Blocks and it's temp vars

> Hi there,
>
> working with callbacks in Seaside I've found this code don't
> evaluate properly in Squeak
>
> stuff := #(foo bar).
> results := OrderedCollection new.
> 1 to: 2 do:[:i|
> results add: [Transcript cr; nextPutAll: (stuff
> at:i);flush]].
> results do:[:e| e value].
>
> It complains about stuff array not having 3 elements (?!). Seems
> that it has some problem with block temps. I saw a Seaside method trying  
> to
> separate the temps of the block from the ones of the method and at that
> point the block temps was already malformed. The previous code is to  
> expose
> this problem.
>
> Tested on 3.9.final.7067 and 3.10 gamma.7159 (a couple of non Squeak
> smalltalks works as expected).
>
> I need this feature badly so if anyone can give me any clue or
> workaround would be great,
>
> thanks,
>
> Sebastian
>
>
>

>
> [copied from Seaside list, where this was asked and answered as  
> follows]
>
>>>>>> "Sebastian" == Sebastian Sastre <[hidden email]> writes:
>
> Sebastian> At first I was suspecting about the #fixCallbackTemps  
> but the block
> Sebastian> was already malformed when reach that point.  
> Investigating about this I'm
> Sebastian> surprised to found that it's related to Squeak being  
> unable to do properly
> Sebastian> this smalltalk code:
>
> Sebastian> stuff := #(foo bar).
> Sebastian> results := OrderedCollection new.
> Sebastian> 1 to: 2 do:[:i|
> Sebastian> results add: [Transcript cr; nextPutAll: (stuff at:
> Sebastian> i);flush]].
> Sebastian> results do:[:e| e value]
>
> I believe that this is a place where you need the fixTemps call on  
> the block.
>
> Did you try this as:
>
> 1 to: 2 do: [ :i |
>  results add: [
>    Transcript cr;
>       nextPutAll: (stuff at: Sebastian> i);flush
>       ] fixTemps
>  ].
>
> As in, notice the difference between:
>
> ((1 to: 3) collect: [:i | [i]]) collect: [:aBlock | aBlock value]
>
> => #(3 3 3)
>
> and
>
> ((1 to: 3) collect: [:i | [i] fixTemps]) collect: [:aBlock | aBlock  
> value]
>
> => #(1 2 3)
>
> --
> Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503  
> 777 0095
> <[hidden email]> <URL:http://www.stonehenge.com/merlyn/>
> Perl/Unix/security consulting, Technical writing, Comedy, etc. etc.
> See PerlTraining.Stonehenge.com for onsite and open-enrollment Perl  
> training!
>

> -----Mensaje original-----
> De: [hidden email]
> [mailto:[hidden email]] En
> nombre de Mathieu Suen
> Enviado el: Sábado, 08 de Diciembre de 2007 21:41
> Para: The general-purpose Squeak developers list
> Asunto: Re: Blocks and it's temp vars
>
> Hi,
>
> I don't think that #fixTemps should fix the problem:
>
> The to:do: block is create once but #fixTemps copy the home
> context This copy change the semantic scope of i.
>
> => Even with full closure your code would not work without #fixTemps.
>
> An other solution is to use curification:
>
> stuff := #(foo bar).
> results := OrderedCollection new.
> 1 to: 2 do:[:i|
> results add: ([:show | [Transcript cr;
> nextPutAll: show;flush]]
> value: (stuff at: i))
> ].
> results do:[:e| e value].
>
> This work with full closure but don't work with squeak block.
> So you still need a #fixTemps if you are not using full closure:
>
> stuff := #(foo bar).
> results := OrderedCollection new.
> 1 to: 2 do:[:i|
> results add: ([:show | [Transcript cr;
> nextPutAll: show;flush]] fixTemps value: (stuff at: i))
> ].
> results do:[:e| e value].
>
> Cheers
>
> On Dec 8, 2007, at 8:23 PM, Randal L. Schwartz wrote:
>
> >
> > [copied from Seaside list, where this was asked and answered as
> > follows]
> >
> >>>>>> "Sebastian" == Sebastian Sastre <[hidden email]> writes:
> >
> > Sebastian> At first I was suspecting about the #fixCallbackTemps
> > but the block
> > Sebastian> was already malformed when reach that point.  
> > Investigating about this I'm
> > Sebastian> surprised to found that it's related to Squeak being
> > unable to do properly
> > Sebastian> this smalltalk code:
> >
> > Sebastian> stuff := #(foo bar).
> > Sebastian> results := OrderedCollection new.
> > Sebastian> 1 to: 2 do:[:i|
> > Sebastian> results add: [Transcript cr;
> nextPutAll: (stuff at:
> > Sebastian> i);flush]].
> > Sebastian> results do:[:e| e value]
> >
> > I believe that this is a place where you need the fixTemps
> call on the
> > block.
> >
> > Did you try this as:
> >
> > 1 to: 2 do: [ :i |
> >  results add: [
> >    Transcript cr;
> >       nextPutAll: (stuff at: Sebastian> i);flush
> >       ] fixTemps
> >  ].
> >
> > As in, notice the difference between:
> >
> > ((1 to: 3) collect: [:i | [i]]) collect: [:aBlock | aBlock value]
> >
> > => #(3 3 3)
> >
> > and
> >
> > ((1 to: 3) collect: [:i | [i] fixTemps]) collect: [:aBlock | aBlock
> > value]
> >
> > => #(1 2 3)
> >
> > --
> > Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503
> > 777 0095
> > <[hidden email]> <URL:http://www.stonehenge.com/merlyn/>
> > Perl/Unix/security consulting, Technical writing, Comedy, etc. etc.
> > See PerlTraining.Stonehenge.com for onsite and open-enrollment Perl
> > training!
> >
>
> Mth
>
>
>
>