Then the change in 20460 is rather simple. It make sure that the MC cache does not return old method information when snaphotting. So I do not see why we cannot fix the problems now. Stef On Dec 30, 2012, at 12:07 AM, Camillo Bruni wrote:

looks like the integration server is not failing:

https://ci.lille.inria.fr/pharo/job/Pharo-2.0/

looks like we have to rollback the MC changes from 20460

On 2012-12-29, at 18:19, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:

...

20461 -----

- Issue 4018: Color class enhancement. Thanks Alexander Bergel. http://code.google.com/p/pharo/issues/detail?id=4018

- Issue 7187: Doit is broken in spec. Thanks Benjamin van Ryseghem http://code.google.com/p/pharo/issues/detail?id=7187 Diff information: http://ss3.gemstone.com/ss/Pharo20/Spec-Tools-StephaneDucasse.77.diff http://ss3.gemstone.com/ss/Pharo20/Graphics-Primitives-StephaneDucasse.97.di...

Stef

Looking at the log I do not see how this is related to the changes we did. I'm puzzled Now I'm rerunning the tests flushing the cache MCMethodDefinition flushMethodCache first But I expect exactly the same result. Stef On Dec 30, 2012, at 10:57 AM, Stéphane Ducasse wrote:

So I downloaded the latest version and run the tests locally Here are the issues that we should fix

- so far I have only MCStReaderTest>>testCommentWithStyle and testMethodWithStyle that are red.

- Now I do not understand why the RB tests are so long. (spellchecking the complete system does not look like a good idea) I have testPullUWithMethodThatCannot… failing

- Why do we have the spell checker rules or tests in the image? (or why do we get the rb-spelling.dat file

- Then a large number of subscriptOutOfBounds: related to the progress bar problems. Camillo if you want to have a look I could never integrate your changes and changing the dispatch of Mouse event or related was not a good idea from my point of view.

- the VM crash and I do not see how this can be related to changing a tests in a cache for MC.

I took the version 20459 and run the tests

- I get the same MCStReaderTest>>testCommentWithStyle and testMethodWithStyle that are red. I have testPullUWithMethodThatCannot… failing a couple more ReleaseTest red RB tests are too long.

- Then a large number of subscriptOutOfBounds: related to the progress bar problems. Camillo if you want to have a look I could never integrate your changes and changing the dispatch of Mouse event or related was not a good idea from my point of view.

- I do not understand why the VM does not crash may this is related to the fact that now we fill up more the WeakRegistry?

I think that we should coordinate and streamline the process We started to address fixes that are burning us: - MC cache because it messed up the method recategorisation and it was introducing pain at each of our movement. -

Stef

<PharoDebug.log>

I took again the same image and did nothing but run the tests and it did not crash. Strange. On Dec 30, 2012, at 12:51 PM, Stéphane Ducasse wrote:

On Dec 30, 2012, at 12:32 PM, Stéphane Ducasse wrote:

...

Looking at the log I do not see how this is related to the changes we did. I'm puzzled Now I'm rerunning the tests flushing the cache MCMethodDefinition flushMethodCache first But I expect exactly the same result. The vm does not crash. I will redo the experiment but I wonder if this is not related to

- http://code.google.com/p/pharo/issues/detail?id=5276

Stef