On Fri, 28 Jul 2000 11:46:20 -0700 Tim Rowledge tim@sumeru.stanford.edu wrote:
So the 'waste' time is high with change sorters but not browsers? Now that's pretty wierd. Does anyone have any idea what change sorters might be doing that is so different to browsers?
They do a little more work on #activate. You might also check to see roughly how often ChangeSorter>>update is being called under these circumstances.
Cheers, Bob
On Fri, 28 Jul 2000 11:46:20 -0700 Tim Rowledge tim@sumeru.stanford.edu wrote:
So the 'waste' time is high with change sorters but not browsers? Now that's pretty wierd. Does anyone have any idea what change sorters might be doing that is so different to browsers?
Bob Arning arning@charm.net relied...
They do a little more work on #activate. You might also check to see roughly how often ChangeSorter>>update is being called under these circumstances.
It should be informative to run the MessageTally for a while (screen menu / debug... / start MessageTally) and then move the mouse to the top of the screen to stop it. Changing the setting of the 'smartUpdating' preference may have an effect.
- Dan
I just tried using the MessaageTally idea that Dan mentioned nad was toldthat with either a browser or a changesorter open that the system spends 97% of it's time in Delay>wait.
Whoever it was that reported the original problem will need to give me some more information about what windows were open, what was done, how the 'waste' time was worked out etc.
tim
Tim,
Tim Rowledge wrote:
I just tried using the MessaageTally idea that Dan mentioned nad was toldthat with either a browser or a changesorter open that the system spends 97% of it's time in Delay>wait.
Whoever it was that reported the original problem will need to give me some more information about what windows were open, what was done, how the 'waste' time was worked out etc.
tim
-- Tim Rowledge, tim@sumeru.stanford.edu, http://sumeru.stanford.edu/tim Strange OpCodes: RPM: Read Programmer's Mind
Morphic! I have just checked that in MVC there is not such a problem.
OK, some more to the already posted infos below:
- Squeak 2.9alpha #2447, - Squeak started with flags (-nojit should be superfluous): squeak -memory 64m -xshm -nojit - Morphic project, - switching off browseWithDrag&Drop and smartUpdate doesn't help; - with EventImageStuff and TPR-EventFix1 changesets filed in; - VM compiled with last version from SourceForge at Wed Aug 2 00:54:32 MEST 2000, - SuSE 6.2 Linux, - sr@Klaus:~/Squeak/work > gcc -v Reading specs from /usr/lib/gcc-lib/i486-linux/egcs-2.91.66/specs gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release), - XFree with fvwm2 window manager and multiple windows, -> Squeak fills one of them.
Just close all windows in the current project and then open a simple change sorter by the menu. This leads to a significant processor load at my machine, with or without the mouse inside the Squeak window. The percentages are measured by 'top' and ca. ones, of course.
Please ask what you want to know, if this is not sufficient yet.
Greetings,
Stephan
Typical output of top for one simple change sorter open:
12:57am up 1:15, 7 users, load average: 0.57, 0.60, 0.46 67 processes: 65 sleeping, 2 running, 0 zombie, 0 stopped CPU states: 34.3% user, 4.1% system, 0.0% nice, 62.1% idle Mem: 159496K av, 155420K used, 4076K free, 66380K shrd, 28824K buff Swap: 136040K av, 388K used, 135652K free 55320K cached
PID USER PRI NI SIZE RSS SHARE STAT LIB %CPU %MEM TIME COMMAND 2773 sr 15 0 21008 20M 3964 R 0 30.2 13.1 3:39 squeak 1170 sr 1 0 4356 4356 3292 S 0 2.3 2.7 0:18 LinCVS_0.2.5 1143 sr 1 0 1224 1224 1004 S 0 1.7 0.7 0:26 xosview.bin 1159 sr 0 0 792 792 612 S 0 1.1 0.4 0:12 top 2782 root 2 0 788 788 612 R 0 1.1 0.4 0:00 top 232 root 0 0 24256 23M 4552 S 0 0.9 15.2 1:45 X 1018 sr 0 0 18768 18M 9368 S 0 0.5 11.7 0:20 netscape 235 sr 0 0 1476 1476 980 S 0 0.1 0.9 0:03 fvwm2 ...
If top shows below 5% without beeing in the Squeak window, that would be OK for me (MVC needs less inside, if you don't move the mouse).
Stephan Rudlof wrote:
This bug report applies, if using event mode (with an event VM).
The processor time used by Squeak is about 4%, if the windows are closed and there is no activity.
The processor time increases from this 4% to about
34%, if a simple ChangeSorter window, 50%, if a dual ChangeSorter window, 57%, if both,
has/have been opened.
This waste of processor time isn't the case for opening a SystemBrowser window.
These problems occure with or without applying the TPR-EventFix1 changeset.
Stephan
P.S.: I'm not actively seeking for this kind of bugs, but my CPU cooler starts in my notebook, if there is significant processor activity over some time...
-- Stephan Rudlof (sr@evolgo.de) "Genius doesn't work on an assembly line basis. You can't simply say, 'Today I will be brilliant.'" -- Kirk, "The Ultimate Computer", stardate 4731.3
I took a brief look at this problem and found that on my machine 'top' claimed squeak was taking 92+% of cpu ! Until I built a VM with stephan(?)'s fix for the no-sockets-loaded-yet-sleep, when the cpu usage dropped to <1% no matter how many browsers and changesorters I opened.
tim
Maybe someone else can try this. If you take a current 2.8 image and run it without the source file around you get the ok message up stating it can't find the source file. However you can't click on the ok, or use the keyboard. You must file quit Squeak. Also It seems the ok message is clipped to a smaller rectangle.
Now I think it's related to the change that was made to auto-size the screen area, but I'm not sure since when we get to the failure point then you can't do any debugging.
So just in case I've got a bad image, or two. Could someone give this a whirl with a 2.8 image with all the updates and let me know if its for real?
-- =========================================================================== John M. McIntosh johnmci@smalltalkconsulting.com 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com =========================================================================== Custom Macintosh programming & various Smalltalk dialects PGP Key: DSS/Diff/46FC3BE6 Fingerprint=B22F 7D67 92B7 5D52 72D7 E94A EE69 2D21 46FC 3BE6 ===========================================================================
I just tried this with a fresh Squeak2.8-2348 image on Windows (2000) with the current May 23 VM. It seems to work just fine... I can hit OK when the message comes up and continue along.
I noticed a couple of problems with the message itself, though. It says "Further explanation can found in the startup window, 'How Squeak Finds Source Code'." The word "be" seems to be missing here, but also, I don't see any further explanation in either the Welcome To or Getting Started windows. Some cleanup is probably in order.
- Doug Way dway@mat.net, @riskmetrics.com RiskMetrics Group, Ann Arbor, MI http://www.riskmetrics.com
John M McIntosh wrote:
Maybe someone else can try this. If you take a current 2.8 image and run it without the source file around you get the ok message up stating it can't find the source file. However you can't click on the ok, or use the keyboard. You must file quit Squeak. Also It seems the ok message is clipped to a smaller rectangle.
Now I think it's related to the change that was made to auto-size the screen area, but I'm not sure since when we get to the failure point then you can't do any debugging.
So just in case I've got a bad image, or two. Could someone give this a whirl with a 2.8 image with all the updates and let me know if its for real?
--
John M. McIntosh johnmci@smalltalkconsulting.com 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com =========================================================================== Custom Macintosh programming & various Smalltalk dialects PGP Key: DSS/Diff/46FC3BE6 Fingerprint=B22F 7D67 92B7 5D52 72D7 E94A EE69 2D21 46FC 3BE6 ===========================================================================
John M McIntosh wrote:
So just in case I've got a bad image, or two. Could someone give this a whirl with a 2.8 image with all the updates and let me know if its for real?
Tried it on the latest Mac VM (home-compiled but standard latest sources) and it worked as it should. I could click OK and go on working.
Henrik
squeak-dev@lists.squeakfoundation.org
-
Bob Arning -
Dan Ingalls -
Doug Way -
Henrik Gedenryd -
John M McIntosh -
Stephan Rudlof -
Tim Rowledge