[ltp] Debian on a 380D

[Tod Harter]

On Tuesday 23 October 2001 14:12, you wrote:
> On Tue, 23 Oct 2001, Tod Harter wrote:
> > From: Tod Harter <tharter@rhombus.net>
> > Subject: Re: [ltp] Debian on a 380D
> > X-Mailer: KMail [version 1.2]
> >
> > I was just really disturbed by Debian's attitude towards certain
> > software, namely KDE. It didn't leave me with a good feeling that they
> > were putting
>
> Note that this attitude is already changed. And I rather disappointed by
> this change. In my opinion KDE and Gnome are too much Windows-like
> to be productive.

Well, they are somewhat "windows-like" but then again I don't see any other 
UI's out there that are demonstrably superior in design. While Apple loves to 
tout their desktop I know one of the big Apple magazines did some objective 
studies (putting users in front of windows and MacOS and having them perform 
various routine operations like installing software, connecting to an ISP, 
etc) and they had to admit that users had no better or worse luck on average 
with either system.  I know there are projects like Berlin and some others 
that are aimed at creating better GUIs, but they are far from useably 
complete, and even when they ARE complete they will have to pass muster in 
objective testing.

>
> Gnome is partially exused by pure openness, which gives theoretical
> possibility to fix broken things (alas, only theoretical - many-many
> man years are needed to fix all the design flaws in gtk, and first they
> have to be admitted tto be design flaws). Kde with dependencyu  on
> proprietary GUI library is worse. Especially becouse it seems that
> some KDE developers seems to misunderstand Qt i18n (which is unneccessary
> complicated).

Well, Qt is now as open as anything, and the whole issue with the GPL vs the 
QPL and KDE's stand on the linking issue was never resolved anyway. RS had a 
big hissy fit about the QPL and came up with some VERY arcane logic as to why 
you couldn't link QPL and LGPL code together. Frankly I thought his logic was 
entirely flawed and it was rather disheartening to see the Debian people 
blindly following RS's lead. The guy is a flaming lunatic if you ask me. Plus 
its not necessary to worry about "viral licensing". ANY library built by one 
group of people (so having one copyright) can be taken out of the public 
domain any old time by its owners. I mean Trolltech could just say tomorrow 
"sorry, Qt 3.x is no longer GPL from here on out". Same is true of a lot of 
code. Thats just life. 
>
> > user's real needs ahead of some philosophical/political agenda. The only
> > other thing is Debian seems to always be way behind in updating packages.
> > Its a two-edged sword, but that is one great thing about Mandrake, esp
> > for us KDE users out there, they are always close to the current edge of
> > development.
>
> It is rather bad that software is in such a poor state now that user
> has to be on the current edge of development just to get away from
> show-stopper bugs. I'd rather prefer situation when I can use same
> good old software for years until my requirements change and I would
> really need to learn something new. Alas, it is virtually impossible.

YES. I agree. Personally I think what we need is maybe for the Linux 
Standards Project or some sort of "Linux Core Team" to take charge of 
defining the basic standard functionality of a "vanilla" Linux system. IE, 
create a base which is generally workeable and provides the core capabilities 
required of an OS for it to be successful. That would mean kernel, file 
systems, common subsystems like HTTP, FTP, SMTP, system management interface, 
network configuration, startup, and a "standard desktop" environment. Certain 
versions of various subsystems could be selected as "base level" and every 
developer could then depend on things  working a certain way. Plus the base 
level would provide a basis for bug fixes. Desktops are especially 
problematical in that they are very hard to upgrade and have tons of 
dependencies. NOTHING made me more irritated than installing a new Mandrake 
and finding that Kmail was totally broken. The Kmail developers only answer 
is "well, we're 5 revs past that in development now, go get the newest CVS 
archive of KDE, build it, and THEN report bugs, otherwise go away". I can 
understand that, but its ridiculous! I need a Kmail that works, and the one I 
have is the newest one available in any stable release of a Linux distro. 
Somehow we HAVE to solve this mess if we expect to put Linux in the hands of 
regular people. 
>
> Anyway, if you want to be at the current edge, Debian have testing and
> even unstable for you. I'd say that later is not for users, but for
> very adventurous developers, but former is good enough for any
> non-critical system.
>
> > I think ALSA is the way to go for sound. It is quite superior to OSS
> > quality wise, and I couldn't get OSS to reliably support my 380D's sound
> > chip anyhow. It would just randomly stop working in KDE. Not sure if it
> > was really an OSS problem, but ALSA fixed it...
>
> It may be. And Debian excellent module-building system is quite
> good for things like tpctl or alsa which have to be in your running
> kernel, but do not fit in the standard kernel source tree.

Yeah, well in the case of ALSA its pretty easy to add it. If you had OSS 
installed properly you can replace it with ALSA with no kernel rebuilds (at 
least under 2.4.x kernels, 2.2.x is a different situation). 

----- The Linux ThinkPad mailing list -----
The linux-thinkpad mailing list home page is at:
http://www.bm-soft.com/~bm/tp_mailing.html