new version of MophicWrappers

Daniel Vainsencher danielv at netvision.net.il
Wed Oct 7 21:37:25 UTC 1998


Gerardo Richarte wrote:

> [looking for more direct menu invocation]

>   You suggested using shift, cmd, etc. as the key to different levels

> of interaction, but what can be done in a Mac, for example, where the
> mouse has only one button, and this keys are already used? (to turn
> halos on, for example)

Oops. How Wintel-centric of me, I forgot about that. In truth, I don't know.

> >    * When you drag around an ellipse with satellites you sometimes
> >      leave traces of the satellite around.
> >
>     Is this with your version of Satellites or with our? This is usually
> produced with a submorph being drawn outsinde owner's bounds.

Either version. I see... off the cuff it seems like there should be a
general solution for all Morphs with errant animated submorphs, but so long
as it doesn't, I guess we'll just have to make the satellites move closer.


> >    * I've looked at the way the MorphicWrapper class is built, and I
> >      see a few interesting things. There are a lot of delegation
> >      methods, some of which are very simple. Why have an "objectClass"
> >      (for example) method?
> >
>     Well this message is needed bye ClassWrapper (yeah!): Leandro
> Caniglia and Valeria Murgia are working on this (otgether with
> MethodWrapper) they said: "We are, almost, no longer using Browsers"
> This will be realeased with MorphicWrappers next time, but I can ask
> them for a beta version if somebody wants it.

I'd like to know more about the intended design (for example, so I'd
understand why this is needed - it looks very inelegant from my limited
perspective). Is there any more info available, old mail, scanned used
napkins with class relation diagrams and such? I'd be glad to read them, I'd
even write them up on the Swiki, if you wish.

I want a beta of CW!

> >    *  The selector list is created on the fly, whenever something is
> >      dropped on a MW. Doesn't seem to hurt responsiveness on my
> >      machine, but it's something to consider.
> >
>     Well, this is a must, as a method can be created anytime, and you
> don't know when this happens... unless you get insde Class and Behaviour
> classes, what I think we don't want to do...

Maybe some sort of notification service could be arranged, but it's beyond
me ...
And probably not worth it for this project. Though I'm thinking that it
would be useful for updating Browsers of all kinds. It's rather annoying to
edit out-of-date views of classes. Maybe this'll attract the attention of
bigger fish ...

> > * An optional stack that, if it exists, is the default target for
> > results. (I'd put it down the right
> > edge of the world).
>
> > * A graphic representation of variables known in the WrapperWorld. (A
> > list down the left side? maybe boxes with values or arrows to them?)
>
>     Why not a Graph? this is already implemented, and will be included
> in next version...

I'd have to see it to know what you mean. Hint, hint. ;-)

> > * A recorder that once turned on notices what I'm doing and creates an
> > equivalent code snippet, or
> > some other representation of the process that can be then run step by
> > step.
>     A lot of things need to be solved for this, what about arguments?
> and arguments' messages?

I agree that this IS in the "blue plane" for this project, and might require
a lot of thought to implement...
I'll see if I can get back to you with a more concrete idea.

>     I didn't get the mails you sent directly to me, don't know why...
> (what address are you using?)

I'm sending to Gerardo Richarte <gera at core-sdi.com>
Is there another address?
Maybe you've a filter that doesn't like me?





More information about the Squeak-dev mailing list