[Idle-dev] idle 1.0.2 cannot bind (again) to port 8833 w/python 2.3.3

[Al Schapira]

On Thu, 2004-02-26 at 17:17, Neal Norwitz wrote:

> I'll try to describe in a general way that isn't exact:
> 
>         This is your windowing (graphical) system (GNOME or Bonobo).
>         It allows GNOME applications to talk a common language (CORBA).
> 
>         Strickly speaking it isn't necessary, unless/until some
>         applications need to communicate, e.g., when you make
>         a configuration change to the window manager.

Thanks.  This explanation does help.

> 
> > Does killing oafd shed any light on what the problem with idle really
> > is?
> 
> You can restart the process if you wish.  IIRC, it's started when
> necessary.  The problem doesn't haven anything to do with IDLE
> specifically.  When you started X (perhaps by logging in),
> GNOME started up, which started oafd and it took port 8833.
> Unfortunately, IDLE is hard coded (IIRC) to use the same port.

Shall I file a bug report (enhancement request) for idle to use another
port if 8833 is busy?

> So there is a conflict since each program is trying to use
> the same port.  Whichever starts first wins.  I think oafd
> can use any port.  

This would explain things except for fact that idle was working fine in
subprocess mode for many invocations before it failed.  Perhaps oafd
started up in between two idle runs.
> 
> Just in case, you can verify that 8833 is not in /etc/services.
> If it is in there, change the value to another number.  The
> new number must be unique in /etc/services (at least for all
> the TCP services).

"grep -rs 8833 /etc" yielded no relevant matches.  Where else should I
look? I do not have an /etc/services file or directory. (RH 9)

> 
> HTH,
> Neal

Thanks again.
	-Al