[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-smalltalk] Smalltalk MUD
From: |
kraehe |
Subject: |
Re: [Help-smalltalk] Smalltalk MUD |
Date: |
Sun, 29 May 2005 02:44:10 +0200 |
User-agent: |
Mutt/1.5.9i |
Moin Mike Anderson,
> I'm no guru, but it seems to me that you close the socket and it will
> never get re-opened. #waitForConnection just returns if the socket is
> closed. That means that your code is in an endless loop creating
> MUDClients that terminate immediately because the socket is closed.
this would result either in a "Connection closed by foreign host."
message, or prevent further telnet connection. Instead an existing
telnet connection hangs (does not read characters) till i kick
waitForConnection by opening an additional telnet.
> I've not done much with networking, so I can't offer a solution with
> certainty, but I notice that the NetClients NetServer/NetSession
> implementation doesn't close the socket, so perhaps it is as simple as that.
my current guess, is that the TCP namespace is an add on and not
an integral part of the gst kernel. Its therefore likely that the
process scheduler does not check the server socket, and it even
looks that a blocking read from console wont yield a select() on
all open file descriptors (including the server socket) to deceide
for the next co-process to run. Thats the only explaination I have
why i need to type something into transcript/console to kick the
the co-process to start working. I fear that NetClients NetServer
only works, because a httpd does'nt need much concurency.
BTW: if co-processses and sockets are working together: would
the limit be 250 users (255-stdio-logfile-savefile) - or does
gst cascade select() calls to support more than 250 co-processes
waiting on more than 255 open files - or does gst use real posix
threads to implement co-processes (so it does'nt need to select
the io in process scheduler, because a pthread can just block) ?
Bye Michael
--
mailto:address@hidden UNA:+.? 'CED+2+:::Linux:2.4.29'UNZ+1'
http://www.xml-edifact.org/ CETERUM CENSEO WINDOWS ESSE DELENDAM