output = [fp.read()]
               fp.close()

which causes entire content is loaded to the memory (and not only once
by my observation). This is unacceptable for large files. I found this
to be much more memory & CPU efficient:

   class BlockIterator(object):
       def __init__(self, fp):
           self.fp = fp

       def __iter__(self):
           return self

       def next(self):
           chunk = self.fp.read(20*1024)
           if chunk:
               return chunk
           self.fp.close()
           raise StopIteration

   ...
               output = BlockIterator(fp)

The iterator spits the content by 20KiB blocks to the network socket and
closes the file at EOF.

I have servers that are using Apache/mod_wsgi and ’startup.wsgi’ scripts that essentially look like this:

#!/usr/bin/python
import os, sys
sys.path.append('/export/home/users/djangoprojects/OurSite')
sys.path.append('/export/home/users/djangoprojects')
os.environ['DJANGO_SETTINGS_MODULE'] = 'OurSite.settings'
import django.core.handlers.wsgi
application = django.core.handlers.wsgi.WSGIHandler()

I don’t know enough about CherryPy? to make comparisons, but could this method replace the startup.wsgi code cleanly? I’m thinking about ease of deployment here.

Cheers, Tone

At least, according to some comments at http://www.eflorenzano.com/blog/post/hosting-django-site-pure-python/, CherryPy is the tool for the job.