3.0 almost green

very nice, 3.0 is almost green :)

https://ci.inria.fr/pharo/job/Pharo-3.0-Update-Step-2-Validation/lastCompletedBuild/testReport/

- some strange unknown processes only under mac and win
- 3 Zinc failures under win
- a chunkimport failure under win

On Jul 16, 2013, at 12:14 PM, Camillo Bruni <[hidden email]> wrote:

> very nice, 3.0 is almost green :)
>
> https://ci.inria.fr/pharo/job/Pharo-3.0-Update-Step-2-Validation/lastCompletedBuild/testReport/
>
> - some strange unknown processes only under mac and win

I vote for changing the test to just check for number of process <15 or something like that.

> - a chunkimport failure under win

This is randomly failing on all architectures:

Error Message

Got '(#testing testImportAMethod testImportAClass) (#auxiliar importAClass)' instead of '(#auxiliar importAClass) (#testing testImportAMethod testImportAClass)'.


I think the solution is to fix the test to not check for order.

        Marcus

On 16 Jul 2013, at 12:51, Marcus Denker <[hidden email]> wrote:

>> very nice, 3.0 is almost green :)
>>
>> https://ci.inria.fr/pharo/job/Pharo-3.0-Update-Step-2-Validation/lastCompletedBuild/testReport/
>>
>> - some strange unknown processes only under mac and win
>
> I vote for changing the test to just check for number of process <15 or something like that.

I have to defend my test ;-)

No seriously: ReleaseTest is there to guard quality (but #testMethodsWithUnboundGlobals is skipped). They all pass in my 3.0 image that I have been developing in for some days now. So that is really nice.

Ideally, we should not weaken the rules, but try to understand why it behaves so strangely. Maybe some bright soul can rewrite the test so that is less brittle.

Sven