Re: Please do not use Qt (fwd)
On Tue, 26 Nov 1996, Craig Sanders wrote:
Thanks Lars and Craig for your great inputs on the subject.
This will be my last follow up to this subject on the mailing lists.
I do think that the discussion should not continue on these ml's.
If you want to continue you are welcome to join the kde ml, or mail me
privately.
> MANY people (including many of the debian development team) will just
> plain refuse to use, recommend, or support kde IF it is based on a
> non-freeware library like Qt. This is true in the debian community, the
> linux community, and also in the rest of the unix using community.
I do understand this point fully.
A gpled gui lib is to be prefered no way!
> This being the case, the only noticeable result of kde will be further
> fragmentation of the unix/X GUI "standards".
I do still believe that it is a step in the right direction.
I would like to achieve something similar to the license use by Alladin
for ghostscript.
> IMO, the goal of kde is well-meaning but severely flawed. For it to be
> THE killer GUI for X it has to:
>
> a) be truly free in the Debian or FSF or BSD or Artistic license sense
> of the word "free". Qt automatically disqualifies kde from this.
Yes you are right. Qt can be used without limits for any gnu/debian
projects, but if people want to use their skills they aquired by
developing for GNU/FSF/Debian they have to pay for a commercial license
in order to be allowed to distribute their non gpled, non source
payware.
Another severe limitation is the fact that you are not allowed to
distribute altered versions of the library.
GNU people want to fix the bugs they do discover. They tend to send the
patches to the main maintainers.
The Troll people tend to include reasonable patches but do not give any
guaranty.
Due to the fact that the internal structures of Qt are open it is more
easy to work around some limitations/problems. Also the fact that Qt is
object oriented give the programmers some relief.
The kde project tries to encapsulate moste of the functionality by
kclasses.
In the beginning most of these classes are wrappers around the original
qclasses.
Some stuff has already been completely replaced.
E.g. there is now a kfileselector which natively understands URLs.
It is beeing used by the kde apps including the kde filemanager.
So any app can laod/save to http: file: and ftp: ressources...
If the kde people get fed up by Qt they have mainly to change their
kdelib(which provides the extra functionality and the wrappers).
> b) be *at least* as functional & complete as Motif, Tk, Athena, and fvwm.
Please have a look at kwm (the kde windowmanager) and especially at
kfm(the very nice looking AND feeling filemanager)
Qt got quite far until now. It is actively developed.
The functionality increases at rapid speed.
> c) support or be mostly compatible with existing standards, allowing
> for trivially easy ports from old libraries to kde.
It is always a diffcult task to port X stuff to different GUIs.
> d) be fast
First test seem to show that Qt is surprisingly fast.
It uses shared libs widly. It is definetely much faster than tk.
> e) be significantly better and easier to program in that any of the
> alternatives.
Oh this seems to be really the case. Please have a look at the source
code of some of the programs
ftp://fiwi02.wiwi.uni-tuebingen.de/pub/kde
> f) not get in the way of (or make arbitrarily difficult) the ability
> to install, use, and develop for non-kde environments. I sure as
> hell am not going to install kde if it means I have to give up
> fvwm95 or any of my tk/motif/athena based apps.
YOu definetely do not have to give up anything.
But you may prefer to have a common look & feel on your desktop.
This might be the reason for some people to change their their acustomed
apps to kapps.
This is especially true for X11 newbies.
But you can without hassle share kde and non kde apps on the same desktop.
Cut&Paste via X11 & Co. will still work...
> g) have an enormous library of available applications.
kde is still very young (beginning of october)
There is already:
a calculator kcalc (far more functional than xcalc!!)
a cdrom player kscd (based on workman, about 6hours work)
a clock kclock (one of the very first apps :-)
a filemanager kfm (fully functional, BeBox inspired, fully URL!!)
a minesweeper kmines (better than the Win Version ;-))
a solitaire kpat (another Win inspired game)
a tetris clone ktetris
a corel clone colia (soon to be integrated with LyX...)
a free xv kview (no manipulation so far, just viewing of
graphics)
a help system khelp
a xterm kvt
a Sound Mixer kMix
a kde library libkde (developed from Kalle Dalheimer (Staroffice)
(This lib handles the Xresources like colors,
common code to the apps,...)
a window manager kwm
a command line interface cli (simialar to the one of Nextstep)
Please have a look at the outdated but nice screenshot from Nov, 10
ftp://fiwi02.wiwi.uni-tuebingen.de/pub/kde/demos/kdescreenshot1.jpg.gz
> In other words, it has to be a backwards-compatible upgrade rather than
> a revolutionary, new, and incompatible change. I suspect that this is
> close to impossible given the existing fragmentation and incompatibility
> between the various x gui libraries.
You can still run any X11 app on a kde desktop! There are no limitaitons.
Just don't expect them to be configurable via the kdelib or to be able
to use an URL from their fileselector....)
It is more like gv/ghostview :-)
> If these points are not met by kde, then kde will go the way of all
> other attempts to unify X GUI programming under one glorious scheme: it
> will be just another one of the many available options. (this, btw, is
> both the beauty AND the bane of X - it's wonderfully customisable and
> configurable...the trouble is that you MUST customise & configure it)
This is the point which led to kde. The look and feel shall be
orthogonal. (therefor the URL approach). It shall be easily configurable
to anyones taste. (Keybindings,colors, geometries... are configurable in
kdelib..)
> in fact, even if kde DOES satisfy all of the above points it is extremely
> likely that that will be the case anyway. Just because something new and
> wonderful exists doesn't mean that you should throw away stuff that is old
> and wonderful or just old and useful.
You are perfectly right here. There is so much very good stuff out there
which can hardly be met.
Just imagine (X)emacs or the powerfull bash.
But why not configure your emacs via a central GUI kdecontrol?
Why not launch gimp from the kfm?
Why not paste your favorite Linux ftp URL into kless?
> There is a lot of time and effort and programming hours invested in
> motif and athena (and other x library) based applications. The kde
> people may want to think of these as "legacy applications" but to the
> rest of the world, they are still very much alive and kicking.
Right, but they tend to be hard to program. They are not free (Motif) for
developing gpl apps..)
> the existence of this "legacy software" is one of the things that makes
> unix and linux so attractive to so many people. free operating system,
> free applications...all you have to do is download them off the net
> and compile (or install a pre-compiled debian package). Windoze users
> have their expensive off-the-shelf applications library. We have our
> compile-it-yourself off-the-net applications library. (i know what i
> prefer :-)
Shure I got your point.
How about having a look at the kde apps and mail me privately (or the kde
lists) about your experiences?
Your input is very much welcome! A lot of thingd still have to be
discussed.
The kde is still VERY young and evolving.
Yours,
-- martin
P.S. Please NO follow ups to this message on the debian lists any more.
Originally I started this thread to ask for an GUI based alternativ
to dselect, which can be more easily used by novices.
I do have the impression that the discussion now just went to much
general...
Personally I would also be happy about an easily used and fully
functional xdeslect based on anything else than Qt.
// Martin Konold, Muenzgasse 7, 72070 Tuebingen, Germany //
// Email: konold@tat.physik.uni-tuebingen.de //
Linux - because reboots are for hardware upgrades
-- Edwin Huffstutler <ehuff@sedona.intel.com> --
Just go ahead and write your own multitasking multiuser os !
Worked for me all the times.
-- Linus Torvalds --
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-REQUEST@lists.debian.org . Trouble? e-mail to Bruce@Pixar.com
Reply to: