Graham Dumpleton
grahamd at dscpl.com.au
Thu Feb 10 23:37:25 EST 2005
Rob Nichols wrote .. > On Thu, Feb 10, 2005 at 06:30:32PM -0600, Shawn Harrison wrote: > > Put the following in httpd.conf: > > > > SetHandler mod_python > > PythonHandler module.path.to.handler.file > > > > Then in the handler file, def a handler that looks for python scripts > > and loads them based on the req.uri. The attached very crufty handler > > might be a good source of amusement. > > Are you using this in a production environment? If so, can you give a > rough > estimate of the cost of "a little python and usually req.sendfile()" compared > to apache serving up text files? > > I'm quite new to mod_python, just toying with it for some pet projects. > It > seems really cool, but I don't have a feel for the performance yet. I'll > do > some tests "some day" (before I do anything serious), but curious about > what > others have found. When you use "req.sendfile()", the job of sending back the contents of the file is carried out by Apache. Thus, it isn't strictly being done by any Python code. The alternative to using "req.sendfile()" in this case where the physical file resides at the same spot as the URL would match to anyway, is to simply return "apache.DECLINED". By doing this, Apache will continue to process the request and will serve up the contents of the file as per normal. The only real reason to use "req.sendfile()" is where the physical file is situated in a different location to that where the URL would map to if you let Apache do its job. For example, if the physical file actually resided outside of the document tree. Graham
|