Q: senders of inlined #to:do:

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

Q: senders of inlined #to:do:

Klaus D. Witzel
I'm planning for an automated analysis of methods in the collection  
hierarchy which could be candidate for a possible #applyTo:from:to  
normalization (only if performance >= ).

The analysis will be done by running Goya[1] and coloring the respective  
callers and callees, driven by a dedicated set of TestCases.

But for inlined #to:do: and friends (#whileTrue:'s etc), another approach  
is needed. Is there perhaps something that someone can share, or do I have  
to write such a query with bytecode analysis tools. Any examples for the  
latter?

Thank you in advance for comments, examples, pointers, all appreciated.

/Klaus

[1] http://map1.squeakfoundation.org/packagebyname/Goya


Reply | Threaded
Open this post in threaded view
|

Re: Q: senders of inlined #to:do:

Bert Freudenberg
Klaus D. Witzel wrote:

> I'm planning for an automated analysis of methods in the collection
> hierarchy which could be candidate for a possible #applyTo:from:to
> normalization (only if performance >= ).
>
> The analysis will be done by running Goya[1] and coloring the respective
> callers and callees, driven by a dedicated set of TestCases.
>
> But for inlined #to:do: and friends (#whileTrue:'s etc), another
> approach is needed. Is there perhaps something that someone can share,
> or do I have to write such a query with bytecode analysis tools. Any
> examples for the latter?
>
> Thank you in advance for comments, examples, pointers, all appreciated.

Just have a look what the Decompiler does.

- Bert -


Reply | Threaded
Open this post in threaded view
|

Re: Q: senders of inlined #to:do:

Klaus D. Witzel
On Fri, 15 Sep 2006 21:51:39 +0200, Bert Freudenberg wrote:

> Klaus D. Witzel wrote:
...
>>  But for inlined #to:do: and friends (#whileTrue:'s etc), another  
>> approach is needed. Is there perhaps something that someone can share,  
>> or do I have to write such a query with bytecode analysis tools. Any  
>> examples for the latter?
>>  Thank you in advance for comments, examples, pointers, all appreciated.
>
> Just have a look what the Decompiler does.

Yes, that's what I wanted to avoid <sic/>

/Klaus

> - Bert -
>
>
>



Reply | Threaded
Open this post in threaded view
|

Re: Q: senders of inlined #to:do:

Philippe Marschall
In reply to this post by Klaus D. Witzel
2006/9/15, Klaus D. Witzel <[hidden email]>:

> I'm planning for an automated analysis of methods in the collection
> hierarchy which could be candidate for a possible #applyTo:from:to
> normalization (only if performance >= ).
>
> The analysis will be done by running Goya[1] and coloring the respective
> callers and callees, driven by a dedicated set of TestCases.
>
> But for inlined #to:do: and friends (#whileTrue:'s etc), another approach
> is needed. Is there perhaps something that someone can share, or do I have
> to write such a query with bytecode analysis tools. Any examples for the
> latter?
>
> Thank you in advance for comments, examples, pointers, all appreciated.

The AST package [1] adds #isInlineToDo to RBMessageNode. Yeah, it will
be slow but it's convenient.

Philippe

[1] http://www.squeaksource.com/AST/

Reply | Threaded
Open this post in threaded view
|

Re: Q: senders of inlined #to:do:

Klaus D. Witzel
Hi Philippe,

on Sat, 16 Sep 2006 12:36:13 +0200, you wrote:
> 2006/9/15, Klaus D. Witzel <[hidden email]>:
>> I'm planning for an automated analysis
... bla, bla, bla.

>>
>> But for inlined #to:do: and friends (#whileTrue:'s etc), another  
>> approach
>> is needed. Is there perhaps something that someone can share, or do I  
>> have
>> to write such a query with bytecode analysis tools. Any examples for the
>> latter?
>>
>> Thank you in advance for comments, examples, pointers, all appreciated.
>
> The AST package [1] adds #isInlineToDo to RBMessageNode. Yeah, it will
> be slow but it's convenient.

Great, that was what I was looking for, #isInlineToDo :) Thank you  
Philippe.

BTW, OT: how is primitive 71 going, did is disappear from the radar.

/Klaus

> Philippe
>
> [1] http://www.squeaksource.com/AST/
>
>



Reply | Threaded
Open this post in threaded view
|

Re: Q: senders of inlined #to:do:

Philippe Marschall
2006/9/16, Klaus D. Witzel <[hidden email]>:

> Hi Philippe,
>
> on Sat, 16 Sep 2006 12:36:13 +0200, you wrote:
> > 2006/9/15, Klaus D. Witzel <[hidden email]>:
> >> I'm planning for an automated analysis
> ... bla, bla, bla.
> >>
> >> But for inlined #to:do: and friends (#whileTrue:'s etc), another
> >> approach
> >> is needed. Is there perhaps something that someone can share, or do I
> >> have
> >> to write such a query with bytecode analysis tools. Any examples for the
> >> latter?
> >>
> >> Thank you in advance for comments, examples, pointers, all appreciated.
> >
> > The AST package [1] adds #isInlineToDo to RBMessageNode. Yeah, it will
> > be slow but it's convenient.
>
> Great, that was what I was looking for, #isInlineToDo :) Thank you
> Philippe.
>
> BTW, OT: how is primitive 71 going, did is disappear from the radar.

http://bugs.impara.de/view.php?id=4709

Philippe