On Thu, 2014-07-24 at 23:31 +0200, Andre Klapper wrote:
Could you please elaborate a bit what your judgement that it's faster and more reliable is based on? (Even if a small library, I wonder how well that code is tested if it's not packaged by major distros.)
We're hoping that depending on qqwing will allow us to close these bugs, which would otherwise be difficult to solve: https://bugzilla.gnome.org/show_bug.cgi?id=619190 https://bugzilla.gnome.org/show_bug.cgi?id=725797 We'll also close these related bugs: https://bugzilla.gnome.org/show_bug.cgi?id=580055 https://bugzilla.gnome.org/show_bug.cgi?id=580056 https://bugzilla.gnome.org/show_bug.cgi?id=731215 The qqwing project is almost 10 years old and is solely dedicated to Sudoku generation, whereas none of the GNOME Sudoku maintainers (myself and Parin) are Sudoku generation experts. Most notably, our generator sometimes creates unsolvable puzzles (and also solves them :). I'm also hoping qqwing will be better at rating puzzles' difficulty, though we can't know for sure until we release and get feedback from users. We already have a good relationship with qqwing upstream (they accepted a significant patch from us recently) so we can get changes merged if we do run into problems. qqwing can generate over 100 Sudoku puzzles per second on my machine, whereas our current generator sometimes take a couple seconds to generate a single puzzle (this is unscientific, yes, and that is due to some bug we haven't found, but that's the point). If we need to wait until 3.16, that's not a big deal, it'd just be nice to close those bugs prior to 3.14, since 3.14 is a major rewrite/redesign and it'd be great for it to make as good a first impression as possible. qqwing upstream distributes an RPM and a deb, so many distros already have a good starting point for their packages. Also it has no dependencies. :) Cheers, Michael
Attachment:
signature.asc
Description: This is a digitally signed message part