Cuis-Smalltalk / Cuis-Smalltalk-Dev

Active development of Cuis Smalltalk
MIT License
435 stars 71 forks source link

Layout broken? TerseGuide broken? #145

Closed Phantasus closed 5 years ago

Phantasus commented 5 years ago

Hi,

is the Terse Guide or the layouts within Morphic in some way broken?

For comparison, this image (version 3430) seems to be more correct:

git checkout 56a268eaa18e26659784d58e934e67c8f5963d85
squeak Cuis5.0-3430.image

Here in this image the TerseGuide looks like this:

terseguide_cuis5_3430

In this "correct" image the TerseGuide has an extent:

terseguide_cuis5_3430_extentpresent

The wrong image (version 3470) can be reproduced by:

git checkout 3ff94bb4d08d2479dfb098ba1167bc47800b0242
squeak Cuis5.0-3470.image

Now in the "wrong" image (version 3470) seems "something" fishy:

terseguide_cuis5_3470

the textarea is "squished" and not anymore filling the rest of the window. As an interesting finding the TerseGuide has no extent anymore, it sits at zero:

terseguide_cuis5_3470_zeroextent

I have no clue why this happens or what exact change caused this behavior or else I would provide a Changeset. I hope this helps.

I also used the build result of the opensmalltalk-vm:

git clone https://github.com/OpenSmalltalk/opensmalltalk-vm.git
git checkout a28a7546d6dc8e1f366c5f9848c78e5fad400547
cd opensmalltalk-vm/build.linux64x64/squeak.cog.spur/build
./mvm

and using the product:

opensmalltalk-vm/products/sqcogspur64linuxht/squeak

Cheers.

jvuletich commented 5 years ago

I could not reproduce exactly what you see, but...

I think there is something weird going on on the VM. If I run Cuis with the recommended VM for 64 bit Linux (see GettingStarted.md) I can evaluate and get:


0.2 hex. '3FC999999999999A'.```

But if I pull openSmalltalkVM master branch, build, and run:
```0.2 storeString. '0.0026124986455922'.
0.2 hex. '0000559B8DA2EB88'.

0.2 storeString.  "repeated evaluations"
'0.00000002706508972' 
'0.00261249864559487' 
'0.00261249864559487' 
'0.00261249864559354' 

0.2 hex. "repeated evaluations"    
'0000559B8E2E78E0' 
'0000559B8E2E78E0' 
'0000559B8E2E78E0' 
'0000559B8E3C7430' ```
I get results that change over time! And if I run Float Tests (in Tests package) most fail.

Please try again with the recommended VM: https://github.com/OpenSmalltalk/opensmalltalk-vm/releases/download/201807260206/squeak.cog.spur_linux64x64_201807260206.tar.gz and see if problems persist. Also try evaluating `0.2 storeString` and `0.2 hex` several times in a workspace.

Thanks!
Phantasus commented 5 years ago

Well that's interesting. It seems to be working with the binary from github. But not when building from source of the source code of which the binary is claimed to be the result.

That's highly not intuitive and doesn't increase my trust in the development process of the opensmalltalk-vm.

When evaluating 0.2 hex -> '0.0'

Interesting, whatever happens there. It also depends on the machine.

hernanwilkinson commented 5 years ago

Could you report it to the VM list? We should report this weird behavior

On Thu, Oct 18, 2018 at 3:34 PM Phantasus notifications@github.com wrote:

Well that's interesting. It seems to be working with the binary from github. But not when building from source of the source code of which the binary is claimed to be the result.

That's highly not intuitive and doesn't increase my trust in the development process of the opensmalltalk-vm.

When evaluating 0.2 hex -> '0.0'

Interesting, whatever happens there. It also depends on the machine.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/Cuis-Smalltalk/Cuis-Smalltalk-Dev/issues/145#issuecomment-431115603, or mute the thread https://github.com/notifications/unsubscribe-auth/AADk99zqSBXyfHamvMIlZ_fO0SbaJDqcks5umMm9gaJpZM4XlCxK .

--

Hernán WilkinsonAgile Software Development, Teaching & Coaching Phone: +54-011-4893-2057 Twitter: @HernanWilkinson site: http://www.10Pines.com http://www.10pines.com/ Address: Alem 896, Floor 6, Buenos Aires, Argentina

Phantasus commented 5 years ago

But still how can it happen, that in the old image version 3430, the TerseGuide works? And also the evaluations of:

0.2 printString. -> '0.2'
0.2 hex.           -> '0000000'

These things intermingle. VM -> Image -> VM. Still a curious behaviour. Image wrong? VM wrong? How should that be formulated and who is best equipped to write a mail to the VM list?