Hi,
Does anyone know what kind of errors may cause PD to become completely unresponsive (i.e.: you click on any GUI object and nothing happens, ctrl+e does not change cursor, clicking on any object doesn't open it, pd's window won't close, etc) but without consuming any CPU at all (0%)???
It's something triggered by something in my patch, but as usual I am not able to isolate it, as all tests with small quantities of things always worked and never had any error, and just when things get bigger (only a matter of quantity: more instances of the very same things) something weird happens....
Windows XP here.
What if a message tree becomes HUGE not in depth (which would give a stack overflow) but in "breadth"? A part from the time needed to execute it (if that was the problem, pd would hang consuming a lot of cpu), is it possible that it gets just too big and everything breaks and stop working without triggering a stack overflow or whatever error? Or must it be something else?
Thanks, m.
Symptoms are somewhat similar to an unstopped [until], except that:
- an unstopped until eats up almost 100% cpu (50% on a dual core),
while I'm experiencing 0% cpu consumption
- a while, an unstopped until is automatically stopped or something
(is it the watchdog?) and pd becomes responsive again, while I am not experiencing this.
2008/2/11, matteo sisti sette matteosistisette@gmail.com:
Hi,
Does anyone know what kind of errors may cause PD to become completely unresponsive (i.e.: you click on any GUI object and nothing happens, ctrl+e does not change cursor, clicking on any object doesn't open it, pd's window won't close, etc) but without consuming any CPU at all (0%)???
It's something triggered by something in my patch, but as usual I am not able to isolate it, as all tests with small quantities of things always worked and never had any error, and just when things get bigger (only a matter of quantity: more instances of the very same things) something weird happens....
Windows XP here.
What if a message tree becomes HUGE not in depth (which would give a stack overflow) but in "breadth"? A part from the time needed to execute it (if that was the problem, pd would hang consuming a lot of cpu), is it possible that it gets just too big and everything breaks and stop working without triggering a stack overflow or whatever error? Or must it be something else?
Thanks, m.
-- Matteo Sisti Sette matteosistisette@gmail.com http://www.matteosistisette.com
Some versions of Pd behave like this if you try to print certain characters like '{'.
matteo sisti sette wrote:
Symptoms are somewhat similar to an unstopped [until], except that:
- an unstopped until eats up almost 100% cpu (50% on a dual core),
while I'm experiencing 0% cpu consumption
- a while, an unstopped until is automatically stopped or something
(is it the watchdog?) and pd becomes responsive again, while I am not experiencing this.
2008/2/11, matteo sisti sette matteosistisette@gmail.com:
Hi,
Does anyone know what kind of errors may cause PD to become completely unresponsive (i.e.: you click on any GUI object and nothing happens, ctrl+e does not change cursor, clicking on any object doesn't open it, pd's window won't close, etc) but without consuming any CPU at all (0%)???
It's something triggered by something in my patch, but as usual I am not able to isolate it, as all tests with small quantities of things always worked and never had any error, and just when things get bigger (only a matter of quantity: more instances of the very same things) something weird happens....
Windows XP here.
What if a message tree becomes HUGE not in depth (which would give a stack overflow) but in "breadth"? A part from the time needed to execute it (if that was the problem, pd would hang consuming a lot of cpu), is it possible that it gets just too big and everything breaks and stop working without triggering a stack overflow or whatever error? Or must it be something else?
Thanks, m.
-- Matteo Sisti Sette matteosistisette@gmail.com http://www.matteosistisette.com
I could finally isolate it.
It is a bug in netsend (again!), not print.
If you send too many message to netsend in 0 logical time, pd freezes and stops responding, but without eating up CPU.
Here's a simple patch that reproduces it. I guess the critical quantity of messages may be platform dependent (or not), so if it does not reproduce the bug try increasing the number in the message box.
2008/2/11, matteo sisti sette matteosistisette@gmail.com:
Hi,
Does anyone know what kind of errors may cause PD to become completely unresponsive (i.e.: you click on any GUI object and nothing happens, ctrl+e does not change cursor, clicking on any object doesn't open it, pd's window won't close, etc) but without consuming any CPU at all (0%)???
It's something triggered by something in my patch, but as usual I am not able to isolate it, as all tests with small quantities of things always worked and never had any error, and just when things get bigger (only a matter of quantity: more instances of the very same things) something weird happens....
Windows XP here.
What if a message tree becomes HUGE not in depth (which would give a stack overflow) but in "breadth"? A part from the time needed to execute it (if that was the problem, pd would hang consuming a lot of cpu), is it possible that it gets just too big and everything breaks and stop working without triggering a stack overflow or whatever error? Or must it be something else?
Thanks, m.
-- Matteo Sisti Sette matteosistisette@gmail.com http://www.matteosistisette.com
-
Claude Heiland-Allen -
matteo sisti sette