Smalltalk › Pharo › Pharo Smalltalk Developers

IslandVMTweaksTestCase and Memory corruption

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

5 messages Options

Clément Béra

IslandVMTweaksTestCase and Memory corruption

Hello,

I was looking for VM tests in the Pharo image and I found the class IslandVMTweaksTestCase. The strange thing is its comments :

"Test case for some tweaks to the VM that Islands requires. These tests are largely for documentation; with an un-tweaked VM, the tests mostly still succeed, albeit with possible memory corruption."

Does that really mean if I run those tests it might result into memory corruption ?

I don't think it is a good idea to ship Pharo with this memory corruption tests in IslandVMTweaksTestCase. I suggest that if the tests may really corrupt memory we remove them from Pharo, or if they cannot corrupt memory then we fix the memory corruption comment.

Regards,

Marcus Denker-4

Re: IslandVMTweaksTestCase and Memory corruption

On Feb 28, 2013, at 3:09 PM, Clément Bera <[hidden email]> wrote:

Hello,

I was looking for VM tests in the Pharo image and I found the class IslandVMTweaksTestCase. The strange thing is its comments :
"Test case for some tweaks to the VM that Islands requires. These tests are largely for documentation; with an un-tweaked VM, the tests mostly still succeed, albeit with possible memory corruption."

Does that really mean if I run those tests it might result into memory corruption ?

I don't think it is a good idea to ship Pharo with this memory corruption tests in IslandVMTweaksTestCase. I suggest that if the tests may really corrupt memory we remove them from Pharo, or if they cannot corrupt memory then we fix the memory corruption comment.

Yes, we should not ship code like that.

Marcus

Ciprian Teodorov-3

Re: IslandVMTweaksTestCase and Memory corruption

No, I think that those tests should be shipped... And fix the VM to not corrupt the memory. That's only in the case these test are wrote in pure smalltalk. If they use ffi then yeah they should be removed

On Feb 28, 2013 3:11 PM, "Marcus Denker" <[hidden email]> wrote:

On Feb 28, 2013, at 3:09 PM, Clément Bera <[hidden email]> wrote:

Hello,

I was looking for VM tests in the Pharo image and I found the class IslandVMTweaksTestCase. The strange thing is its comments :
"Test case for some tweaks to the VM that Islands requires. These tests are largely for documentation; with an un-tweaked VM, the tests mostly still succeed, albeit with possible memory corruption."

Does that really mean if I run those tests it might result into memory corruption ?

I don't think it is a good idea to ship Pharo with this memory corruption tests in IslandVMTweaksTestCase. I suggest that if the tests may really corrupt memory we remove them from Pharo, or if they cannot corrupt memory then we fix the memory corruption comment.

Yes, we should not ship code like that.

Marcus

Marcus Denker-4

Re: IslandVMTweaksTestCase and Memory corruption

On Feb 28, 2013, at 4:15 PM, Ciprian Teodorov <[hidden email]> wrote:

No, I think that those tests should be shipped... And fix the VM to not corrupt the memory. That's only in the case these test are wrote in pure smalltalk. If they use ffi then yeah they should be removed

They have primitives in methods for which the primitive was not written.

On Feb 28, 2013 3:11 PM, "Marcus Denker" <[hidden email]> wrote:

On Feb 28, 2013, at 3:09 PM, Clément Bera <[hidden email]> wrote:

Hello,

I was looking for VM tests in the Pharo image and I found the class IslandVMTweaksTestCase. The strange thing is its comments :
"Test case for some tweaks to the VM that Islands requires. These tests are largely for documentation; with an un-tweaked VM, the tests mostly still succeed, albeit with possible memory corruption."

Does that really mean if I run those tests it might result into memory corruption ?

I don't think it is a good idea to ship Pharo with this memory corruption tests in IslandVMTweaksTestCase. I suggest that if the tests may really corrupt memory we remove them from Pharo, or if they cannot corrupt memory then we fix the memory corruption comment.

Yes, we should not ship code like that.

Marcus

Fernando olivero-2

Re: IslandVMTweaksTestCase and Memory corruption

In reply to this post by Ciprian Teodorov-3

Island are part of the TeaTime implementation in Croquet. If the tests
are referring to that "Island" then why should Pharo include such
tests?

Fernando

On Thu, Feb 28, 2013 at 4:23 PM, Marcus Denker <[hidden email]> wrote:

>
> On Feb 28, 2013, at 4:15 PM, Ciprian Teodorov <[hidden email]<mailto:[hidden email]>> wrote:
>
>
> No, I think that those tests should be shipped... And fix the VM to not corrupt the memory. That's only in the case these test are wrote in pure smalltalk. If they use ffi then yeah they should be removed
>
> They have primitives in methods for which the primitive was not written.
>
> On Feb 28, 2013 3:11 PM, "Marcus Denker" <[hidden email]<mailto:[hidden email]>> wrote:
>
> On Feb 28, 2013, at 3:09 PM, Clément Bera <[hidden email]<mailto:[hidden email]>> wrote:
>
> Hello,
>
> I was looking for VM tests in the Pharo image and I found the class IslandVMTweaksTestCase. The strange thing is its comments :
> "Test case for some tweaks to the VM that Islands requires. These tests are largely for documentation; with an un-tweaked VM, the tests mostly still succeed, albeit with possible memory corruption."
>
> Does that really mean if I run those tests it might result into memory corruption ?
>
> I don't think it is a good idea to ship Pharo with this memory corruption tests in IslandVMTweaksTestCase. I suggest that if the tests may really corrupt memory we remove them from Pharo, or if they cannot corrupt memory then we fix the memory corruption comment.
>
> Yes, we should not ship code like that.
>
> Marcus
>
>