[Idle-dev] [ idlefork-Patches-661363 ] Run multiple IDLEforks on the same machine simultanously

[SourceForge.net]

Patches item #661363, was opened at 2003-01-02 17:57
Message generated for change (Comment added) made by kbk
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=309579&aid=661363&group_id=9579

Category: None
Group: None
Status: Open
>Resolution: Accepted
>Priority: 4
Submitted By: Noam Raphael (noamr)
>Assigned to: Kurt B. Kaiser (kbk)
Summary: Run multiple IDLEforks on the same machine simultanously

Initial Comment:
  Instead of defining a specific port (8833), a list of
legitimate ports is defined. When IDLE starts, it goes
over the list and tries each of the ports. When it
finds a port which is ok, it starts the subprocess.
  The list of ports which I defined is very arbitrary,
and should be reviewed.
  A more minor change, which is calling the
spawn_subprocess method only after the RPCClient class
was succesfully initialized, solves the bug that when
opening a second IDLE, an error message is displayed
but a malfunctioning shell window is displayed, and
when it's closed, the subprocess keeps on running until
the first IDLE is shut down.

----------------------------------------------------------------------

>Comment By: Kurt B. Kaiser (kbk)
Date: 2003-05-16 22:34

Message:
Logged In: YES 
user_id=149084

I have implemented the second change: spawn after
successful port binding.

Having done this, I no longer see the spurious
processes being spawned on Linux.

However, on W2K I have at least once seen a second 
copy of IDLEfork start listening on the 8833 port used
by the original copy.  Then it went ahead and spawned
a subprocess in spite of this change. 

Until we figure that out (it's not supposed to happen, as 
far as I know) I don't want to create a bigger mess by 
allowing multiple IDLEforks to run.

----------------------------------------------------------------------

Comment By: Noam Raphael (noamr)
Date: 2003-04-12 18:01

Message:
Logged In: YES 
user_id=679426

This better patch tries the ports at a random order, which
is much faster when some IDLEfork instances are already running.

----------------------------------------------------------------------

Comment By: Kurt B. Kaiser (kbk)
Date: 2003-01-02 22:55

Message:
Logged In: YES 
user_id=149084

Converted Noam Raphael's file to a context patch



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=309579&aid=661363&group_id=9579