Morphic thoughts and Questions
Alan Kay
Alan.Kay at disney.com
Sat Jul 8 07:59:31 UTC 2000
The plan is to eventually drive the scripting tiles UI from standard
methods. This will cut the bloat for transport, and, as Lex points
out, eventually the total Squeak code for scripting support will
settle down and be amenable to optimization for speed and space ...
Cheers,
Alan
At 11:38 AM +0300 7/7/00, Lex Spoon wrote:
>[...scripting bloats morphic...]
>
>Ah, scripting certainly does effect the amount of *code*. It's hard to
>fix this, though, until the design of the scripting stuff settles
>down....
>
>
>
>
>> >They would make a
>> >terrific basis, I believe, for a full system to do business GUI's. We
>> >just need (a) more of them and (b) layout morphs.
>>
>> I partly disagree. For an easy to use UI, I'd request
>>
>> (c) true models, that is objects which send notifications if one change
>> them and so the UI automatically updates itself.
>>
>> The need to call these messages by hand right now is annoying. As well as
>> the fact that a list always gets assigned a completely new list or a text a
>> new text object, even if one could modify the existing pseudo-model.
>>
>
>I don't follow how the current pluggable morphs are not based on true
>models; they are classic model-view as done back in Smalltalk-80 and
>probably before. Are you requesting things like ValueHolder and
>StringHolder, which automatically call #changed: when their contents are
>modified? I would still call that a change in volume rather than quality--
>we need more stuff, but a lot of things are done at least tolerably well....
>
>
>
>-Lex
More information about the Squeak-dev
mailing list
|