Author: Sune Fischer
Date: 04:58:27 04/14/04
Go up one level in this thread
On April 14, 2004 at 07:50:57, Fabien Letouzey wrote:
>
>>>It's not supposed to hang if the pipe is broken (it should report an error, of
>>>course).
>>>It could do if the pipe was alive, but full.
>>>I don't know what's the default buffer size for pipes, 4K or something?
>>>I don't think that's what happens in your case.
>
>>That could be exactly what happens, 4k is nothing in multipv mode.
>
>The OS is supposed to close that pipe if Arena crashes.
>Then printf() shouldn't hang anymore.
>But, once more, I don't know the behaviour on Windows.
It is really odd.
I'm doing a check like this:
fprintf(log,"printing now to arena\n");
fflush(log);
printf("%s\n",pv);
fflush(stdout);
fprintf(log,"printing was successful\n");
fflush(log);
and after the crash the last line in the log
is "printing now to arena\n".
I can only conclude that printf hangs in a bad way.
>>'ctrl-Z' does nothing unless I also press return, then the engine is killed.
>
>Could you please try it with other engines (Fruit, Yace)?
>Just ctrl-Z.
Same behavior with Yace and Fruit, nothing happens until I press return.
-S.
>Fabien.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.