Morph browser discussion

Adriano apeluso at peg.it
Mon Nov 22 15:47:43 UTC 1999


At 21:20 +0100 21-11-1999, Stefan Rieken wrote:

>Alan Kay wrote:
>>
>> constructed, but there is still not a real Morph browser -- moreover, I
>> think all would agree that there could be much more of a theory for larger
>> useful tools, and then we could build a very nice
>> browser/inspector/explorer/debugger that would make crystal clear how
>> things are organized and used.
>>      This is now -- IMO -- a very pressing need. Please contribute ideas.
>>
>
>Let me give some feedback on this by telling my experience. I really
>have problems with having too little control on the class instances
>(objects) themselves, for example:

[big snip]

>Just see me in this context as the kid that should be able to play with
>Smalltalk, and these comments as things that this kid doesn't yet quite
>manage. I hope they are helpful,
>
>Greets,
>
>Stefan
>
>P.S.: Marcus Denker wrote: "It should be somehow more explicit: We need
>some kind of "meta-instance" that discribes how a particular instance
>was put together.". Let me conclude my words by saying: we need to get
>control over the class of an object, as well as over the objects
>belonging to a class.

Also, in a later message, Stefan wrote:

>Bob Arning wrote:
>>
>> Granted that these may not be obvious to the new user, but they are not
>>far off the beaten path. Do you see a way that
>> - they could be useful to you?
>> - they could be made more accessible?
>>
>> Cheers,
>> Bob
>
>Well, YES. Paff. What you wrote was all new to me.

....

>Luckily my ignorance has also been the occasion for thoughts on
>"overviewing Squeak". I guess that what Marcus calls relationship, the
>OO bookstore calls "aggregation".
....

>I think this (new-to-me) Explorer utility is a good starting point for
>creating a system overview. Although the current implementation seems a
>little heavy-duty on first view (much numbers, not much words), it is
>actually a good tool for an "end user". It shows relations in a quite
>good manner (submorphs and instance variables). When this interface is
>combined with that of Whisker

Er... what's Whisker?

> (good for finding class methods IMHO), you
>could get an interface that tells of each object _individually_  its
>current methods and variables.
>
>This point of view on objects reminds me of the Self discussion we had
>lately.

Yes, because of the habit of showing every object as an outline, in Self
you'd have been able to access the single istances in a more intuitive and
quick way.

As command line tools hide their power to beginners, so does Squeak, sometimes.

A GUI describing VISUALLY the concepts would make the learning curve much
smoother, I think.

An outline could give you a bird's eye view over the whole matter, even on
the aggregations between objects, it could allow you to go deeper and
deeper (at any level you want) AND it would give you the ability to point
an element and say "I want to modify THIS" (as Markus Denker noted) without
need to tinker under the hood (having no docs  ;-)   )

The only lack of outliners I see is that outlines are not suitable to show
the dinamicity of the things (Bob Arning noted this and suggested some
ideas).

The Self outlines even have slots with pointers to super objects (super
classes)

Maybe Golgi could be endowed with some machinery to detect objects and show
them as outlines?

Also the Steve Wart's idea of a sinergy between MorphBooks and Golgi
outlines is terribly attracting.

And an implementation of Fabrik would be lovely, too (see the last message
from Dan Ingalls "Re: Let's Get Bizzy")

Well, as correctly Alan Key noted, the lack of documentation was explicitly
choosen by the few guys who built all this because of the fear to be
overwhelmed by questions and claims before being ready for it.

I understand it. But having a little more accessible system would be lovely :-)

Well, now I'm terribly eager to see what this "net Squeak" is going to be !!

Oh, sorry for being so long :(

Just my thoughts
Bye
Adriano Peluso

P.S. Also, this matter of being able to deal directly with the single
istances could be considered important by the direct programming folks (see
the Gliphic web site)





More information about the Squeak-dev mailing list