I finally figured out how to get mongrel2 + uwsgi + web2py running and I
*definitely* need the final slash after the route.
mongrel2.conf exerpt:
...
hosts = [
Host(name="127.0.0.1", routes={
'/tasks/static/': Dir(
base='./tasks/static/',
index_file='index.html',
default_ctype='text/plain'
),
'/tasks/': Handler(send_spec='tcp://127.0.0.1:3033',
send_ident='54c6755b-9628-40a4-9a2e-cc82a816345e',
recv_spec='tcp://127.0.0.1:3032', recv_ident='',
protocol='tnetstring')
})
]
...
Leave out the final slash in '/tasks/' and the request for
http://localhost/tasks/edit/index will be taken by the static handler as
per the mongrel2 documentation.
In my web2py code I throw away the useless PATH_INFO ( '/edit/index' or
something similar ) and build the PATH_INFO and the QUERY_STRING from
REQUEST_URI.
QUERY_STRING is set to '' if there is no '?' in the request, (like
mod_wsgi does), which web2py relies upon.
I suggest not setting the SCRIPT_NAME to the PATTERN (stripped of the
final '/' or not). If you have parenthesis in your pattern you get that
along as well.
SCRIPT_NAME in mod_wsgi is the empty string, just take that.
Take the full 'PATH' to set 'PATH_INFO' and parse.
Set QUERY_STRING to "" if QUERY is not found.
If I would need to leave out '/tasks' in my web2py handling I will do
the url rewrite there. That is not the task of uWSGI interpreting what
mongrel2 is not doing in the first place. I got the impression that the
code in proto/zeromq.c is trying to do exactly that, but correct me if I
am wrong.
_______________________________________________
uWSGI mailing list
[email protected]
http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
