Progress on preventing multiple instances of a programfromrunning in windows

[Dar Scott]

On Dec 3, 2004, at 11:17 AM, Lynch, Jonathan wrote:

> What I am finding is that it is forcing the TCP listen without even
> using the accept command. The following script:
>
> Open socket to "127.0.0.1:48953"
> Write "Greetings" to socket "127.0.0.1:48953"
> Close socket "127.0.0.1:48953"
>
> Causes the instance that ran this script to start listening to port
> 48953. Shouldn't it only do that if the script actually uses the accept
> command?

Rev is helping you by not waiting around for the open.

I put a field and a button on a card with this script in the button:

on mouseUp
   put empty into field "field"
   open socket to "127.0.0.1:48953"
   put "open result: " & the result & lf after field "field"
   write "Greetings" to socket "127.0.0.1.48953"
   put "write result: " & the result & lf after field "field"
   put "openSockets(): " & lf & the openSockets & lf after field "field"
   put "netstat report: " & lf & shell("netstat -a -p TCP") & lf after 
field "field"
   close socket "127.0.0.1:48953"
   put "close result: " & the result & lf after field "field"
   put "netstat report: " & lf & shell("netstat -a -p TCP") & lf after 
field "field"
end mouseUp

This put this in the field (with the netstat reports cut out):

open result:
write result: socket is not open
openSockets():
127.0.0.1:48953
netstat report:

Active Connections

   Proto  Local Address          Foreign Address        State
   [no 48953]

close result:
netstat report:

Active Connections

   Proto  Local Address          Foreign Address        State
   [no 48953]


The open only starts the opening sequence.  The socket is put into open 
sockets tentatively.  The write should wait for timeout or success or a 
refusal.  The above example gets the connection refusal and there is an 
error in the write.  The openSockets() still includes the socket even 
after open fails, which I think is a bug.  The close socket gets no 
error, which I think is part of the same bug.

On my system I did not see any reference to 48953 in the netstat report.

You might want to add -n to the netstat command and stop it from 
flashing a window on Windows.

I ran the same test on OS X and it worked the same except it 
(correctly) got an error on the close.

Dar

****************************************
     Dar Scott Consulting
     http://www.swcp.com/dsc/
     Programming Services
****************************************