Re: Pharo - Squeak VM bug?

> I believe I found cool bug (probably in Squeak VM, also affects  
> Exupery
> VM), but also other Smalltalks are affected, incl. Smalltalk/X, Cuis
> and Squeak itself .. not checked with Visual Works though.
>
> Just try to inspect "a:=Array new: SmallInteger maxVal." and see what
> is value of "a" variable - interestingly, primitive (in Squeak VM,  
> primitive
> 71) DOES NOT fail on too big arguments (however, it fails on negative
> arguments). What is interesting is that Smalltalk/X crashes, while  
> Cuis,
> Pharo a Squeak actually assign "a" a value - check yourself what is
> assigned (depends on Smalltalk variant). No Smalltalk variant I  
> checked
> fails a primitive.
>
> Maybe it is on purpose, but I believe this situation shall be somehow
> signalled to developer as "Array>>new:" method did not provide what
> is expected (returns junk).
>
> In case of a miscall, sorry for your time.
>
> Greets,
> Petr Vacha
>
>
>
>

> Hi again,
> when try to find out that Array>>new: bug I found one thing that I  
> think
> may be very confusing - somehow mangled information from hierarchy  
> browsers
> (see figure).
>
> Array>>new: calls primitive and uses self>>basicNew: call as fallback.
> Array does not implement basicNew: so I tried to find implementors of
> it. Behavior has this implementation. However, when you show hierarchy
> of Array, it ends in ProtoObject with no Behavior involved. I think
> this to be is somewhat confusing for beginners especially when they
> are trying to get into method lookup chain.
>
> Petr Vacha
>
>
> <pharo-browsers.jpg>

> I suggest you to read the chapter on the Smalltalk
> model in the Squeak
> or pharo by example book
> the key is:
>
> when you look for a message you look in the CLASS of
> the receiver
> then its superclasses.
>
> So when you look for new sent to Array you should look
> in the Array
> class superclasses and you will see it contains
> Behavior before ProtoObject
>
> Stef
>
> PS: you can sent to the mailing-list just register
> to it. You do not
> need to have a google account.
>
> On Oct 7, 2009, at 1:00 PM, Petr Vácha wrote:
>
> > Hi again,
> > when try to find out that Array>>new: bug I found
> > one thing that I
> > > think
> > may be very confusing - somehow mangled information
> > from hierarchy
> > > browsers
> > (see figure).
> >
> > Array>>new: calls primitive and uses self>>basicNew:
> > call as fallback.
> > > Array does not implement basicNew: so I tried to
> > find implementors of
> > > it. Behavior has this implementation. However, when
> > you show hierarchy
> > > of Array, it ends in ProtoObject with no Behavior
> > involved. I think
> > > this to be is somewhat confusing for beginners especially
> > when they
> > > are trying to get into method lookup chain.
> >
> > Petr Vacha
> >
> >
> > <pharo-browsers.jpg>
>

> I know Smalltalk hierarchy, I just wanted to point out, that Hierarchy
> Browser does not provide (maybe I am dumb, but I do not see any)  
> obvious
> way to show the hierarchy to see that Array class inherits from  
> Behavior.
>
> PV.
>
> ----- PŮVODNÍ ZPRÁVA -----
> Od: "Stéphane Ducasse" <[hidden email]>
> Komu: "Petr Vácha" <[hidden email]>
> Předmět: Re: Hierarchy browsers & implementers of
> Datum: 7.10.2009 - 21:22:42
>
>> I suggest you to read the chapter on the Smalltalk
>> model in the Squeak
>> or pharo by example book
>> the key is:
>>
>> when you look for a message you look in the CLASS of
>> the receiver
>> then its superclasses.
>>
>> So when you look for new sent to Array you should look
>> in the Array
>> class superclasses and you will see it contains
>> Behavior before ProtoObject
>>
>> Stef
>>
>> PS: you can sent to the mailing-list just register
>> to it. You do not
>> need to have a google account.
>>
>> On Oct 7, 2009, at 1:00 PM, Petr Vácha wrote:
>>
>>> Hi again,
>>> when try to find out that Array>>new: bug I found
>>> one thing that I
>>>> think
>>> may be very confusing - somehow mangled information
>>> from hierarchy
>>>> browsers
>>> (see figure).
>>>
>>> Array>>new: calls primitive and uses self>>basicNew:
>>> call as fallback.
>>>> Array does not implement basicNew: so I tried to
>>> find implementors of
>>>> it. Behavior has this implementation. However, when
>>> you show hierarchy
>>>> of Array, it ends in ProtoObject with no Behavior
>>> involved. I think
>>>> this to be is somewhat confusing for beginners especially
>>> when they
>>>> are trying to get into method lookup chain.
>>>
>>> Petr Vacha
>>>
>>>
>>> <pharo-browsers.jpg>
>>
>
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>

> I know Smalltalk hierarchy, I just wanted to point out, that Hierarchy
> Browser does not provide (maybe I am dumb, but I do not see any)  
> obvious
> way to show the hierarchy to see that Array class inherits from  
> Behavior.
>
> PV.
>
> ----- PŮVODNÍ ZPRÁVA -----
> Od: "Stéphane Ducasse" <[hidden email]>
> Komu: "Petr Vácha" <[hidden email]>
> Předmět: Re: Hierarchy browsers & implementers of
> Datum: 7.10.2009 - 21:22:42
>
>> I suggest you to read the chapter on the Smalltalk
>> model in the Squeak
>> or pharo by example book
>> the key is:
>>
>> when you look for a message you look in the CLASS of
>> the receiver
>> then its superclasses.
>>
>> So when you look for new sent to Array you should look
>> in the Array
>> class superclasses and you will see it contains
>> Behavior before ProtoObject
>>
>> Stef
>>
>> PS: you can sent to the mailing-list just register
>> to it. You do not
>> need to have a google account.
>>
>> On Oct 7, 2009, at 1:00 PM, Petr Vácha wrote:
>>
>>> Hi again,
>>> when try to find out that Array>>new: bug I found
>>> one thing that I
>>>> think
>>> may be very confusing - somehow mangled information
>>> from hierarchy
>>>> browsers
>>> (see figure).
>>>
>>> Array>>new: calls primitive and uses self>>basicNew:
>>> call as fallback.
>>>> Array does not implement basicNew: so I tried to
>>> find implementors of
>>>> it. Behavior has this implementation. However, when
>>> you show hierarchy
>>>> of Array, it ends in ProtoObject with no Behavior
>>> involved. I think
>>>> this to be is somewhat confusing for beginners especially
>>> when they
>>>> are trying to get into method lookup chain.
>>>
>>> Petr Vacha
>>>
>>>
>>> <pharo-browsers.jpg>
>>
>
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project