[squeak-dev] Latest VM appears to mess with initial window size

Marcel Taeumel marcel.taeumel at hpi.de
Wed Jun 1 09:19:37 UTC 2022


Hi Tim --

Hmm... I cannot reproduce that in Ubuntu 18. What I can reproduce is a very strange behavior of primitiveHostWindowSizeSet in X11. I will look into that.

Best,
Marcel
Am 31.05.2022 19:07:53 schrieb tim Rowledge <tim at rowledge.org>:
Pi-4-1
screen size 1600 at 1200
VM /home/pi/Documents/Squeak/sqcogspur64ARMv8linuxht/lib/squeak/5.0-202205110711-64bit/squeak
image squeak 6 alpha 21829
opens with display extent = 1016 at 702
increase to 1448 at 986 by usual dragging

save and quit

restart
display extent is 1016 at 702.

Copy this file to Goldskin (another Pi 4)
screen size 1920 at 1440
VM /home/pi/Documents/Squeak/sqcogspur64ARMv8linuxht/lib/squeak/5.0-202205110711-64bit/squeak
opens with display extent1448 at 986
drag to extent 1632 at 1092
save and quit
reopen
display extent is 1632 at 1092

copy *that* back to Pi-4-1
display extent ... 1016 at 702

> On 2022-05-31, at 8:26 AM, Marcel Taeumel wrote:
>
> Okay. Let my try to understand what is happening here.
>
> Platform: LinuxARMv8
>
> 1. Resize window to ??? then save-and-quit.
> 2. Open on a 1920 by 1080 screen but window is smaller than ???.
> 3. Open on a 1920 by 1440 screen and window is exactly ???.
>
> So, what is ??? in this scenario? :-)
>
> And what are the results of...
>
> Display platformScaleFactor.
> Display uiScaleFactor.
>
> Best,
> Marcel
>> Am 30.05.2022 19:02:24 schrieb tim Rowledge :
>>
>> Sorry if I wasn't clear - it's the *host* window size that is messed up for me.
>>
>> > On 2022-05-30, at 2:40 AM, Marcel Taeumel wrote:
>> >
>> > Hi Tim --
>> >
>> > Are you referring to the host-window size or the system-window size? I think you mean the latter. Whether or not the scale factor is involved, you can check by looking at what is selected in "Extras > Scale Factor" menu.
>> >
>> > I am not sure that I really understand what you mean. However, there are two hard-coded numbers in RealEstateAgent class >> #strictlyStaggeredInitialFrameFor:... which we did not adjust to the scale factor.
>> >
>> > ...
>> > "Number to be staggered at each corner (less on small screens)"
>> > maxLevel := allowedArea area > 300000 ifTrue: [3] ifFalse: [2].
>> > "Amount by which to stagger (less on small screens)"
>> > grid := allowedArea area > 500000 ifTrue: [40] ifFalse: [20].
>> > ...
>> >
>> > Would you apply a "* self scaleFactor" to those areas and see if the problem gets fixed? :-) Maybe also the grid.
>> >
>> > ---
>> >
>> > This is most likely not a VM issue but related to a feature in newer VMs. It might be that a newer VM will actually deliver that platformScaleFactor which will then set the image-side uiScaleFactor which will then modify the "RealEstateAgent scaleFactor" which will then result in bigger default-window areas and thus hit those magic boundaries. Not sure.
>> >
>> > Best,
>> > Marcel
>> >> Am 29.05.2022 20:54:38 schrieb tim Rowledge :
>> >>
>> >> I'm just trying the latest release VM (ARMv8 on a Pi in this case) and it seems to be doing Odd Things with the window size.
>> >>
>> >> The same image on two different Pis opens with different window sizes. One Pi has a 1920 at 180 screen and the initial window is notably smaller than what it was when I saved the image, the other has a 1920 at 1440 screen and the initial window size is what I set. That size is not larger than the 1920 at 1080 screen, by the way. To the best of my recollection this is new (mis)behaviour.
>> >>
>> >> We've always done some startup checks in the VM to fit the window within the available display limits but I don't remember previously over-riding the requested size this aggressively. Has anyone changed this code recently? Or is this a side effect of recent scale factor stuff? I can't spot any suspicious code in the image right now.
>> >>
>> >> tim
>> >> --
>> >> tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
>> >> A misplaced modifier walks into a bar owned a man with a glass eye named Ralph
>> >>
>> >>
>> >
>>
>>
>> tim
>> --
>> tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
>> Useful random insult:- Always responds to "Make Money Fast" postings on the Net.
>>
>>
>>
>


tim
--
tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
Strange OpCodes: VMB: Verify, then Make Bad



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20220601/fd31291a/attachment.html>


More information about the Squeak-dev mailing list