marksibly wrote: > Found it in the FAQ; > [quote] > If you're using mod_python but not using Django's request handler, > you'll need to work around a mod_python bug related to the use of > SetEnv; before you import anything from Django you'll need to do the > following: > > os.environ.update(req.subprocess_env) > [/quote]
The FAQ isn't really correct in saying that is a mod_python bug, it is how Apache works and populating os.environ with the contents of req.subprocess_env is wrong and could cause all sorts of problems, especially in a multithreaded MPM and/or where there are multiple Django installations running on the one Apache web server and distinct Python interpreters instances are not used to separate applications properly. What the 'subprocess_env' table in the Apache 'request_rec' structure is for, is to hold any environment variables which should be passed to subprocesses which may be created for things like running CGI scripts. Some values in the 'subprocess_env' table may be specific to the current request being handled especially if req.add_common_vars() has been called. Because of this, the variables are passed direct to the exec call used to run the CGI script and are never pushed into the normal Apache process environment variables as you are doing. If you do push them into os.environ (even if in Python os.environ is local to the Python interpreter and doesn't actually change the process environment) and something is relying on reading the values from there, the problem you will have is that in a multithreaded MPM where requests can be serviced in parallel, a distinct request, possibly against a distinct Django application, could update the values of the environment variables to something different between the time that the first request handler set them and subsequently went to use them. This could result in things like a request being processed wrongly where it depends on an environment variable sourced from os.environ. So, if the FAQ does say that it should be changed as it is not strictly correct. If Django relies on getting configuration through environment variables or values stuffed into os.environ, then I would suggest it is possibly broken. At least now I can finally see why the mod_python/Django documentation might say that the worker MPM should not be used. I have queried in the past why the worker MPM could not be used but no one could give an answer as no one really knew where that statement originated and what the basis for it was. Some people would say it worked okay and some would say it wouldn't. I wander if the people for who it didn't work had a setup where they were relying on updating os.environ with a fiddle like the FAQ describes. FWIW, I have never seen anyone who has identified themselves as being involved with Django development on the mod_python mailing lists. I have never been able to understand this. To me, if you are going to provide support for running Django on top of mod_python, a good understanding of how mod_python and Apache works should be a requirement. :-/ Graham --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/django-users?hl=en -~----------~----~----~----~------~----~------~--~---
