Re: [Bug-gnubg] gnubg-multi on Windows

[Holger]

At 19:36 29.07.2003 +0200, Olivier Baur wrote:

> From the output log you provided, it seems that the accept() call in 
> Slave() doesn't succeed. I've fixed a bug on line 3609 that prevented 
> gnubg from displaying the error in that case.

Yes, I think there's something odd for Windows in Slave(). I'm not sure 
about the condition of the do loop and the if-statement before.

> I've just committed the changes; can you try to compile and run it, please?

(No game) pu slave
Host set to SLAVE mode.
  Hit ^C to kill pending tasks and return to MASTER mode.

Available processing units:
  Id  Type    Status     Queue  Tasks                 Address
   1  Local   Ready          1  rollout analysis      n/a

Tasks (todo/prog/done)  Rollouts (rcvd/sent/fail)  Evals 
(rcvd/sent/fail)  Statu
s
   0 (   0/   0/   0)     0 (   0/   0/   0)     0 (   0/   0/   0)  N/A
   0 (   0/   0/   0)     0 (   0/   0/   0)     0 (   0/   0/   0)  N/A
   0 (   0/   0/   0)     0 (   0/   0/   0)     0 
(   0/   0/   0)  Waiting for
*** RPU refused connection on local IP address 255.255.255.255 from remote 
IP ad
dress 255.255.255.255.

   0 (   0/   0/   0)     0 (   0/   0/   0)     0 
(   0/   0/   0)  Waiting for
*** RPU refused connection on local IP address 255.255.255.255 from remote 
IP ad
dress 255.255.255.255.

> I also see that you have commented out many calls to 
> gdk_threads_enter()/gdk_threads_leave() in real_main(); now my gnubg-multi 
> freezes at startup :-) These calls are necessary once the GDK lock is 
> inited; why did you comment them out?

I've put the gdk_threads_enter()/gdk_threads_leave() pairs in the 
appropriate GTK+ output functions, such as PushSplash, GTKOutputErr and so 
on, thus they aren't necessary in e.g. real_main anymore. Make sure you 
update all files from the CVS.
And obviously I forgot to remove those lines. If your problems go away I'll 
remove them.

Regards,

        Holger