There appears to be a glitch where the server thinks the program is still
logged in,  but still isn't communicating with it.

I am actually interested in building a new server around html5 and
specifically websockets and a much nicer web page and reporting system.
 It will keep its relatively simple format,  no major changes, just a much
nicer experience and less bugs.     The protocol would be GTP except for the
custom addition of login support so you could add that to your program or
just use the client.

The only problem is that websockets and html5 not quite ready or
standardized yet.

I would like to see if I can produce the buggy behavior at will.    The
latest theory is that this happens when a program logs in that is already
logged in, so I can at least take a look at that part of the code and/or see
if I can hang the server.     I am pretty limited on my time but I really
want to fix this so that I don't have to constantly monitor the server
(which usually means getting an email from someone telling me the server is
not working.)

I also wrote a server in C a long time ago (2 years?)  which I have not
deployed.  I don't remember the status but I was playing games on it and my
recollection was that I was just about ready to deploy it.    I
procrastinated because of the large amount of work involved - the first 2
weeks after deploy would be a major drain on my time and I would need to be
ready to respond to any problems.     But my feelings were that it was
better (even though written in C I felt better about the robustness of the
code.)     The current server is written in TCL.

You are probably busy too,  but if you want to try the login experiment I'll
be standing by to restart the server most of the day, not going out until
about 5:00 PM EST today.

Don




On Wed, Jul 6, 2011 at 7:48 AM, Rémi Coulom <[email protected]> wrote:

> Hi,
>
> This is an excerpt from my cgos log:
>
> 06:07:37    S->C genmove b 54749
> 06:07:37    C->E time_left b 54 0
> 06:07:37    E->C =
> 06:07:37    C->E genmove b
> 06:07:39    E->C = J2
> 06:07:39    C->S J2
> 06:07:39    S->C gameover 2011-07-06 W+Time
> 06:07:39    C->S ready
> 06:08:51    S->C info Maximum time until next round: 09:59
>
> Obviously there is a problem: Crazy Stone had 54 seconds left, played a
> move in 2 seconds, and lost on time.
>
> After this loss on time, it was never paired again.
>
> Rémi
> _______________________________________________
> Computer-go mailing list
> [email protected]
> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
>
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go

Reply via email to