htoolkit-users

Re: [htoolkit-users] Is GTK2 required? and glade question

[Krasimir A. <ka2...@ya...>]

--- Nick Name <nic...@in...> wrote:
> I was in doubt because due to recent develpments on
> the GUI ML I thought 
> that every people seeking to write a high-level GUI
> had given up; this 
> is the only reason why I publicily invited you to an
> heartbeat signal.

Don't worry! The HToolkit is a live and will continue
to evolve. I hope that it will be possible to release
the next version very soon - probably this or next
week.

> For the proposal about attributes, there are non-h98
> solution wich will 
> allow a search on a list of types (probably you
> already know, since 
> many people appear to use this to model multiple
> inheritance in ffi's). 

I am not very worried about Haskell 98 comptibility
but some people will do. If you have simple idea which
uses only features proposed for Haskell 2. I am 
inclined to incorporate it in the GIO design.

> Apart from that, I still think that the direction of
> htoolkit was the 
> right one, and if you don't mind, when I will have
> some time, I can 
> post some possible addition to the library on the
> mailing list, but BTW 
> what's your position regarding to complex state
> handling (e.g. ability 
> to change the monad from IO to something else, or
> ability to watch for 
> state changes in a "Var" by obtaining an event
> handler from an 
> attribute), so that I avoid bothering you with such
> things in case you 
> are absolutely against them, and wasting my time too
> ? :) 

    I don't like to change the monad from IO to
something else. The ObjectIO library is implemented
with two monad types: GUI and Draw. Each time when I
want to run some IO action I need to lift the action
with 'liftIO'. I think that the absence of specific
monads in HToolkit is an advantage.
    The ability to watch the state change (i.e. the
change of some attribute value) is another thing. This
can make the library design more clear but I have
never thought about it.

Bye,
Krasimir

    P.S. As you can see the definition of Process type
in CVS is changed as in your proposal.

__________________________________
Do you Yahoo!?
The New Yahoo! Shopping - with improved product search
http://shopping.yahoo.com

Re: [htoolkit-users] Is GTK2 required? and glade question

[Nick N. <nic...@in...>]

Alle 12:42, luned=EC 20 ottobre 2003, Krasimir Angelov ha scritto:

> =A0 =A0 I don't like to change the monad from IO to
> something else. The ObjectIO library is implemented
> with two monad types: GUI and Draw. Each time when I
> want to run some IO action I need to lift the action
> with 'liftIO'. I think that the absence of specific
> monads in HToolkit is an advantage.

Oh, yes, that's a design decision wich one can agree or disagree, and I=20
find it more comfortable, too. What I was talking about is of less=20
interest in the project, but is about attributes in general (since=20
someone mentioned somewhere the possibility to include the attribute=20
library in the hslibs, and I think that would be a good thing): the=20
ability for the user of the Attr library to change the IO monad to=20
something else - it would not affect htoolkit since it has ffi calls=20
and so it would choose the IO monad.

> =A0 =A0 The ability to watch the state change (i.e. the
> change of some attribute value) is another thing. This
> can make the library design more clear but I have
> never thought about it.
>
> Bye,
> Krasimir
>
> =A0 =A0 P.S. As you can see the definition of Process type
> in CVS is changed as in your proposal.

Thanks ;) I still have to take your gnome2.4 route, though, but I think=20
this will happen really soon.

V.