InputEventFetcher starving all other processes

Hi:

I am currently trying to get Pharo running smoothly on our RoarVM.
However, there seem to be some changes in Pharo which lead to the image not making any progress.

For my experiments, I am using the latest 1.2 core image.

Do you know about any changes in Pharo, compared to Squeak 4.1, with respect to input event retrieval and perhaps the event processing?

Thanks
Stefan

--
Stefan Marr
Software Languages Lab
Vrije Universiteit Brussel
Pleinlaan 2 / B-1050 Brussels / Belgium
http://soft.vub.ac.be/~smarr
Phone: +32 2 629 2974
Fax:   +32 2 629 3525

Yes
mike rueger changes the poll behavior of the sensor.
The idea is to get rid of polling.

Stef

On Oct 31, 2010, at 5:06 PM, Stefan Marr wrote:

Hi:

On 31 Oct 2010, at 17:32, Stéphane Ducasse wrote:

> mike rueger changes the poll behavior of the sensor.
> The idea is to get rid of polling.
Well, I have not identified the source of the problem yet.

But the symptoms are that InputEventFetcher>>primGetNextEvent: delivers 500-1000x more events than EventSensor>>primGetNextEvent: in MVC or Squeak 4.1 images deliver.

Could there be something else in PharoCore which interacts with the VM in unexpected ways?
In case that is relevant, the RoarVM uses X11 for the interface. (-headless has the same issues)

Thanks
Stefan

--
Stefan Marr
Software Languages Lab
Vrije Universiteit Brussel
Pleinlaan 2 / B-1050 Brussels / Belgium
http://soft.vub.ac.be/~smarr
Phone: +32 2 629 2974
Fax:   +32 2 629 3525

Hi:

On 31 Oct 2010, at 19:15, Stefan Marr wrote:

> But the symptoms are that InputEventFetcher>>primGetNextEvent: delivers 500-1000x more events than EventSensor>>primGetNextEvent: in MVC or Squeak 4.1 images deliver.
Turns out that the array which is supposed to be filled in by '>>primGetNextEvent: array' does not get filled in correctly.
For some reason the old values stay in the array when there isn't any new event.

However, the very same code works for the non-Pharo images.

Was there any change to the code in or related to this primitive (<primitive: 94>) lately?

Thanks
Stefan


--
Stefan Marr
Software Languages Lab
Vrije Universiteit Brussel
Pleinlaan 2 / B-1050 Brussels / Belgium
http://soft.vub.ac.be/~smarr
Phone: +32 2 629 2974
Fax:   +32 2 629 3525

what is strange is that pharo does not change the vm code.
One which OS are you?

> Hi:
>
> On 31 Oct 2010, at 19:15, Stefan Marr wrote:
>
>> But the symptoms are that InputEventFetcher>>primGetNextEvent: delivers 500-1000x more events than EventSensor>>primGetNextEvent: in MVC or Squeak 4.1 images deliver.
> Turns out that the array which is supposed to be filled in by '>>primGetNextEvent: array' does not get filled in correctly.
> For some reason the old values stay in the array when there isn't any new event.

may be a semaphore or something like that prevents some actions to happen.