Perhaps your web server is providing the correct SCRIPT_NAME,
and the problem is introduced in Publisher.process(), which attempts
to work around a common, similar SCRIPT_NAME/PATH_INFO problem in
a way that would make matters worse for your setup.
Here's the code I am talking about:
def process(self, stdin, env):
"""
Return a Hit with a complete response.
Subclasses that want to customize the Hit class may do so by
overriding
this method.
"""
if env.get('SCRIPT_NAME') and env.get('PATH_INFO') == '':
# The "script" here is your QP application, so the
SCRIPT_NAME
# should be the part of the url, after the host name, that
# causes requests to be handled by your QP application.
# The PATH_INFO part of the path after the SCRIPT_NAME.
# We expect applications running at the top of the URL
space
# to get requests where the SCRIPT_NAME is the empty string
# and the PATH_INFO starts with a '/'.
# Your HTTP server is not providing the expected CGI
environment.
# The full path part of the URL is being put in the
SCRIPT_NAME.
# We'll try to make it work anyway, but we are assuming
here
# that your application is running at the top of your
URL space.
# If you want your application to run somewhere else in
your URL
# space, you must change your server so that it provides
the
# correct environment variables. Alternatively, your
publisher
# class can override this method and hack the
environment as
# needed to fit QP's expectations.
env['PATH_INFO'] = env['SCRIPT_NAME']
env['SCRIPT_NAME'] = ""
hit = Hit(stdin, env)
self.process_hit(hit)
return hit
In your case, I think you could try overriding this method and
changing any empty PATH_INFO to "/" and removing the "if" statement that
appears above.
Even easier, perhaps, would be to put a redirect in your web server
config
so that '/app.cgi' always redirects to '/app.cgi/' before qp sees it.
_______________________________________________
QP mailing list
[email protected]
http://mail.mems-exchange.org/mailman/listinfo/qp
