Andy McKay wrote:
Of course the only problem with a hit counter is that high traffic requests mean continually writing the object into the ZODB each time it gets hit / downloaded, whatever. Thats why I release FSCounter which uses the file system. I would like to expand this to cover multiple objects...
The problem is slightly more complicated than this: We have Apache cache file and image objects. This is a good thing, because they are static, and Apache will *always* be better than Zope at serving static content. This is a bad thing, because direct requests for files don't ever make it to Zope. Nonetheless, this doesn't affect the /view method. Further, we now have the feature that the listing of products shows the latest version, no /view method in the middle, for your direct download. I may be able to do some parsing of the apache log files for those things that end in tgz, tar.gz , and zip. This would be somewhat canonical, except for the fact that because we have cache headers set on the file objects, they are cached downstream, and so many requests for cache objects will never make it to our Apache at all. ~ethan p.s. I'd really appreciate it if requests for improvements of zope.org be sent to zope-web@zope.org.