In rec.games.chess.computer help bot wrote:
On Sep 11, 10:59 am, "Thomas T. Veldhouse" wrote:

Perhaps your team should get a chess player onboard

Maybe Sanny thinks *he is* a chess player?


and test it before you
release software. Better yet, build unit tests to test various functions in
the engine so that you KNOW before you release that your software at least
does the things you expect it to do under the test conditions. You will get a
lot less heat from the people in this group if you put a little QC around your
product.

Functions? Like say, a modular program in which
a move-generator passes a list to the next function,
which screens for legality and then passes off the
revised list (legal moves only) to an analysis function,
and on down the line? I don't think that is how
Sanny's program works.

My guess is that Sanny's program has a bunch
of patched-together code which incorporates a
multitude of Band-Aid style "fixes", and spits out a
move (legal or not) only after going through a series
of contortions and painful exercises, many of which
would "kill" any normal program the same way a
lemming is killed when he jumps off a cliff into the
sea. For Sanny, "testing" is where he goes back
after the game and meekly asks where the program
went wrong, so he can then type in another "fix" in
which the program will play some improvement by
rote from then on whenever it reaches the same
position, in exactly the same move order.

In short, the way to help him is by mailing him
cases of Band-Aids, glue, rubber bands and best
of all, duck tape.


Agreed, it sounds like that. Never-the-less, a released program should have a
QA process and a QA process usually follows a unit testing process [or
incorporates one in batch].

If he doesn't take the suggestion, then perhaps he won't be pestering this
group much longer, but if he does, perhaps he will put out a better product.

--
Thomas T. Veldhouse

We have more to fear from the bungling of the incompetent than from the
machinations of the wicked.