Friday 17 November 2000 21:11 wrote John Levon:
> On Fri, 17 Nov 2000, Matthias Ettrich wrote:
[snip]
> True, but then you lose the KDE2 added bonuses. I'm sure *you* are aware
> of that :)
>
> I'm not averse to a pure-Qt port, but personally I'm not interested in
> it ...
In KDE2, pure Qt and KDE are very similar. It's possible to support both with
a few #ifdefs. The biggest difference will currently be the KDE widget style
engine and the network transparency. This would also make a MS-Window port
trivial.
>
> > I don't believe you can use any of the code for KDE 1.1. It's simply too
> > different.
>
> OK, I will bow to your far greater knowledge here. I admit I still don't
> know what to do about KDE2 ...
Focus on Qt2 right now. The Qt Designer makes life soooooo much easier. When
you feel comfortable, extend the thing to use some nifty KDE2 features.
[snip]
> I believe this is mostly due to lack of time of the developers. I know
> personally I have very little time to spend on LyX ... and I am just a
> serf compared to Allan etc.
Another reason to use existing resources as wise as possible. Maximize fun,
minimize work. Object-orientation, code-reuse and utilizing existing
development frameworks help achieving that :)
[snip]
> > Sorry, that's more than lame. KDE is an application development
> > framework. Nobody has to run the window manager in order to run
> > applications written with that framework.
>
> I'm aware of this. Have you ever tried running a KDE app under gnome or
> vice versa with 40Mb of memory ? Maybe you aren't bothered, but I still
> care about those users (maybe because I am one). And you exaggerate the
According to the latest surveys, the vast majority of linux users has
machines with more than 300 Mhz and at least 64MB ram. Today, you cannot even
buy something that slow and small. The current standard for new systems is at
least 128MB ram.
Forcing the 2% potential users of LyX on machines with only 32MB of ram to
run either no desktop or KDE2 doesn't sound like a terrible constraint to me.
With such a machine, you should run KDE2 anyway if you want a decent web
browser :)
> inter-operability between KDE and Gnome. To pick an example, the component
> architectures are still not inter-operable AFAIK. You can bring up further
> examples much better than I could I am sure. To pretend that a
> Qt-targetted program is exactly equivalent to a Gnome-targetted one seems
> blatantly untrue to me.
I'm not aware of any plans of the LyX team to utilize either KPart or Bonobo
components, so this doesn't really apply. It might make sense to embed an
image or postscript KPart, that's true. Those will work just fine on a Gnome
desktop.
Apart from that: I'm pretty sure that we'll see a way to utilize kparts in
bonobo applications and vice versa before LyX2 comes out. Providing
interoperability like this is much less work than a toolkit-indepent text
processor backend.
To clearify my statement: I haven't said there is no work involved in making
a Qt application behave as closely to a native Gnome application as possible.
What I said is that it involves less work than full GUI independence and that
it's actually useful work that will benefit many more projects, not just LyX.
I sometimes have the impression that LyXers believe a KDE port would just
reach a tiny share of possible users. On Linux, that's just not true.
Approximately 70% of all users use KDE, many more have it installed to be
able to use at least some KDE applications. With the KDE2 release, I'm
confident we'll strengthen that position. There will be packages available
for Solaris and other commercial unixes (Sun even ships KDE with their
systems). Kylix (when finally released) will bring even more Qt based
applications to Linux. Many more big applications users use have been ported
from MS-Windows to Linux using Wine, that's not Gtk+ either. OpenOffice, like
I already said, uses its own toolkit and is even dubbed "core gnome
component".
In other words, a single-toolkit desktop will not happen anytime soon, but
we'll see a tighter integration of many toolkits including shared components.
RAM doesn't matter with 128MB, what users are concerned with is look and feel
and common themes.
How many potential users do you think a Qt-quality MS-Windows version can
reach? On my university, the missing MS-WIndows port was reason enough to not
offer LyX as standardsoftware (plain LaTeX was available, though). That was
years ago. Today, universities are even more windows-centric. LyX would
immediately become a standard package for technical or mathmatical papers.
But the best thing with a KDE2 port: thanks to ViaVoice, you'll soon get
decent speech control almost for free ;)
[snip]
Thanks for arguing, John. Sleep about it. Maybe you'll change mind and will
be able to convince Lars and Asger :)
Good night,
Matthias
