Re: OO & GNOME, summary & an action request
- From: Havoc Pennington <hp redhat com>
- To: Bill Haneman <bill haneman ireland sun com>
- Cc: seth eazel com, gnome-hackers gnome org, mjs eazel com
- Subject: Re: OO & GNOME, summary & an action request
- Date: 18 Feb 2001 10:35:24 -0500
Bill Haneman <bill haneman ireland sun com> writes:
>
> [among other things, about converting VCL to use native widgets]
>
> There are some engineering problems with using "native" widgets
> underneath a platform-independent layer. Some of these are
> *big* problems.
>
> One of the motivations for Java-Swing, as I understand it, was to
> get out from under this problem. I repeatedly have heard
> that making the Java AWT widgets behave consistently with
> various peer widget sets (MS, Motif, Mac) has been one of the
> Java group's enduring headaches. Wrapping the widgets so that they
> look the same/expose the same API may be straightforward,
> but trying to make their _behavior_ equivalent can be
> considerably more challenging.
>
> I don't mean to deny the advantages of "going native"
> ... but they come at a cost.
>
I would consider porting VCL to GTK+ to be a really bad idea, for
exactly this reason. AWT and similar toolkits show that it's a
failure.
Have posted longer mails on this topic in the past, but in essence the
Swing/GTK+/Qt approach of making the cross-platform layer be simply
low-level drawing, then having pluggable look and feel to emulate the
native kit, has proved the most feasible approach. The AbiWord
approach which amounts to "write each frontend separately" (with what
code sharing can be devised) is also pretty feasible when open source
teams are doing the work, since parallelized work is essentially free
in the open source world.
Havoc
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]