Smalltalk › Pharo › Pharo Smalltalk Users

Question about alias in traits

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

6 messages Options

Hernan Wilkinson-3

Question about alias in traits

Hi,

I may have understood incorrectly the idea behind and alias when using traits, but I was expecting that when using an alias, the aliased method would be removed and all the sender of that message would be change to the new one (the alias), but I don't see that happening.

Here is some code:

Trait named: #SomeTrait

uses: {}

SomeTrait>>m1

^ 1

SomeTrait>>m2

self m1

Object subclass: #SomeClass

uses: SomeTrait @ { #m3 -> #m1}

It generates:

SomeClass>>m1

^ 1

SomeClass>>m2

self m1

SomeClass>>m3

But I was specting:

SomeClass>>m2

self m3

SomeClass>>m3

and the method SomeClass>>m1 removed.

Was my interpretation of Alias incorrect? or is the current implementation of alias incorrect? (maybe due to some implementation problems?)

Thanks!

Hernan.

Hernán Wilkinson
Agile Software Development, Teaching & Coaching

Phone: +54 - 011 - 6091 - 3125
Mobile: +54 - 911 - 4470 - 7207
email: [hidden email]
site: http://www.10Pines.com

Address: Alem 693, Floor 5 B, Buenos Aires, Argentina

Damien Cassou

Re: Question about alias in traits

On Thu, Jan 10, 2013 at 1:21 PM, Hernan Wilkinson
<[hidden email]> wrote:
> Was my interpretation of Alias incorrect? or is the current implementation
> of alias incorrect? (maybe due to some implementation problems?)

@ is an Alias operator, not a Rename operator. That means when you
alias { #m3 -> #m1 } you get a new name for the same method. So what
you see is the expected behavior, as defined in the paper. And the
direction of the arrow makes sense: #m3 -> #m1 is read as "m3 is a new
name for m1", whereas for a Rename operator, we would probably have
something like "#m1 -> #m3", i.e., m1 is renamed to m3. In a fully
dynamic language, a Rename operator would be quite hard to implement I
guess, IIRC Eiffel has a rename operator.

--
Damien Cassou
http://damiencassou.seasidehosting.st

"Success is the ability to go from one failure to another without
losing enthusiasm."
Winston Churchill

Stéphane Ducasse

Re: Question about alias in traits

In reply to this post by Hernan Wilkinson-3

>
> I may have understood incorrectly the idea behind and alias when using traits, but I was expecting that when using an alias, the aliased method would be removed and all the sender of that message would be change to the new one (the alias), but I don't see that happening.

You cannot do that in a dynamically typed language.
Why because you do not know how to scope the method renaming.

Aliasing is not renaming.

Stef

> Here is some code:
>
> Trait named: #SomeTrait
> uses: {}
>
> SomeTrait>>m1
> ^ 1
> SomeTrait>>m2
> self m1
>
> Object subclass: #SomeClass
> uses: SomeTrait @ { #m3 -> #m1}
>
> It generates:
> SomeClass>>m1
> ^ 1
> SomeClass>>m2
> self m1
> SomeClass>>m3
> ^1
>
> But I was specting:
> SomeClass>>m2
> self m3
> SomeClass>>m3
> ^1
> and the method SomeClass>>m1 removed.
>
> Was my interpretation of Alias incorrect? or is the current implementation of alias incorrect? (maybe due to some implementation problems?)
>
> Thanks!
> Hernan.
>
> --
> Hernán Wilkinson
> Agile Software Development, Teaching & Coaching
> Phone: +54 - 011 - 6091 - 3125
> Mobile: +54 - 911 - 4470 - 7207
> email: [hidden email]
> site: http://www.10Pines.com
> Address: Alem 693, Floor 5 B, Buenos Aires, Argentina

Stéphane Ducasse

Re: Question about alias in traits

In reply to this post by Damien Cassou

On Jan 10, 2013, at 2:17 PM, Damien Cassou wrote:

> On Thu, Jan 10, 2013 at 1:21 PM, Hernan Wilkinson
> <[hidden email]> wrote:
>> Was my interpretation of Alias incorrect? or is the current implementation
>> of alias incorrect? (maybe due to some implementation problems?)
>
> @ is an Alias operator, not a Rename operator. That means when you
> alias { #m3 -> #m1 } you get a new name for the same method. So what
> you see is the expected behavior, as defined in the paper. And the
> direction of the arrow makes sense: #m3 -> #m1 is read as "m3 is a new
> name for m1", whereas for a Rename operator, we would probably have
> something like "#m1 -> #m3", i.e., m1 is renamed to m3. In a fully
> dynamic language, a Rename operator would be quite hard to implement I
> guess, IIRC Eiffel has a rename operator.

Not quite hard. doing it fully and correctly is no possible because you should distinguish the methods
that should be renamed.

Hernan Wilkinson-3

Re: Question about alias in traits

thanks, I don't know why I thought that the alias was a rename too (maybe the name suggested me that...)

I think that having a "rename operator" that acts on the trait being composed in messages send to self would help in most of the cases, but as you say, it is not possible to take care of all the cases.

Thanks!

Hernan.

On Thu, Jan 10, 2013 at 4:17 PM, Stéphane Ducasse <[hidden email]> wrote:

On Jan 10, 2013, at 2:17 PM, Damien Cassou wrote:

> On Thu, Jan 10, 2013 at 1:21 PM, Hernan Wilkinson
> <[hidden email]> wrote:
>> Was my interpretation of Alias incorrect? or is the current implementation
>> of alias incorrect? (maybe due to some implementation problems?)
>
> @ is an Alias operator, not a Rename operator. That means when you
> alias { #m3 -> #m1 } you get a new name for the same method. So what
> you see is the expected behavior, as defined in the paper. And the
> direction of the arrow makes sense: #m3 -> #m1 is read as "m3 is a new
> name for m1", whereas for a Rename operator, we would probably have
> something like "#m1 -> #m3", i.e., m1 is renamed to m3. In a fully
> dynamic language, a Rename operator would be quite hard to implement I
> guess, IIRC Eiffel has a rename operator.

Not quite hard. doing it fully and correctly is no possible because you should distinguish the methods
that should be renamed.

Hernán Wilkinson
Agile Software Development, Teaching & Coaching

Phone: +54 - 011 - 6091 - 3125
Mobile: +54 - 911 - 4470 - 7207
email: [hidden email]
site: http://www.10Pines.com

Address: Alem 693, Floor 5 B, Buenos Aires, Argentina

Damien Cassou

Re: Question about alias in traits

On Mon, Jan 14, 2013 at 9:11 PM, Hernan Wilkinson
<[hidden email]> wrote:
> I think that having a "rename operator" that acts on the trait being
> composed in messages send to self would help in most of the cases, but as
> you say, it is not possible to take care of all the cases.

I guess it would be possible to rename all the self-sends and only them.

--
Damien Cassou
http://damiencassou.seasidehosting.st

"Success is the ability to go from one failure to another without
losing enthusiasm."
Winston Churchill