Hi ppl, Thanks for your answers. I've looked to everything you suggested, but many of the solutions were in Java, Pascal, or even worse, proprietary software. It can be a good example for a front-end, but i think that it's for the best if the engine is directly integrated in Koha, and therefore to use a backend in Perl, and a frontend with templates, with export possibilities (rss, csv, etc). I've begun working on a couple of script that would work with the frontend of Nicolas Morin: http://morinn.free.fr/koha/canned_reports.htm because this interface can easily be used by a librarian with no IT skills, but is flexible enough to extract almost any interesting piece of information from the database. Any crits or ideas would be welcome, btw, from next week on, i'll be in vacation, far from any kind of network, and i'll be back by the end of August, so feel free to work on the reports engine even if i can't answer or give my opinion during this time :) /pierre