RE: [Aisleriot] Freecell game selection, Redmond style

[Alan Horkan <horkana maths tcd ie>]

On Sun, 18 Feb 2007, Bastien-F Jonathan wrote:

> Date: Sun, 18 Feb 2007 21:30:01 -0500
> From: Bastien-F Jonathan <J Bastien-F cgodin qc ca>
> To: Alan Horkan <horkana maths tcd ie>
> Cc: games-list gnome org
> Subject: RE: [Aisleriot] Freecell game selection, Redmond style
>
> > This is a good thing but I should mention that the ability to start games
> > based on a specific seed was previously removed from Gnome Games.  It was
> > a rather a fiddly option and there were some calls in bugzilla for seeds
> > which actually matched the pattern of Microsoft Freecell as you have now
> > implemented.  If you can put a nice unobtrustive user interface on this
> > feature it should be good.  For what it is worth Pretty Good Solitaire
> > (Goodsol) sticks to the same pattern, as much as is possible.

> Ok, I'll look into how goodsol does it, and make it as
> unobtrusive/friendly as possible. I will start by re-adding seed support
> in the UI, that will be my first set of patches. The second set will add
> redmond-style game generation to Freecell (it will probably add a
> checkbox or something, too). How does that sound ?

I was suggesting you have one and only one seed type and that it would
match the numbers used by Microsoft as far as possible (which is what
goodsol does, but in their case they took it as far as possible and then
went further offering many more possible seeds than the old version of
Freecell provides).

> > Being able to start specific games with easily predicted out comes (like a
> > completely unshuffled deck) might serve as useful test cases to smoke test
> > Aisleriot.

> Does gnome-games currently have unit testing ?

Parts do, but the mix of C, Python, and Lisp means different approaches
are required.  I was thinking even in the simplest sense of having a list
of seed values for a few games which should be perfectly easy to solve
(not shuffled at all) would be helpful just to see if things are working
as expected.  More detailed unit tests would be great too if you are
interested but I wouldn't want to you to take it as anything other than a
suggestion.

-- 
Alan