Perhaps an example of something you'd actually like to do in Squeak 
would help clarify the issue.

On 2/28/12 12:48 PM, Lawson English wrote:
> On 2/28/12 9:11 AM, Bob Arning wrote:
>> The first question is what it would mean in a Smalltalk environment. 
>> His tree-drawing example seems to be more of a complete "program" 
>> idiom. What would be the counterpart in Smalltalk? If you were 
>> looking at a method and made some changes, when and where would you 
>> expect to see the changes reflected? If the method were already being 
>> run frequently (say as a #step method), then you would see the 
>> changes in the World. Recompile time is negligible, of course, and 
>> one could think about using sliders to change numeric bits - if you 
>> think that improves the experience significantly. But what if the 
>> method is not being executed already, how do you say "I've made a 
>> change, now run it?" What's the context? And if you think about his 
>> tree example, that would likely be a number of methods in Smalltalk, 
>> so there's not a natural way to look at the source all at once.
>>
>
> Actually, it should be relatively trivial to provide a custom "folded 
> method" editing pane that is indexed by the methods in the method list 
> of the regular system browser. You can open individual methods or all 
> methods and scroll to them manually, or click on the method name and 
> scroll to it specifically.
>
> More importantly, recompiling during slider changes could also be done 
> reasonably easily, again, with a custom pane that evokes the slider 
> when you click on a number. The total "liveness" of executing changes 
> as they are typed might be more difficult to recreate. Perhaps typing 
> the enter key...
>
>> Of course, what does work a bit like that now is EToys. Also, 
>> inspectors and explorers can often be used to tweak values of running 
>> code, so the dichotomy of compile-and-run vs immediate-feedback is 
>> one that has already been largely bridged.
>>
> Sure.  This would be a somewhat specialized tool anyway. Most 
> applications don't need to record user events for playback, but again, 
> one could modify the standard browser to evoke the behavior if 
> specific classes and their subclasses were being edited.
>
> Lawson
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20120228/1f789459/attachment.htm