hi all
Some updates for 3.9alpha to warn up again. :)
6601HexFixes-laza -- Boris Gaertner -- 16 March 2005 with changes set #6217 the method Integer>>hex was deprecated. This change set is an attempt to replace the remaining senders of hex with either storeStringHex or printStringHex"
6602CleaningUtilitiesAgain-sd -- stephane ducasse -- 30 March 2005 slowly we will get there..."
6603CleaningUtilities-sd -- stephane ducasse -- 30 March 2005 Move some utilities close to the class they use, i.e. Scanner"
6604-prefers-sd Disable the preference for horizontal scrollbars that made everything so sluggish in addition to the rest :)
6605-FixeScrollbars-sd -- stéphane ducasse -- 1 April 2005 fixing old instance of scrollbars now there is no popped up debugger anymore. '
Now we can start to work with 3.9.
Stef
On [DATE], "[NAME]" <[ADDRESS]> wrote:
6604-prefers-sd Disable the preference for horizontal scrollbars that made everything so sluggish in addition to the rest :)
Are you reallu sure ?
I founding several things in Preferences , what when finish collecting and test submit to Mantis, but IMHO Preferences should be a user taste thing.
Edgar
Propose others. I could not work with the horizontal scrollbar on. This is why I disable them. And 3.9 should not be sluggish. because we should not trade beauty for a system that nobody uses because it is too slow.
Stef
On 1 avr. 05, at 23:39, Lic. Edgar J. De Cleene wrote:
On [DATE], "[NAME]" <[ADDRESS]> wrote:
6604-prefers-sd Disable the preference for horizontal scrollbars that made everything so sluggish in addition to the rest :)
Are you reallu sure ?
I founding several things in Preferences , what when finish collecting and test submit to Mantis, but IMHO Preferences should be a user taste thing.
Edgar
I like the functionality of the horizontal scrollbar in the list panes, but I have to admit that it does cause a significant performance hit.
I thought most of the slowdown in 3.9alpha (when opening browsers etc) was due to the antialiased fonts, but I guess the horizontal scrollbars are a bigger factor. Maybe with some MessageTally investigation there's a way to speed those up.
Anyway, I tried fiddling with various preferences while running the following benchmark:
Time millisecondsToRun: [10 timesRepeat: [Browser fullOnClass: Browser selector: openBrowser]]
To test, start with a fresh image, open one workspace. (Also open one browser first to select green "#Browser" type and then close the browser.) Do a print-it on the benchmark, and then do it again twice, leaving the extra windows open. (How many windows are open has a huge effect on performance.) Tested on a Mac G5 - 1.8GHz.
Here are the results I saw. ("1-10" means opening the first 1 to 10 browsers, "11-20" is the next 10, etc)
fresh Squeak3.9alpha-6599 (alwaysHideHScrollbar is off at this update level): 1-10: 12994ms 11-20: 18554ms 21-30: 23814ms
fresh Squeak3.9alpha-6599 with text/list/code fonts set to Accuny-10: 1-10: 10252ms 11-20: 13857ms 21-30: 18047ms
fresh Squeak3.9alpha-6605 (alwaysHideHScrollbar is on at this update level): 1-10: 6698ms 11-20: 11627ms 21-30: 17238ms
fresh Squeak3.9alpha-6605 with roundedWindowCorners turned off: 1-10: 6671ms 11-20: 7976ms 21-30: 3704ms !! (probably because of window placement)
fresh Squeak3.9alpha-6605 with text/list/code fonts set to Accuny-10: 1-10: 5980ms 11-20: 9345ms 21-30: 13153ms
fresh Squeak3.9alpha-6605 with reverseWindowStagger turned off: 1-10: 6471ms 11-20: 11486ms 21-30: 16859ms
fresh Squeak3.9alpha-6605 with all three above settings: 1-10: 3584ms 11-20: 3318ms 21-30: 3325ms
So, bottom line, if you change these particular preferences in 3.9alpha, opening a browser is about eight times as fast if you already have a couple dozen windows open.
Anyway, there needs to be a balance between performance and appearance. Probably some significant performance gains could be had by fiddling with MessageTally and improving some Morphic behavior.
- Doug
On Apr 2, 2005, at 2:45 AM, stéphane ducasse wrote:
Propose others. I could not work with the horizontal scrollbar on. This is why I disable them. And 3.9 should not be sluggish. because we should not trade beauty for a system that nobody uses because it is too slow.
Stef
On 1 avr. 05, at 23:39, Lic. Edgar J. De Cleene wrote:
On [DATE], "[NAME]" <[ADDRESS]> wrote:
6604-prefers-sd Disable the preference for horizontal scrollbars that made everything so sluggish in addition to the rest :)
Are you reallu sure ?
I founding several things in Preferences , what when finish collecting and test submit to Mantis, but IMHO Preferences should be a user taste thing.
Edgar
On Apr 3, 2005, at 3:28 PM, stéphane ducasse wrote:
fresh Squeak3.9alpha-6605 with reverseWindowStagger turned off: 1-10: 6471ms 11-20: 11486ms 21-30: 16859ms
Hi Doug
What does exactly this preference?
This preference covers where new SystemWindows are opened on the screen. #reverseWindowStagger (which is on by default) causes windows to be opened in scattered areas of the screen, so that previously opened windows are less likely to be covered. If it's off, new windows always open in the upper-left corner. It doesn't have a huge impact on performance, but there seems to be less need for window redrawing when it's off, since it's always covering the previously opened window.
Anyway, I wouldn't mind a third type of window-placing behavior, something more like the Mac OS, where new windows are opened ~30 pixels below the previous window.
- Doug
Hi,
May I suggest to reduce a little the default size of the list font, and/or, make system windows a little bigger? I like the bigger fonts of 3.9 but the 4 columns browsers are an usability pain. And if we are going to remove the horizontal scrollbars it will make that aspect a lot worse.
My 2 cents.
Regards, Hernán
Doug Way wrote:
I like the functionality of the horizontal scrollbar in the list panes, but I have to admit that it does cause a significant performance hit.
I thought most of the slowdown in 3.9alpha (when opening browsers etc) was due to the antialiased fonts, but I guess the horizontal scrollbars are a bigger factor. Maybe with some MessageTally investigation there's a way to speed those up.
Anyway, I tried fiddling with various preferences while running the following benchmark:
Time millisecondsToRun: [10 timesRepeat: [Browser fullOnClass: Browser selector: openBrowser]]
To test, start with a fresh image, open one workspace. (Also open one browser first to select green "#Browser" type and then close the browser.) Do a print-it on the benchmark, and then do it again twice, leaving the extra windows open. (How many windows are open has a huge effect on performance.) Tested on a Mac G5 - 1.8GHz.
Here are the results I saw. ("1-10" means opening the first 1 to 10 browsers, "11-20" is the next 10, etc)
fresh Squeak3.9alpha-6599 (alwaysHideHScrollbar is off at this update level): 1-10: 12994ms 11-20: 18554ms 21-30: 23814ms
fresh Squeak3.9alpha-6599 with text/list/code fonts set to Accuny-10: 1-10: 10252ms 11-20: 13857ms 21-30: 18047ms
fresh Squeak3.9alpha-6605 (alwaysHideHScrollbar is on at this update level): 1-10: 6698ms 11-20: 11627ms 21-30: 17238ms
fresh Squeak3.9alpha-6605 with roundedWindowCorners turned off: 1-10: 6671ms 11-20: 7976ms 21-30: 3704ms !! (probably because of window placement)
fresh Squeak3.9alpha-6605 with text/list/code fonts set to Accuny-10: 1-10: 5980ms 11-20: 9345ms 21-30: 13153ms
fresh Squeak3.9alpha-6605 with reverseWindowStagger turned off: 1-10: 6471ms 11-20: 11486ms 21-30: 16859ms
fresh Squeak3.9alpha-6605 with all three above settings: 1-10: 3584ms 11-20: 3318ms 21-30: 3325ms
So, bottom line, if you change these particular preferences in 3.9alpha, opening a browser is about eight times as fast if you already have a couple dozen windows open.
Anyway, there needs to be a balance between performance and appearance. Probably some significant performance gains could be had by fiddling with MessageTally and improving some Morphic behavior.
- Doug
On Apr 2, 2005, at 2:45 AM, stéphane ducasse wrote:
Propose others. I could not work with the horizontal scrollbar on. This is why I disable them. And 3.9 should not be sluggish. because we should not trade beauty for a system that nobody uses because it is too slow.
Stef
On 1 avr. 05, at 23:39, Lic. Edgar J. De Cleene wrote:
On [DATE], "[NAME]" <[ADDRESS]> wrote:
6604-prefers-sd Disable the preference for horizontal scrollbars that made everything so sluggish in addition to the rest :)
Are you reallu sure ?
I founding several things in Preferences , what when finish collecting and test submit to Mantis, but IMHO Preferences should be a user taste thing.
Edgar
The suffix falsely gives me credit for the changeset, but this was done by Boris and I just looked at it on mantis.
Alex
stéphane ducasse schrieb:
6601HexFixes-laza -- Boris Gaertner -- 16 March 2005 with changes set #6217 the method Integer>>hex was deprecated. This change set is an attempt to replace the remaining senders of hex with either storeStringHex or printStringHex"
Hi,
Thanks, this now lives at:
http://box1.squeakfoundation.org/files/3.9/Squeak3.9a-6606.zip
cheers
bruce
=?ISO-8859-1?Q?st=E9phane_ducasse?= ducasse@iam.unibe.ch wrote:
Date: Fri, 1 Apr 2005 21:40:29 +0200 From: stZphane ducasse ducasse@iam.unibe.ch Subject: [UPDATES] 3.9 alpha 5 more To: The general-purpose Squeak developers list squeak-dev@lists.squeakfoundation.org reply-to: The general-purpose Squeak developers list squeak-dev@lists.squeakfoundation.org content-length: 826
hi all
Some updates for 3.9alpha to warn up again. :)
6601HexFixes-laza -- Boris Gaertner -- 16 March 2005 with changes set #6217 the method Integer>>hex was deprecated. This change set is an attempt to replace the remaining senders of hex with either storeStringHex or printStringHex"
6602CleaningUtilitiesAgain-sd -- stephane ducasse -- 30 March 2005 slowly we will get there..."
6603CleaningUtilities-sd -- stephane ducasse -- 30 March 2005 Move some utilities close to the class they use, i.e. Scanner"
6604-prefers-sd Disable the preference for horizontal scrollbars that made everything so sluggish in addition to the rest :)
6605-FixeScrollbars-sd -- stZphane ducasse -- 1 April 2005 fixing old instance of scrollbars now there is no popped up debugger anymore. '
Now we can start to work with 3.9.
Stef
squeak-dev@lists.squeakfoundation.org