RE: [boost] Re: Re: Re: GUI/GDI template library

From: Brock Peabody (brock.peabody_at_[hidden])
Date: 2003-08-05 13:39:49

> -----Original Message-----
> From: boost-bounces_at_[hidden]
[mailto:boost-bounces_at_[hidden]]
> On Behalf Of Bohdan
> Sent: Tuesday, August 05, 2003 12:08 PM
> To: boost_at_[hidden]
> Subject: [boost] Re: Re: Re: GUI/GDI template library
> 
> 
> "Brock Peabody" <brock.peabody_at_[hidden]> wrote in message
> news:000b01c35b66$cee4d180$580a10ac_at_npci.com...
> > > On Behalf Of Bronek Kozicki:
> > > Please, do NOT use MFC. Just native Win32 API as a starting point.
MFC
> > > is 1) not available on mingw 2) big and full of traps.
> >
> > If we can do Win32 I guess there isn't any point in doing MFC.  I
will
> > not miss it much :)
> 
> Honestly ... i can't see any  reason why you may want MFC.
> Rather contrary, as Bronek said your library can't be used with
> other compilers. Besides, plenty of MFC code causes many people
> don't want to look at your gui library trial at all. They don't want
to
> waste time.

I don't know that Win32 will be much prettier to look at, but I agree
that using MFC is not decessary.

> IMHO, it would be nice to concentrate on:
>         1. design
>         2. interface + examples
>         3. Portability. IMHO, initially it should be implemented at
least
>             for two platforms (POSIX and win32).
>             Otherwise, when porting to other environment, you may be
>             forced to redesign your library completely.
>             With this two platform support you may really draw
>             community attention and ... i'm 99% sure ... obtain a lot
of
>             help in discussing and developing.

Agreed.  There's no such thing as a POSIX GUI standard is there?

>         4. It is not work for 1-2 persons. I think it is enough
serious
>             project even for 10 developers.
>             What about ? :
>                     a. opening new project ( on sourceforge ?).
>                     b. trying to find new develpers
>                     c. preparing to spend next few years for it :o)
>                ?

I think we need more people involved just to make the design better, but
one person with a lot of time could implement it (for one platform, at
least).

>         5. Implement at least one serious application.
>             If i don't mind boost needs kind of installation
>             which can track library dependencies and install
>             only required libs.

Agreed.

>         6. Do you really think that lisp-like planty of parentheses
>             will be very convenient for coding ?
>             If you really wan't to invent kind of "dialog language"
>             ( i really don't understand why one may need it ?),
>             wouldn't it be better  to use operators like spirit or BLL
do.

Although I do love Lisp, I would never argue that it is easy to read,
and it is important that the syntax for our library is.

I'm definitely open to improvements in the syntax.  Thinking of it as a
dialog language might make it easier to discuss.

With Spirit, the operators made sense because they replicated the syntax
of the problem domain.  I think operators will have their place in this
library (especially if they help reduce reliance on macros), but I don't
think we will want to use them as shortcuts for domain specific terms
like 'edit_box' or 'button' or 'row'.

Brock