gzz-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gzz] (Lately)


From: Tuomas Lukka
Subject: Re: [Gzz] (Lately)
Date: Mon, 22 Jul 2002 10:45:51 +0300
User-agent: Mutt/1.4i

On Sun, Jul 21, 2002 at 09:33:52PM +0200, address@hidden wrote:
> 
> > > Restructured views: I've made View and FallbackView into one class
> > > (View) by introducing the ViewContext abstraction, allowing you
> > > to retrieve stuff like
> > > the cursor position and the dimension set-- that can be implemented
> > > reading that from the structure or from a Java object.
> > 
> > For the moment, let's not be too concerned with the structure: first let's
> > make a thing that works right to the user. Then we'll build the turtles
> > inside.
> 
> Hm. Are you suggesting to implement complete bindings in Fallback-- i.e.,
> not only a minimal set that allows us to get things to work, but all the bells
> and whistles (like real marking etc.)?

Yes: as I said in Fallback.java, we need to have a complete working system
without any metadata in the structure *first*, and *then* we can start working
on the bootstrapped system.

> > > My request:
> > 
> > > Tuomas, could you please factor out / rewrite the algorithm from the
> > > ExternalEditor class? We need it for Fallback, because we cannot edit
> > Jython scripts
> > > (for the real client) in one-line cell vobs. 
> > ...
> > > Anyway, do you think it would be a great hassle to implement this?
> > 
> > The problem is that the algorithm is really clumsy and isn't really good
> > to use: it doesn't find the most natural way to do things.
> 
> Well, a reasonably-well working algorithm is better than no algorithm at
> all. I'd suggest that we make an interface and make this the first
> implementation of the interface.

Ok.

> > Is jython-speed ok?
> 
> I don't know-- it's certainly not enough for an edithack-style window (that
> syncs on every keypress), but it could well be good enough for an external
> editor/dialog box interface, so let's have that for now. If we have big
> problems in the end, we can still make a Java implementation of the interface.

Yes, and jython should make the prototyping quick anyway.

I'll try to look in to this within a few days. If I forget, nudge me again.

        Tuomas



reply via email to

[Prev in Thread] Current Thread [Next in Thread]