Hi all!
This is an E-mail I sent as a response to Larry Hastings. I'm posting it here
because I think it may be of interest to the general public.
Regards,
Shlomi Fish
---------- Forwarded Message ----------
Subject: Re: Why not just ship Freecell Solver 2.10?
Date: Wednesday 12 January 2005 18:15
From: Shlomi Fish <shlomif_at_iglu.org.il>
To: Larry Hastings <larry_at_hastings.org>
On Wednesday 12 January 2005 15:56, you wrote:
> Freecell Solver 2.9.9 has been in "beta" since about March of 2003.
Right.
> I
> dimly remember it being a speed improvement over the previous version
> I'd been using, 2.8.6.
That's quite unlikely. I did not introduce any speed optimizations in the
2.9.x tree. What I did was introduce more features in an attempt to support
Patsolve's board and move prioritazation logic. I finished implementing what
I believed would suffice for it, in a not entirely identical manner, but the
performance was not as good as Patsolve, so I wanted to go for a better
emulation of it. That was not implemented yet.
Once Freecell Solver could duplicate Patsolve's moves exactly, there should
be a significant speed increase in solving using only Atomic Moves. It would
also be possible to construct meta-scans that will consist of several
optimized atomic moves scan to possibly even surpass Patsolve.
But none of this is finished yet.
> Why don't you just declare it done, bump the
> version number to 2.10, and ship it?
Because the current snapshot of the trunk in the CVS isn't done yet. There
are possibly bugs lurking, and I know that I am constantly annoyed by a lot
of debugging print statements, which I've inserted into it. So the code
needs to be cleaned up a lot. Add to that that the entire purpose of the
branch was to provide Patsolve-like scans, and you'll understand that
someone has to complete the job.
I should note that somewhere in the middle of the 2.9.x branch, I seem to
have lost interest in maintaining Freecell Solver, and have diverted my time
to other avenues. While there's still some tasks that an eager developer can
perform, I feel that Freecell Solver 2.8.x is good enough as it is for most
of the needs to which it is intended. If someone wishes to further improve
it, I'll gladly make him a project co-maintainer at berlios, once he proves
that he is eager enough to improve it. (by sending patches).
Regards,
Shlomi Fish
---------------------------------------------------------------------
Shlomi Fish shlomif_at_iglu.org.il
Homepage:
http://www.shlomifish.org/
Knuth is not God! It took him two days to build the Roman Empire.
-------------------------------------------------------
--
---------------------------------------------------------------------
Shlomi Fish shlomif_at_iglu.org.il
Homepage: http://www.shlomifish.org/
Knuth is not God! It took him two days to build the Roman Empire.
Received on Thu Jan 13 2005 - 10:27:19 IST