I have a few web applications that use Quixote [1] to perform path traversal (Quixote having borrowed the idea from Zope). In my setup, Apache passes the request on to one of a number of long-running SCGI [2] (similar to Fast CGI) processes, which then call upon Quixote. I'm interested in alternatives to using several SCGI processes. There has been discussion in Quixote circles on having an integrated HTTP server, and some work has been done with Medusa and Twisted. The attraction with both of these is the possibility of using a lighter mechanism than multiple processes for dealing with concurrent requests. However my experience with the Quixote/Medusa approach had some troubles, and I now have some uncertainties about whether Twisted would be the right approach [1]. From what I've read on this list, Zope 3's HTTP server looks like it has a good design; my question is whether it is in a ready enough state yet to do simple object publishing on a production server, either using Zope 3's own path traversal mechanisms or by passing the request on to Quixote. I wouldn't need Zope's security mechanisms, acquisition, object database, or other stuff, so it wouldn't matter if those parts of Zope 3 are not yet ready. [1] http://www.mems-exchange.org/software/quixote/ [2] http://www.mems-exchange.org/software/scgi/ [3] http://lists.zope.org/pipermail/zope3-dev/2002-February/000516.html Hamish Lawson