[squeak-dev] Browsers slow in 3.9? Don't use rounded window corners.

> Hello,
>
>  several people have observed slow browser behavior in version 3.9, in
>  particular that browsers get slower the more of them you open [1].
>  Benchmarks showed that things were faster in 3.8 [2]. Jerome Peace
>  narrowed it down to 3.9 alpha, between 6684 and 6686. I have seen the
>  slowdownn disappear when switching off rounded window corners in my
>  3.9.1 dev image, see [1] for details.
>
>  So for now, if you want faster screen updates with multipe browsres
>  open, do not use rounded window corners:
>
>   Preferences disable: #roundedWindowCorners.
>  thanks to everyone who helped with this.
>
>
>
>  [1] Mantis issue 0002084: Each overlaping browser slows down screen
>  updates considerably
>     http://bugs.squeak.org/view.php?id=2084
>  [2] Thread "3.9 Browser speed" started by Andreas Raab,
>     http://thread.gmane.org/gmane.comp.lang.smalltalk.squeak.general/108359
>

> Hello,
>
> several people have observed slow browser behavior in version 3.9, in
> particular that browsers get slower the more of them you open [1].
> Benchmarks showed that things were faster in 3.8 [2]. Jerome Peace
> narrowed it down to 3.9 alpha, between 6684 and 6686. I have seen the
> slowdownn disappear when switching off rounded window corners in my
> 3.9.1 dev image, see [1] for details.
>
> So for now, if you want faster screen updates with multipe browsres
> open, do not use rounded window corners:
>
>  Preferences disable: #roundedWindowCorners.
> thanks to everyone who helped with this.
>
>
>
> [1] Mantis issue 0002084: Each overlaping browser slows down screen
> updates considerably
>    http://bugs.squeak.org/view.php?id=2084
> [2] Thread "3.9 Browser speed" started by Andreas Raab,
>    http://thread.gmane.org/gmane.comp.lang.smalltalk.squeak.general/108359
>
>

> I get another factor of 2 speedup when using TrueType fonts in  
> Browsers with:
>
> Installer mantis bug: 6089 fix:'TTFSpeedUp-dgd.2.cs'.
>
> Now browser speed with truetype fonts is comparable to speed with
> bitmapped fonts.
>
> The Mantis issue  http://bugs.squeak.org/view.php?id=6089 says this
> improvement will be in 3.10, might be a good idea to have it included
> in the next dev image based on 3.9.1 as well :-)

>
>
> Cheers
>
> Matthias
>
> On Fri, May 2, 2008 at 1:28 PM, Matthias Berth
> <[hidden email]> wrote:
>> Hello,
>>
>> several people have observed slow browser behavior in version 3.9, in
>> particular that browsers get slower the more of them you open [1].
>> Benchmarks showed that things were faster in 3.8 [2]. Jerome Peace
>> narrowed it down to 3.9 alpha, between 6684 and 6686. I have seen the
>> slowdownn disappear when switching off rounded window corners in my
>> 3.9.1 dev image, see [1] for details.
>>
>> So for now, if you want faster screen updates with multipe browsres
>> open, do not use rounded window corners:
>>
>>  Preferences disable: #roundedWindowCorners.
>> thanks to everyone who helped with this.
>>
>>
>>
>> [1] Mantis issue 0002084: Each overlaping browser slows down screen
>> updates considerably
>>    http://bugs.squeak.org/view.php?id=2084
>> [2] Thread "3.9 Browser speed" started by Andreas Raab,
>>    http://thread.gmane.org/gmane.comp.lang.smalltalk.squeak.general/108359
>>
>
>

>
>  On May 2, 2008, at 3:21 PM, Matthias Berth wrote:
> > I get another factor of 2 speedup when using TrueType fonts in Browsers
> with:
> >
> > Installer mantis bug: 6089 fix:'TTFSpeedUp-dgd.2.cs'.
> >
> > Now browser speed with truetype fonts is comparable to speed with
> > bitmapped fonts.
> >
> > The Mantis issue  http://bugs.squeak.org/view.php?id=6089 says this
> > improvement will be in 3.10, might be a good idea to have it included
> > in the next dev image based on 3.9.1 as well :-)
> >
>
>  Yes I tried but because there is always a but in squeak.
>  The wonderful pops that are used when MC loads a package use the fonts...
>  so it broke and I gave up because not the time to find how to stop the pop
> up behavior.
>  Popup menu and the like are a real plague....
>
>  Stef
>
>

>
>  On May 3, 2008, at 8:24 AM, Matthias Berth wrote:
>
>
> > I see, so it is another case of the system stepping on its own feet
> > when updating? I guess using the change set was not an option because
> > you want everything based on Monticello? I haven't followed all the
> > discussions around releases, so I'm just guessing.
> >
>
>  Yes this was that.
>  I like to eat our own dog food because this is the only way to really
>  push it to the limits.
>
>
>
> > Would an option for monticello, like "load the next package silently"
> > or "load the next package with the most primitive progress information
> > (i.e. no popups)" help in situations like this?
> >
>
>  Yes (atomic loading would be good too) but yes!
>  I wanted to have a look at that but was flooded by work.
>  If you have that feature I would love to use it.
>  Stef
>
> >
> >
>
>
>

>
> Matthias
>
>
> On Sat, May 3, 2008 at 5:58 PM, stephane ducasse
> <[hidden email]> wrote:
>  
>>  On May 3, 2008, at 8:24 AM, Matthias Berth wrote:
>>
>>
>>    
>>> I see, so it is another case of the system stepping on its own feet
>>> when updating? I guess using the change set was not an option because
>>> you want everything based on Monticello? I haven't followed all the
>>> discussions around releases, so I'm just guessing.
>>>
>>>      
>>  Yes this was that.
>>  I like to eat our own dog food because this is the only way to really
>>  push it to the limits.
>>
>>
>>
>>    
>>> Would an option for monticello, like "load the next package silently"
>>> or "load the next package with the most primitive progress information
>>> (i.e. no popups)" help in situations like this?
>>>
>>>      
>>  Yes (atomic loading would be good too) but yes!
>>  I wanted to have a look at that but was flooded by work.
>>  If you have that feature I would love to use it.
>>  Stef
>>
>>    
>>>      
>>
>>    
>
>
>  

>
>  On May 2, 2008, at 3:21 PM, Matthias Berth wrote:
>
>
> > I get another factor of 2 speedup when using TrueType fonts in Browsers
> with:
> >
> > Installer mantis bug: 6089 fix:'TTFSpeedUp-dgd.2.cs'.
> >
> > Now browser speed with truetype fonts is comparable to speed with
> > bitmapped fonts.
> >
> > The Mantis issue  http://bugs.squeak.org/view.php?id=6089 says this
> > improvement will be in 3.10, might be a good idea to have it included
> > in the next dev image based on 3.9.1 as well :-)
> >
>
>  Yes I tried but because there is always a but in squeak.
>  The wonderful pops that are used when MC loads a package use the fonts...
>  so it broke and I gave up because not the time to find how to stop the pop
> up behavior.
>  Popup menu and the like are a real plague....
>
>  Stef
>
>
>
>
>
> >
> >
> > Cheers
> >
> > Matthias
> >
> > On Fri, May 2, 2008 at 1:28 PM, Matthias Berth
> > <[hidden email]> wrote:
> >
> > > Hello,
> > >
> > > several people have observed slow browser behavior in version 3.9, in
> > > particular that browsers get slower the more of them you open [1].
> > > Benchmarks showed that things were faster in 3.8 [2]. Jerome Peace
> > > narrowed it down to 3.9 alpha, between 6684 and 6686. I have seen the
> > > slowdownn disappear when switching off rounded window corners in my
> > > 3.9.1 dev image, see [1] for details.
> > >
> > > So for now, if you want faster screen updates with multipe browsres
> > > open, do not use rounded window corners:
> > >
> > >  Preferences disable: #roundedWindowCorners.
> > > thanks to everyone who helped with this.
> > >
> > >
> > >
> > > [1] Mantis issue 0002084: Each overlaping browser slows down screen
> > > updates considerably
> > >   http://bugs.squeak.org/view.php?id=2084
> > > [2] Thread "3.9 Browser speed" started by Andreas Raab,
> > >
> http://thread.gmane.org/gmane.comp.lang.smalltalk.squeak.general/108359
> > >
> > >
> >
> >
> >
>
>
>