)
> 
> And while I'm here asking dumb questions, is it possible to run the 
> twisted service along with other services on the same port?  Does it 
> integrate nicely with other twisted services trying to run on port 8080, 
> or do we need a new port for every twisted service?
> 

I guess you should be able to run it same as any other twisted resource.
e.g. something like:

from twisted_wsgi import WSGIResource
from wsgi_app import application as pyblosxom_application

application = service.Application("xyz")
page = SomePage()
resource = WSGIResource(pyblosxom_application, async=False)
page.putChild('pyblosxom', resource)
s = internet.TCPServer(8080, server.Site(page))
s.setServiceParent(application)

Which should give you a Pyblosxom blog at localhost:8080/pyblosxom.
I can't remember the exact syntax for twisted, but it should look
something like the above example.

>> Where WSGI_CGI, twisted and mod_python all run the same program, namely
>> wsgi_app.py. They all use server dependant wrappers/connectors to setup
>> a WSGI compliant environment. Read the source for further infos and
>> dependencies.
> 
> Okay, so, are we agreed that we can remove the pyblosxom.cgi, as it 
> should be logically equivalent to pyblosxom_wsgi.cgi?  (And browsing 
> through the source, they look _extremely_ similar, which worries me as a 
> coder.  Twice the chance to make mistakes.)
> 

The problem here is that pyblosxom_wsgi.cgi depends on the wsgiref
library. I don't think users who only care about CGI should be forced to
install that.

The reason why pyblosxom_wsgi.cgi looks mostly like pyblosxom.cgi is
static rendering. The wsgi_app.py doesn't know anything about that, so
it has to be done in the cgi. I agree that it would be nicer to refactor
the code which is identical to some other place where both scripts have
access to it.
On the other hand, pyblosxom_wsgi.cgi was ment to be another testcase
for the wsgi-related code, not a replacement for pyblosxom.cgi.

>>>It looks like WSGI is a way to write our app, and have it 
>>>"automatically" connect to Twisted or mod_python or CGI, or what have 
>>>you.  If that's really the way it works, it would make our lives (by 
>>>which I mean Steven's life) a lot easier, since we could just write the 
>>>app once, and have the connectors deal with the various backends.
>> As noted above that's exactly how its implemented :-)
> 
> In the words of Will, "w00t!"  ;)
> 
> For a future direction, would it be useful to make the 
> Pyblosxom.pyblosxom.PyBlosxom class be more like an application, so that 
> we could get rid of the wsgi_app.py class entirely?  I think that might 
> be a slightly cleaner design, just from my 50,000' point of view.
> 

I totaly agree on this. I think it would generally be good to move from
the "collection of scripts" design to a more "application" design.

cheers
Steven




-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click