[jawsscripts] Re: Pause versus Delay
- From: "Jim Snowbarger" <Snowman@xxxxxxxxxxxxxxxx>
- To: <jawsscripts@xxxxxxxxxxxxx>
- Date: Thu, 3 May 2012 21:38:59 -0500
Gosh, There is a lot not really understood here, about the real time aspects
of this.
I dont' really know how windows task scheduling works. AT one time, it ,
maybe clear back to 3.1, was, I thought, cooperative. Once you got control
of the cpu, your process kept it until it decided to relinquish it. So, a
process could hog the CPU, kind of like how some people hog the conversation
space? But, since they couldn't count on developers implementing their code
with that in mind, I thought they went to more of a round robbin method,
where processes that want time get multiplexed at a 1 millisecond rate or
something. If that's the case, it may be that this distinction no longer
is relevant. When one says, put the thread to sleep, that would imply, I
suppose, that other processes and threads can still run.
I know stuff keeps happening, even when a delay is active.
For example, when I want to change focus, and then wait until it actually
occurs, I don't want to wait any longer than necessary. So, instead of just
waiting an arbitrary amound of time, sure to be plenty long enough for the
most sluggish environment, I might take the more complicated route, and do
something like this.
globals
int focusChangedFlag
... snip...
Have FocusChangedEvent set the global FocusChangedFlag to 1
; incidentally, I usually prefix, or postFix, my globals with some
app-related letters to try to stay out of other people's name space, because
globals are shared between all scripts, for all applications..
... snip...
; focus in my window
FocusChangedFlag = 0
SetFocus (myHandle)
for safety = 0 to 50
Delay(1) ; or this can be a pause, either way
if (FocusChangedFlag) then
; it happened
Safety = 99 ;; escapes the loop
endif
EndFor
; chekc FocusChangedFlag and see what happened...
While all these delay statements are being executed, the application is,
usually, responding to the focus changed, and the JAWS FocusChangedEvent
rungs, confirming that the OS gave control to the app during my loop, and
that jaws events are truly asynchronous with script execution. So, I'm
speculating that, other than the affect of cancelling delays due to key
presses, pause and delay(1) are essentially equivalent,
Concerning my comment about how delays are aborted, if you press a key, the
delay calls return almost immediately, even faster in jaws 13, though I have
not tick counted it. But, your loop will run to completion, which means
that, if it was doing other things as well, it just did them all many times,
real fast.
I got into this incidentally, because I am having problems with an app that
has worked wonderfully until I installed jaws 13. Now, it misbehaves on all
earlier versions of jaws as well.
The symptom seems to be that you can do a SetFocus in a particular
circumstance, and then focus ChangedEvent will run, and GetFocus confirms
that focus was placed in that window, but my tools show that the window is
not visible, but is obscured by the window that had focus before this little
exercise. And, insert+t says, lost focus.
I have never seen that before. GetFocus says that a window which is not
visible has focus, even though it certainly does not receive keyboard input.
Is that normal?
Even if I take the object route, sellflag_takeFocus, and all that, the same
thing happens.
The Control does not come into the foreground.
Any ideas?
AT this point, I really can't prove it's a jaws fault. I know I would need
to uninstall 13 including shared components, and reinstall 10. But, I'm not
quite that ambitious. <grin>
Guess I don't want to know badly enough yet.
Depends on whether a work around can be established. Work arounds are my
life.
----- Original Message -----
From: "Andrew Hart" <ahart@xxxxxxxxxxxxx>
To: <jawsscripts@xxxxxxxxxxxxx>
Sent: Thursday, May 03, 2012 9:19 AM
Subject: [jawsscripts] Re: Pause versus Delay
I can't speak for Delay, but my understanding is that Pause in JAWS
script is more or less the same as DoEvents in visual Basic, that is,
JAWS relinquishes control back to the OS. The OS stores the JAWS
process context and gives other processes waiting in the queue and
demanding cycles some CPU time to run. Now, there should be some
minimum amount of time betweenJAWS relinquishing control and it
regaining it again. The point is to give the OS and other
processes/applications time to send, receive and process messages.
After this, JAWS can respond by updating the OSM and continuing to run
scripts, etc. I imagine Jim is right about the 100 ms minimum, but that
is probably an FS-imposed constraint. DoEvents in Visual Basic has no
such constraint because I used to use it in the main loop of game code
so that JAWS could be used to monitor on-screen activity, and the game
code was looping 50 times per second. Thus, each loop cycle had 20 ms
to run (of which I was using about 8-12 ms) and DoEvents was only
consuming a maximum of 2 ms (and less than 1 ms most of the time). This
was on a slower machine than what I'm running today.
However, it is interesting that a keypress kills all Delay calls in a
script. I was never aware of that. I often use code like
SpeechOff()
Delay(1)
Pause()
SpeechOn()
To shut JAWS up and prevent it from speaking extraneously when returning
from menus or dialogs that scripts have opened. The Delay is
essential, for without it, the messages end up getting processed after
Pause and hence everything gets spoken. The delay seems to allow Pause
to time to process the messages while Speech is off and before it gets
activated again, but I never truly grokked the relationship between
Delay and Pause, except that I didn't think Pause was necessarily for a
fixed time.
Cheers,
Andrew.
On 3/05/2012 8:59 AM, Doug Lee wrote:
> Interesting concept, pause() waiting for a specific number of
> messages. I do know it seems to wait for a fairly consistent amount of
> time though, so maybe it's an either-or relationship between message
> count and millisecond count?
>
> On Thu, May 03, 2012 at 08:53:24AM -0400, Jamal Mazrui wrote:
> I forget where I got this info, but have the impression that Delay puts
> the current thread to sleep for a prescribed time, whereas Pause waits
> for a certain number of Window messages to be processed before
> continuing. Perhaps that number is 20, and perhaps those are messages
> passed to the active window -- not sure.
>
> Jamal
>
> On 5/2/2012 10:26 PM, Jim Snowbarger wrote:
>> Have any of you guys ever sussed out the subtle differences between pause
>> and delay? I know what FSDN says about it, but that description doesn't
>> seem particularly enlightening or precise to me.
>> What does it do behind the scenes? Does jaws get suspended allowing
>> other programs to run for both? Or, just for the pause. From the
>> description, delay doesn't do that. So, when you are delaying, what
>> happens to the rest of your running applications.
>> I get the impression that they are blocked.
>> I have always modeled this jaws scrip machine in my head as a single
>> execution thread, with a single priority interrupt. The main thread
>> runs your scripts and the functions they call. the interrupts represent
>> events.
>> I know you can run a loop containing a pause statement, waiting for an
>> event to fire.
>> I don't know if delay allows that, or it will forestall the event.
>> Anyway, do you think that is an accurate model?
>>
>> I am poking into this because of some serious problems I am having with
>> some custom scripts that work fine under jaws 5, and jaws 10, but have
>> issues, as they say, with jaws 13. More about that perhaps as I learn
>> it.
>> But, in general, I am finding that lots of things don't work quite the
>> same in jaws 13, and I'm having to chase down the anomalies. Great sport,
>> isn't it? Hunting down the latest batch of jaws quirks? Keep your old
>> versions handy!
>>
>> But anyway, one thing I know about these two builtins is that they behave
>> differently if you press a key.
>>
>> For example, consider the following fragment as part of a script
>>
>> SayString ("start")
>> Let i = 20
>>
>> while (i> 0)
>>
>> ; delay(5)
>>
>> Pause()
>>
>> Let i = i-1
>>
>> EndWhile
>>
>> SayString ("stop")
>>
>>
>> Notice that delay is commented out, and pause is not. You might want to
>> flip that around when experimenting.
>>
>> This total loop takes 2 seconds, no matter which builtin you use,
>> implying that Pause is also a 100 millisecond delay, which is what I
>> always believed.
>>
>> But, one subtle difference is that delay is canceled by keyboard input.
>> Pause is not.
>> With delay, once you press a key, all calls to the delay builtin are
>> canceled until the running script terminates.
>> This means that, if your script really needs to do the delay before it
>> takes some other action, that action is going to be done prematurely if a
>> key is pressed.
>> This is true in jaws 10, my favorite, and jaws 13, and probably most if
>> not all others.
>>
>> So what do you experts know about pause versus delay?
>>
>>
>>
>>
>> __________???
>>
>> View the list's information and change your settings at
>> //www.freelists.org/list/jawsscripts
>>
> __________???
>
> View the list's information and change your settings at
> //www.freelists.org/list/jawsscripts
>
__________�
View the list's information and change your settings at
//www.freelists.org/list/jawsscripts
__________�
View the list's information and change your settings at
//www.freelists.org/list/jawsscripts
Other related posts:
