[Vm-dev] VMMaker small problem for pharo developper

[Nicolas Cellier]

2010/3/19 Levente Uzonyi <leves at elte.hu>:
>
> On Fri, 19 Mar 2010, Henrik Johansen wrote:
>
>>
>>
>> On Mar 19, 2010, at 1:18 17PM, Levente Uzonyi wrote:
>>
>>> On Fri, 19 Mar 2010, Bert Freudenberg wrote:
>>>
>>>>
>>>> On 19.03.2010, at 10:44, Henrik Johansen wrote:
>>>>>
>>>>>
>>>>> the method will probably be renamed again in Pharo, to
>>>>> #primFindSubstring:in:startingAt:matchTable:
>>>>
>>>> I see no compelling reason why we shouldn't just apply the same
>>>> refactoring to Squeak? KISS ;)
>>>
>>> What about changing the primitive to fail if the String is "Wide", or
>>> make it work with WideStrings?
>>>
>>>
>>> Levente
>>
>> Sure, it's an orthagonal issue though.
>> You'd still need the guards in case an old VM is used, and have to skip
>> those if the image the VM was built with included the primitive code which
>> works correctly.
>> IE. it'd be a real mess, and you'd definately want to have to do it in
>> only one place.
>
> I'd call it parallel, not orthogonal. If the primitive is fixed, then in
> Squeak 4.2/Pharo 1.2/etc we can remove the checks/hacks in the image and
> assume that the issue is handled correctly by the vm.
>

... or the primitive fallback code.
I agree with Levente, the first thing to do is to protect the primitive

Nicolas

>
> Levente
>
>>
>> Cheers,
>> Henry
>>
>>
>