Re: [IronPython] Rewrite of WSGI handler

2006-11-02 Thread Mark Rees
On 11/3/06, Christopher Baus <[EMAIL PROTECTED]> wrote: I'm beginning to wonder if we really need the reload IronPythonfunctionality in WSGI.  It adds some complexity for not a lot of realgain.  I thought one reason for reloading the interpreter would be toreload all the modules, but that can be do

Re: [IronPython] Rewrite of WSGI handler

2006-11-02 Thread Christopher Baus
I'm beginning to wonder if we really need the reload IronPython functionality in WSGI. It adds some complexity for not a lot of real gain. I thought one reason for reloading the interpreter would be to reload all the modules, but that can be done via WSGI middlewear. Also I ran into a problem wi

Re: [IronPython] Rewrite of WSGI handler

2006-11-01 Thread Christopher Baus
> In wsgi.py you comment "There isn't an obvious way to get request protocol > from HTTPRequest object". In the wsgi.py under fepy CVS that was checked > in > after IPCEr3 I used request.ServerVariables['SERVER_PROTOCOL'] which seems > to return the correct string. Mark, I updated my code to look

Re: [IronPython] Rewrite of WSGI handler

2006-11-01 Thread Mark Rees
Nice job Christopher. I have no problem with combining WSGI and WSGI-SI into one handler. Also testing with my simple wsgi apps under your code and xsp2 works ok so far. Have a more complex app which I will try later tonight. In wsgi.py you comment "There isn't an obvious way to get request protoco

[IronPython] Rewrite of WSGI handler

2006-11-01 Thread Christopher Baus
I spent the night hacking on the WSGI handler. Frankly the performance I was getting wasn't very good, and I'm making a grab at some low hanging fruit. Primarily the changes involves loading IronPython and the modules once at startup (rather than per instance). The code is a bit strange because