[MacRuby-devel] experimental branch: status update

[Jordan Breeding]

On Apr 06, 2009, at 07:50, Eloy Duran wrote:

>>> As Laurent noted we are now passing most language specs. The ones  
>>> that we don't pass yet are either because we simply fail, or these  
>>> examples (tests) are simply not updated for Ruby 1.9 yet. Which as  
>>> you all know is what MacRuby is based on. This is an area where  
>>> all of you Ruby devs can help out if you feel the need to get  
>>> MacRuby up and running asap but don't want to work on the C(++)  
>>> code.
>>>
>>> If there are people who would like to work on getting examples up- 
>>> to-date, please respond and I will try to give you an outline on a  
>>> workflow.
>>>
>>> If you just want to run the specs use the spec:ci task.
>>>
>>> One last question I have for everyone on the list. If there's  
>>> someone with a 32 bit intel machine, could you please please run  
>>> the spec:ci task and see if you get any failures?
>>> It seems that, at least, because of a 32/64 bit issue some IO  
>>> specs are now failing on my machines and those of Laurent. Thanks  
>>> in advance!
>>
>> I am assuming the spec:ci task uses miniruby to run the tests  
>> currently, and will use miniruby or macruby in the future.
>>
>> Why not make a new task spec:ci32 that forces 32-bit execution on  
>> 64-bit machines using `arch -arch i386 miniruby` instead of  
>> `miniruby`?
>>
>> Just an idea.
>
> The question is not on how to circumvent these failures, rather the  
> question is _if_ these failures don't occur on 32 bit machines.
> Laurent and I both only have 64 bit these machines and we see these  
> failures, whereas Patrick who apparently has a 32 bit machine does  
> not see them.
>
> The specs should ultimately pass on both archs, so we'd like  
> feedback from other people on 32 bit on whether or not they see  
> these failures.
>
> Eloy
> _______________________________________________
> MacRuby-devel mailing list
> MacRuby-devel at lists.macosforge.org
> http://lists.macosforge.org/mailman/listinfo.cgi/macruby-devel

Sorry I only meant to suggest this to aid in testing as you suggest.  
By adding a second task spec:ci32 that forces 32-bit execution of  
tests you and Laurent could see if the problem goes away when running  
the 32-bit code, even on a 64-bit machine. Hopefully if it works for  
Patrick and is tied to 32-bit it is tied to the code and not to the  
hardware, in which case spec:ci would still fail on your machine and  
spec:ci32 would work, which would then hopefully help isolate the  
problem with 64-bit.

This also means spec could become a task that could test to see what  
kind of machine you are on, if you are on a 32-bit machine just run  
spec:ci, if you are on a 64-bit machine run spec:ci and spec:ci32 so  
that the 32-bit build always gets tested.

Hope that makes more sense now.

Jordan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 3847 bytes
Desc: not available
URL: <http://lists.macosforge.org/pipermail/macruby-devel/attachments/20090406/050fc026/attachment.bin>