Problem and solution for DistributionMaps in Pharo 1.1 [WAS] Fwd: problem with CompiledMethod equality

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Problem and solution for DistributionMaps in Pharo 1.1 [WAS] Fwd: problem with CompiledMethod equality

Mariano Martinez Peck
Hi guys. In that thread in the mailing lists, you can see the problem I had in Pharo 1.1 as keys of a Dictionary answers a Set.
I used CompiledMethod as properties, and the equality of CompiledMethod answers true for methods even with different selectors. So I my case I have problems because elementsAndPropertyValue keys answer me less thing that I put.

The solution, was to change elementsAndPropertyValue to a IdentityDictionary instead of Dictionary.
So, I have to subclass DistributionMap, and override initializeInstanceVariables  so that:

initializeInstanceVariables

    partsAndElements := Dictionary new.
    elementsAndPropertyValue := IdentityDictionary new.
    colorMap := Dictionary new.
    colors := Color strongColors


Do you think this is good ? do you have a better solution ?

Thanks!

Mariano




---------- Forwarded message ----------
From: Mariano Martinez Peck <[hidden email]>
Date: Tue, May 18, 2010 at 3:35 PM
Subject: problem with CompiledMethod equality
To: Pharo Development <[hidden email]>


Hi folks. I was debugging a problem with Moose and I realized that 2 different methods can have the same CompiledMethod. This was weird for me. I don't know if this is correct or not.

For example, evaluate:

(InstructionClient>>#methodReturnTop) = (InstructionClient>>#doDup)    ->  true

If you look at

CompiledMethod = aCompiledMethod 
    "Answer whether the receiver implements the same code as aCompiledMethod."

    | numLits |
    self == aCompiledMethod
        ifTrue: [ ^ true ].
    self class = aCompiledMethod class
        ifFalse: [ ^ false ].
    self size = aCompiledMethod size
        ifFalse: [ ^ false ].
    self header = aCompiledMethod header
        ifFalse: [ ^ false ].
    self initialPC to: self endPC do: [ :i |
        (self at: i) = (aCompiledMethod at: i)
            ifFalse: [ ^ false ] ].
    (self sameLiteralsAs: aCompiledMethod)
        ifFalse: [ ^ false ].
    self halt.
    ^ true


It is returning in the last ^ true.

What was weird for me is that (self sameLiteralsAs: aCompiledMethod)  returns true also. But if I print both literals:

self literals ->  {#methodReturnTop. (#InstructionClient->InstructionClient)}

aCompiledMethod literals ->  {#doDup. (#InstructionClient->InstructionClient)}

So...for me they are differnt, but sameLiteralsAs:  is returning true. Debugging that, it seems that in that method it is returning in the last ^ true.

So...any hints? is this the expected behavior ?  or it is a bug ?

Thanks

Mariano




_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: Problem and solution for DistributionMaps in Pharo 1.1 [WAS] Fwd: problem with CompiledMethod equality

Mariano Martinez Peck
Sorry...I forgot to say that the problem with this solution is that not in all the distribution maps of I want IdentityDictionary, but only in some  :(

On Thu, May 20, 2010 at 12:39 PM, Mariano Martinez Peck <[hidden email]> wrote:
Hi guys. In that thread in the mailing lists, you can see the problem I had in Pharo 1.1 as keys of a Dictionary answers a Set.
I used CompiledMethod as properties, and the equality of CompiledMethod answers true for methods even with different selectors. So I my case I have problems because elementsAndPropertyValue keys answer me less thing that I put.

The solution, was to change elementsAndPropertyValue to a IdentityDictionary instead of Dictionary.
So, I have to subclass DistributionMap, and override initializeInstanceVariables  so that:

initializeInstanceVariables

    partsAndElements := Dictionary new.
    elementsAndPropertyValue := IdentityDictionary new.
    colorMap := Dictionary new.
    colors := Color strongColors


Do you think this is good ? do you have a better solution ?

Thanks!

Mariano




---------- Forwarded message ----------
From: Mariano Martinez Peck <[hidden email]>
Date: Tue, May 18, 2010 at 3:35 PM
Subject: problem with CompiledMethod equality
To: Pharo Development <[hidden email]>


Hi folks. I was debugging a problem with Moose and I realized that 2 different methods can have the same CompiledMethod. This was weird for me. I don't know if this is correct or not.

For example, evaluate:

(InstructionClient>>#methodReturnTop) = (InstructionClient>>#doDup)    ->  true

If you look at

CompiledMethod = aCompiledMethod 
    "Answer whether the receiver implements the same code as aCompiledMethod."

    | numLits |
    self == aCompiledMethod
        ifTrue: [ ^ true ].
    self class = aCompiledMethod class
        ifFalse: [ ^ false ].
    self size = aCompiledMethod size
        ifFalse: [ ^ false ].
    self header = aCompiledMethod header
        ifFalse: [ ^ false ].
    self initialPC to: self endPC do: [ :i |
        (self at: i) = (aCompiledMethod at: i)
            ifFalse: [ ^ false ] ].
    (self sameLiteralsAs: aCompiledMethod)
        ifFalse: [ ^ false ].
    self halt.
    ^ true


It is returning in the last ^ true.

What was weird for me is that (self sameLiteralsAs: aCompiledMethod)  returns true also. But if I print both literals:

self literals ->  {#methodReturnTop. (#InstructionClient->InstructionClient)}

aCompiledMethod literals ->  {#doDup. (#InstructionClient->InstructionClient)}

So...for me they are differnt, but sameLiteralsAs:  is returning true. Debugging that, it seems that in that method it is returning in the last ^ true.

So...any hints? is this the expected behavior ?  or it is a bug ?

Thanks

Mariano





_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: Problem and solution for DistributionMaps in Pharo 1.1 [WAS] Fwd: problem with CompiledMethod equality

Simon Denier-3
In reply to this post by Mariano Martinez Peck

On 20 mai 2010, at 12:39, Mariano Martinez Peck wrote:

Hi guys. In that thread in the mailing lists, you can see the problem I had in Pharo 1.1 as keys of a Dictionary answers a Set.
I used CompiledMethod as properties, and the equality of CompiledMethod answers true for methods even with different selectors. So I my case I have problems because elementsAndPropertyValue keys answer me less thing that I put.

The solution, was to change elementsAndPropertyValue to a IdentityDictionary instead of Dictionary.
So, I have to subclass DistributionMap, and override initializeInstanceVariables  so that:

initializeInstanceVariables

    partsAndElements := Dictionary new.
    elementsAndPropertyValue := IdentityDictionary new.
    colorMap := Dictionary new.
    colors := Color strongColors


Do you think this is good ? do you have a better solution ?

Well you could call super initializeInstanceVariables and override only the initializer for elementsAndPropertyValue. At least you will have less copy-paste code. But it sure does not look good (but again, if it works).


Could you produce a perfect hash key for each CompiledMethod and use that instead of relying directly on CompiledMethod>>= ?




Thanks!

Mariano




---------- Forwarded message ----------
From: Mariano Martinez Peck <[hidden email]>
Date: Tue, May 18, 2010 at 3:35 PM
Subject: problem with CompiledMethod equality
To: Pharo Development <[hidden email]>


Hi folks. I was debugging a problem with Moose and I realized that 2 different methods can have the same CompiledMethod. This was weird for me. I don't know if this is correct or not.

For example, evaluate:

(InstructionClient>>#methodReturnTop) = (InstructionClient>>#doDup)    ->  true

If you look at

CompiledMethod = aCompiledMethod 
    "Answer whether the receiver implements the same code as aCompiledMethod."

    | numLits |
    self == aCompiledMethod
        ifTrue: [ ^ true ].
    self class = aCompiledMethod class
        ifFalse: [ ^ false ].
    self size = aCompiledMethod size
        ifFalse: [ ^ false ].
    self header = aCompiledMethod header
        ifFalse: [ ^ false ].
    self initialPC to: self endPC do: [ :i |
        (self at: i) = (aCompiledMethod at: i)
            ifFalse: [ ^ false ] ].
    (self sameLiteralsAs: aCompiledMethod)
        ifFalse: [ ^ false ].
    self halt.
    ^ true


It is returning in the last ^ true.

What was weird for me is that (self sameLiteralsAs: aCompiledMethod)  returns true also. But if I print both literals:

self literals ->  {#methodReturnTop. (#InstructionClient->InstructionClient)}

aCompiledMethod literals ->  {#doDup. (#InstructionClient->InstructionClient)}

So...for me they are differnt, but sameLiteralsAs:  is returning true. Debugging that, it seems that in that method it is returning in the last ^ true.

So...any hints? is this the expected behavior ?  or it is a bug ?

Thanks

Mariano



_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

--
 Simon




_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev