On 3/12/07, Earl Miles <merlin@logrus.com> wrote:
It would be pretty cool if we could script your analysis so that we can just dump the awstats data into it every month.
The analysis is basically a spreadhseet I made that contains a lookup table of all modules, themes, and translations and has some reference data for them. I copy the entire awstats data into that and do a "text to columns" on it. Then I have a few functions that parse out the filename and the version from the filename (e.g. so that trip_search-5.x-1.2.tar.gz and trip_search-5.x-1.1.tar.gz are both labeled as the "5" version of trip_search). And then I manually do a handful of pivot tables to aggregate the data and see which items are most popular. I'll send that to anyone who wants it. I believe the last time we discussed more automated solutions we got stuck at the best implementation. If we inserted a counter between the link on the project page and the ftp server we would have easy data collection into drupal database, but lose some people who get it directly from FTP. If we use awstats output then we get basically everyone but miss folks who pull from CVS directly (which probably isn't statistically significant - especially not now with the new release system). Awstats also has the problem of scripting something to parse that data. I believe dww was in favor of the awstat->project solution and Dries in favor of the "In project.module solution" http://lists.drupal.org/archives/development/2006-11/msg00264.html There are even issues for some of these things http://drupal.org/node/66015 - browse by popularity http://drupal.org/node/52475 - Display usage data for projects http://drupal.org/node/79550 - Automate gathering of quality metrics (though it's kind of a meta issue and should probably be split) I'd love to work on some/all of those, but have lost my round tuit. Maybe I'll find it at Drupalcon ;) Greg