Yes, the Windows priority setting affects. Setting the priority of J
from Normal to BelowNormal makes the J break work immediately.
But interesting observation is this: While J running infinite loop
Normal priority starting Notepad takes also long time, but only for
the first time, starting Notepad again is much faster. And same thing
seems to be with Jbreaker: Breaking the loop at the first time in J
session takes a long time, but some of the the following breaks work
fast but also slowly sometimes.
Closing and starting the J session again resets the situation and the
first launches both Notepad and J breaker take a long time. Same
result with J console and DOS console.
However if the time between launches is longer the delay is again
long. This must be an essential behaviour of the win2000 process
management. For my case I think this problem is solved now. I can
wait or lower the J priority if needed when working with WIN2000.
Thank you for the help.
Anssi
At 19:07 29.8.2006, you wrote:
It looks as if Jbreak is working properly (just very very slowly).
It seems as if a compute bound J makes your win2000 very slow to run
the jconsole.exe task to set the break signal.
I'm not aware of win2000 priority settings that would affect this or
that you can adjust. If you know of things you can adjust to affect
task priority you can try them.
Try a few things to get a handle on what the host scheduler is doing.
Run J exec loop.
How is it to start and run notepad?
How is it to start a DOS console window and do simple commands?
How is it to run jconsole i.5 etc?
If a Dos window runs ok then try the following to signal break to
the J exec loop:
...> cd \j601
...> jconsole -jprofile breaker.ijs
Does this work better than the yellow icon or is it just as sluggish?
Try editing the yellow icon to run j.exe instead of jconsole.exe as
the host rules for scheduling are different for console tasks. Does
this work any better?
----- Original Message ----- From: "Anssi Seppälä" <[EMAIL PROTECTED]>
To: "Beta forum" <[email protected]>
Sent: Tuesday, August 29, 2006 9:18 AM
Subject: Re: [Jbeta] J break hangs Windows for seconds
The result of the tests on Win 2000 are:
At 16:00 29.8.2006, you wrote:
I can't think of any reason for Jbreak to behave differently under
win200 than XP. Try Jbreak on your win2000 system with a few
simple tests. First verify that repeated running of Jbreak (with
nothing to break) run successfully. Then try each of the following
to verify Jbreak proper behaviour at least in simple cases.
J Break without anything to break runs immediately.
NB. double attention (break) required
delay=: 3 : 0
6!:3[20
)
J Break (double) stops this immediately
exec=: 3 : 0
while. 1 do. end.
)
J break hangs and the exec '' stops after 10 ... 40 seconds
output=: 3 : 0
q=.0
while. 1 do. smoutput q=.>:q end.
)
J break is pressed when the number is about 500 the process
interrupts at 18771 (after several tens of deconds). Seconf try made 18417.
input=: 3 : 0
while. 1 do. q=.1!:1[1 end.
)
Stops after the input after the J break.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
