Hi, On Sun, Aug 24, 2008 at 12:24:38PM -0500, Raphael Geissert wrote: > Package: edos-debcheck > Version: 1.0-7 > > Using the attached file: > > $ edos-debcheck -quiet < Packages_board > row1 (= 1.0-1): OK > cell1.1 (= 1): OK > cell1.2 (= 2): OK
Thanks for your bug report and for trying this somewhat unexpected application of edos-debcheck. The problem of solving package dependencies is NP-complete in general, which means that most probably no algorithm can solve efficiently any random instance of it. Edos-debcheck was designed to solve efficiently dependency relations that stem from actual Package files, that is Package files describing a "real" distribution. It would certainly be posible to tweak edos-debcheck to also solve that particular problem (since soduko is solved in no time by a finite-domain constraint solver). However this would yield no advantage for better solving the problem that edos-debcheck was designed to solve. I will hence tag this as "wontfix". Having said that I still find it interesting and amusing to see how one can systematically construct difficult problem instances, even if these are compleley artificial and have noting to do with real-life instances (I have seen Daniel's blog post with the python code). We will probaby use this as hard test case in the future dependency solver competition ... Cheers -Ralf. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
