Future was Re: Contributors v. Writers v. Editors

[Dan Smith]

Mr Carlson

On Thu, Jun 19, 2008 at 10:07 PM, Chris Carlson <cwcarlson at cox.net> wrote:
> Mr. Smith,
>
> I sure hope what you say isn't true.  It may seem like it, but from what
A few months back the Fedora project officially announced it was
bowing out of the Desktop development game :(  I was saddened but not
surprised and that seemed to be the general concencus in all the blogs
I read about the announcement.

> I've seen comparing Fedora 9 to RH7.3 (where I used to be), the desktop has
I bought RH5 and actually got useful work done on it. By RH 6.x I was
switched entirely to LInux and using RH both at work and at home. I
agree the desktop has improved by light years since then. There was no
way I was converting a novice computer user to Linux using the RH 7.x
desktop. By FC3 I was not only converting novice computer users, I was
helping out and answering questions from people to whom the words
TCP/IP sounded like a jet aircraft LOL. I met a 12 yr old kid who
installed Fedora without any help from anybody. Only needed help with
a few things, mostly being pointed to the right repositories for
example. Fedora has greatly advanced the Linux desktop and for years
was the leader in Desktop innovation for the desktop.

> improved considerably.  Also, since this project is a collaborative effort,
> it requires people to focus on the important issues.  If you feel that the
> desktop is the top issue, then I'd like to suggest you help make it so.
> Based on what I've been seeing on the fedora-list, there are gobs and gobs
> of bugs that need to be fixed that have nothing to do with the desktop (your
> issues with sound are an example).

Ah but sound is part of the desktop. There are lots of fundemental
pieces of work that need to be done.  A better example is the ability
to change monitors without reinstalling the OS. That has to be one of
the biggest annoynances, especially for single computer owners. No way
to SSH in and fix things if you have no other machine to SSH in from
and to be honest it's probably easier to reinstall than to get X to
recognize a new monitor.  Sound on laptops is especially flaky under
Fedora. Scanners are poorly supported on Fedora as well and from what
I've read Myth TV can be a bear to get working properly under Fedora.
Jack audio and Fedora just plain do not get along. The list is long on
those areas.

I think we have divergent deffinitions of desktop. The general use is
an overall desktop experience, not the Gnome vrs KDE vrs any other
window manager.  Key things I personally feel would be a big help in
Linux adoption.

Compatability with Windoze specific formats - Here huge progress has
been made. Open Office has especially brought the Linux desktop up to
speed on this aspect but lots of other projects have helped a great
deal as well.

Software installation -  Here the RPM and deb GUI package managers
have turned this from a liability into a major positive for most
instlalations. Tarballs scare novice users. Dependency hell that can
often accompany a tarball installation scare even long time Linux
users LOL. The RPM repositories are still the largest but the .deb
repositories are catching up. What is disturbing is how many packages
exist only on one or the other and how many are woefully antique on
one or the other. Wesnoth for example. No way you install Wesnoth on
Fedora from an RPM if you want anything close to the current game or
even a working version of Wesnoth. Yet Wesnoth is probably the most
advanced and stable game freely availible for Linux. The .deb version
is current and installs a full featured and stable version.  In both
cases the GUI package managers make Windoze installations seem
primitive and a pain, where as a few short years ago installations was
one of the biggest negatives I got back from new Linux users.


> Could it be that individuals are focusing on their pet projects rather than
> the desktop?  My guess is the average individual working on Fedora is
> Linux-savvy and doesn't care about the desktop for themselves, so they put
> their efforts into the bugs or features that interest them.

Difficult to ask somebody to work on something that does not interest
them, especially somebody doing the work for free. It's more an issue
of focus and culture. Fedora is populated by old school Nix people,
many of whom never really ran windoze for a primary environment. Look
back at the tools thread in this group and a surprising number of VI
and Emacs users popped up.  Few windoze converts would ever touch VI
or Emacs. Not since much more user friendly editors have appeared.
Though an easy to use console based text editor with standard keyboard
shortcuts would be a huge help for Linux adoption. Where I lose many
people is trying to walk them through a VI session if something goes
wrong with X and they have to edit configuration files or SSH in to a
machine which doesn't have X installed. They love the idea of remote
administration but VI sends them screaming into the night.  The
learning curve was too steep for me to ever mess with it. One of the
reasons I love Linux is that I spend very little time doing sys admin
tasks and almost all my time doing stuff WITH not TOO my computer. No
defragging, no virus scanning, no constant roll backs, no registry
hacking.  I install it, I do updates, every once in a while I add or
change something. I run occasional rootkit checks and if my machine
behaves oddly in any way I am going through the logs with a fine tooth
comb. Windoze is constant work to keep running. Learning VI and or
Emacs is memorization of tons of keyboard shortcuts. Sure it might be
handy to have but when even the most basic functions are so different
than the CDE shortcuts I've used for years it just doesn't make the
time expenditure all that worthwhile.

The VI/C++ old school nix culture who is quite happy with primitive
interfaces and who long ago spent the many hours learning and
memorizing the many exotic and often complex user interfaces of Unix
is the core culture of Fedora. You see it here in the docs project as
my efforts to go into many areas were thwarted as they were considered
"extra" when to a novice user, especially a windoze convert these
areas are core functionality.  Remember most computer users wouldn't
know a primary partition from a whole in the ground and have been
conditioned by years of M$ propoganda into thinking they shouldn't
have too.

So the languages supported, the documentation written are old school
nix. Joe is probably as close to an easy console text editor you'll
find and the UI for Joe is horrible for even someone like me who have
used dozens of DOS and Linux text based editors. I wind up limping
through enough VI to make simple changes or SSHing the file over to a
machine with X for any major changes rather than deal with extended VI
sessions.  The lack of an easy to use, intuitive fast console based
text editor is a weakness of the Linux desktop. Many windoze converts
are scared enough of the console. Inflicting VI on them is a great way
to send them back to windoze.

Lots on this list are comfortable with VI and love it. Nothing wrong
with that. It's the lack of empathy for those who are not which best
demonstrates the culture that can write off desktop development as
secondary. In my opinion it is crucial to the survival of Linux
period. More so leaving it purely to Ubuntu is a big mistake. For
years Fedora shouldered the main work in desktop development and other
distros benefited from this work. Unlike competitors in commercial
markets Fedora and Ubuntu can and should work closely together to
augment their desktop and other areas of development. Right now I see
little to no direct sharing between most distros. Especially the
Debian world which seems an island all too itself.  To yank out VI
would gall many on this list as much as I and others have been galled
by the lack of empathy for those who are comfortable with CDE and UIs
that are close to CDE. I had to learn almost nothing to use Kedit or
Gambas or a host of other Linux apps. This meant I spent my time doing
stuff with the apps instead of spending time learning how to use them.

A VB/QB drop in can still be a huge conversion tool and Gambas is
perfect for that. The language wars have lingered on since the FIDO
days. There is no room for such arrogance if we want to further Linux.
Support both. Add Gambas to the standard install, hype it. Even add
better VB/QB conversion utils. Think about the billions of lines of VB
code out there that would convert to Gambas far easier than to .net
and you have a very compelling arguement for many companies to switch
to Linux.  Each year that arguement wanes as more and more .net code
is deployed.

Still for the hobby programmer Basic remains the language of choice
and Gambas can do for Linux what QB did for DOS.  Basic made DOS
really. The compatability helped but it could have easily been
replaced by a competing OS but so many users had written so much code
with Basic that they did not want to swtich. Even today you find large
and enthusiastic QB clubs despite the difficutlties in getting QB to
run under modern windoze OS's.  QB and Rekall type utils create the
ability for people to quickly and easily learn how to program and do
useful work.

Multi-media - Face it this is the biggest and most important issue.
The poor Flash support is something that is hard to do anything about
right now short of writing an open source replacement for Flash.
Musicians find almost a total lack of support. Even long time Linux
users like myself have to keep a windoze box around to record on.
Rhapsody, Apple, and most other Music distributors have little or no
support for Linux. Even Amazon's Linux app failed on my FC7 machine.
Wine is nice but it's a mighty slow way of looking at the world.
Propriatory formats suck but they are reality. Replacing mp3 with Ogg
is a worthy cause on any platform and Linux can lead the way but there
is still a real need to much better support multi-media on the Linux
desktop. Eye candy can win many hearts. Educating people about Flac
and Ogg formats is important.

Yes there are lots of bugs to fis. Always will be. Lots of server
issues to deal with, always will be. Linux by being both a great
server, desktop and specialty OS will always see great demand. Windoze
is a desktop first and foremost, even their server versions are
desktops with server functionality added. Widnoze however is a
commercial effort and limited by that. Linux has the ability to
transcend those limits. That is where I come into odds with the old
school Nix crowd who control Fedora. The interest is not there. Many
of my potential and attempted contributions exceeded the defined
boundries of that umwelt of what Linux is. I am not alone in that.
This in turn discourages advancements in those areas. There are of
course practical limits. Because of lawsuits the MP3 format cannot be
officially supported but even if it was possible the culture is that
because it is propriatary it is unclean. Few Linux users like
propriatory software but the idealogical vrs the practical when you
look at the wider scheme of things, it just makes sense to do what you
can to enable users even if they choose to use propriatory formats.
You can't force what's good for people down their throats. You win
them over then introduce them to mature alternates.

A great example of this was when I wanted to document the Yum GUIs. I
was told that was out of scope in no uncertain terms. The vastly
inferior default Yum GUI being the only one which was officially
supported by Fedora and that would be documented. It took all the wind
out of my sails for writing any documentation. The default is pretty
lame. Yum Extendor, Kyum and Gnome-Yum all very good products which
novice users would find invaluable. All products you have to pretty
much discover by accident or be refered too by a friend. Use of a Yum
GUI alone would help adoption tremendously. That is one of the
strengths of Fedora. The ease of installation and the world of
applications that open up before you with those products. Command line
Yum with it's poor fuzzy search support and nearly unworkable search
facilities is all but useless. Without a Yum GUI users are forced back
to the windoze days of browsing websites and individually installing
apps.

That is an example of Fedora dropping desktop support before the
official announcement ever came out. An example of the culture that I
feel is committing distrocide.

> I also believe that the desktop is extremely important.  For Linux to be
> accepted in place of M$, it *must* be idiot-proof.  Unfortunately, the
> underpinnings seem to be changing so rapidly, things that used to work don't
> anymore, and no one wants to go back and fix them (or at least very few
> do).  So, the system gets lots of features, but the desktop drifts.

M$ is far from idiot proof LOL. Folks mangle it pretty bad just
installing and uninstalling software. My first encounter with windoze
I kept seeing this HUGE file eating up people's disk space. So I'd
delete it after looking at it in a hex editor and seeing it was just
random junk. Turned out it was the windoze swap file LOL.  What
Windoze power user can avoid registry hacks? What registry hacker
hasn't killed their system?  I remember once being tasked with support
of an Windoze web server. I followed the hardening guideline from M$.
It hardened all right. So much so I had to wipe and load to even log
in to the machine. I should have known better than to trust M$
documentation but I hadn't admin'd a windoze box in a few years and
had a mental lapse.


> I myself want to work on the desktop (presuming I have the time).
> Unfortunately, it seems that this has become an ominous task as I now have
> to learn yet another language (gnome script or kde script), and that isn't

Dunno about that. With the collapse of Open GL that leaves .qt as the
best portable graphics lib out there. Why not write in .QT instead? If
not .QT, then GTK is supported under all major desktops. Or you can
avoid the low level and write in Gambas or use some of the C++ GUIs
and do RAD development.  I am about to undertake writing a writer's
editor in Gambas. My first big project in it. Still playing with
different functionality to try to get off too the right start instead
of having to re-write core funcitonality.  Most of the functionality
is abstracted away, so really aside from minor quicks you are lib
independent with most of the code.

> on the top of my priority list right now.  Working with X, Xt and Motif was
> so much easier 10 years ago.  I could develop a Motif window much like
> Gnome's or KDEs, but it seems that the world had abandoned Motif.  By the
> time I get Gnome understood, there'll be yet another window manager that
> will be the favorite. :-)

I had to support some Motif based apps and the lesstiff vrs Motif and
what was the third Motif based set of libs? Anyway it was a mess. Some
apps ran great on Lesstif, others HAD to have a specific set of libs,
all 3 of which did not work and play well together. Motiff's cracking
down on patents though is what really killed Motif programming. Caused
many apps to get away from Motiff completely. Shrug I haven't done
much C coding since Win95 came out. I never felt comfortable with
C++'s syntax, never and got paid well to switch to VB around that
time. Found with VB I was worried mostly about actual programming not
about libs and system level aspects, as such got more done faster with
less frustration. Too bad the executable was so bloated and relied on
buggy dlls. Matter of preferences. Some folks like ;s at the end of
their line. OThers like me find them annoying beyond belief. The way
it should be is the language you write it in shouldn't matter much. It
all compiles down to the same machine code eventually. The language
should only be a matter of what works best for an individual.
Unfortunately that's not the case. The C/C++ only syntax of Linux
development languages has kept me from contributing as a programmer to
Linux. Wasn't until I stumbled upon Gambas that I found a language
that wasn't C/C++ based syntax with a decent IDE which I could quickly
jump into without getting tangled up in the
Motif/OpenGL/GTK/QT/whatever else dependency hell thing. As you said
it's just too mcuh work and too much re-learning to mess with. IT is
also a weak point on all platforms right now. M$ does a good job of
abstracting you away from that mess but it changes the language
without warning and drops support for your language without warning.
That's how I got into the DBA and Sys Admin racket. I was tired of
re-learning M$ languages. By the time I became an expert in them they
changed on me. I don't mean improved, I mean the entire language went
away.  Masm and Quick C gave way to QB. QB gave way to Foxpro which
gave way to VB which was replaced by .net. Pointless change. Instead
of creating new languages why not fix the old ones? Grre. Anyway,
Linux would do well to adopt better abstraction to improve programming
output and efficiency. Let the desktop and compiler authors worry
about the widgets. App level developers have enough to do dealing with
features and functionality. No point inflicting the M$ re-learning
curve either. Abstract it away, maybe give the app level programmer
the ability to specify the back end libs but don't make us interface
with them. I want to work on the program not spend %90 of the time
learning lib APIs only to see them change a couple years later and
have to rewrite my app.

Right now there are 3 biggies, but I'd love for some window manager to
come out and blow my personal favorite KDE out of the water. Until
then I remain a KDE fan. The underpinnings are for the window manager
group to worry about in my opinion. When I talk desktop I'm talking
about apps that are happy on any window manager. Core idea is to
abstract away from the details on desktop development. After all isn't
that what the Window manager is supposed to be about in the first
place? OS calls and use of system utilities should be universal. One
of my frustrations with distros is that they do not install both KDE
and Gnome libs by default. To be honest both have much to offer. Both
KDE and Gnome have undertaken a promise of better compatability
between each other. Having libs from both present allow you to use the
best of breed even if your app winds up running on a minimalistic
window manager. Having the libs present allow users to pick and
choose. They can run Gthumb AND K3b on the same desktop and suffer no
penalties. An app can leverage both K3b and Gthumb and HTML libs to
create a static slide show CD with minimal coding and not care what
window manager a user is running as long as they didn't uninstall KDE
or Gnome libs/apps. Not much point re-inventing the wheel and the more
wheels you have to choose from the better. The more wheels the more
likely you are to find the one that fits your needs the best.

That is also what I mean by desktop. Give a windoze convert the
default Fedora desktop and they are lost. Big time lost.  Half of the
apps they would instantly gravitate too are not there because KDE is
not installed. So apps like K3b are not there. Gnome finally has
something close to K3b in functionality but I'd still rate K3b way
better. KDE based distros again do the same thing. Lots of Gnome apps
are really best of breed.  Lots of apps are window manager
independent.  Just by creating a default install which doesn't entail
writing a single line of code but that does a better job of
streamlining the user interface and shortening the learning curve you
are doing work on the desktop. Best of breed discovery and inclusion
is another important part of desktop work. Documentation is a crucial
aspect of desktop work. It is our documentation the user first see's.
Many users do or don't explore Linux based on the distro's default
documentation and thier ability to do the things they did under
Windoze or the Mac by that documentation. Any time we do not point out
best of breed and show alternates we remove one of Linux's big selling
points. That is choice. That is the level of customization that comes
quickly and easily with Linux. The ability to quckly and easily build
a machine that works the way YOU want it too. I disagree users will be
confused by choices. People really are not that dumb no matter how
often M$ tells us they are. We fall trap to M$ propoganda when we
disparage our own user base. We in turn feed that same propoganda.
They might not know the difference between a gig and a terrabyte but
they know what they want to do and only want to see how to do it.
Offering different ways to try something long as it works only gives
them more opportunity to customize and find a unique LInux experience
that best suites them.

> Anyway, I wouldn't say that the desktop is dead; it's just dormant while
> other things are more pressing.

The official release was that Fedora had stepped out of desktop
development. The Linux world was not surprised at all at that
announcement and the feeling is that Fedora has officially ceased such
work for good.

> Thanks for your view, though.  It helps me to understand where efforts need
> to be placed.

Always happy to get on the soapbox :)  M$ users absolutely hate me I
suspect LOL. Here I have lots of positives to accent. When I get lured
to a M$ blog by something I just plain cut them to pieces. Here I hope
to point out a road that eases migrations. I do quite a bit of that as
I often help people try and hopefully stay with Linux. A dozen or so
every year at least. Every so often I get the chance to convert a
school or business to Linux.  Much of what I've written is just plain
feedback from those efforts. Me personally, it's no big deal to add
KDE to my installation. I customize my installations anyway. That's
just a few more boxes to check.  I customize my partitions as well, so
changine the default partitions is not a big deal. I would NEVER set a
new home user to use the default Fedora partitions.  /home should
NEVER EVER be on the same partition as /.  I've had more users wipe
their home dirs while trying different distros than I can count. They
fill up the /home partition with .xsessionerror files or downloads or
other junk and come a running telling me Linux crashed. They corrupt a
partition and with /home on the / parition it makes recovery so much
more difficult.  I also preach the /data partition because of several
reasons.  Searching on /home means hitting those thousands of system
files. Backups are bloated by things like Mozilla cache files and temp
files and installation files.  Again recovery is a pain. If your
walking a disk and trying to reassmble files and have these zillion
temp files to contend with, it makes life SO much more dificult. If on
a /data every file is a real file it makes life much easier. Backups
are easier. Most users really don't customize the desktop much, but
thier data dir is very dynamic. Then there's security issues. Can't
count how many times I've caught users making the /home public read
write to share documents. I firmly believe cfg files should be
seperated from user generated files. As such /home should be split
into /home and /data.

That is where my perspective is coming from. From front line work
getting folks converted over to Linux. I don't know too many old
school Nix users. Most LInux users I know are people I personally
converted to Linux or people just trying it out who've come to me for
help. When I write I write for that user. My suggestions to this and
other Linux lists are for that user for the most part. Every so often
I'll do a rant about something that annoys me personally but doesn't
have a big impact on new Linux users. So my biggest exposure to old
school nix is on lists like this. When I don't have time to read lists
like this I often forget there are people who never ran windoze as a
primary OS or who only did so recently mostly because they are stuck
in old school Linux apps. I've met a couple folks who are comfortable
with VI but who use Windoze now despite coming from a Nix background
because Windoze is "easier". Well yeah if your using old UIs like VI
LOL. Yet they are resistant to trying new Linux UIs which blow Windoze
UIs away. It'd be hilarious if it wasn't tragic. Oh well. Life
happens. It just boggles my mind to see somebody dealing with
Notepad's primitive interface when Kedit provides the same ease of use
but without the bugs and annoying limitations such as attempting to
force an ext down your throat.  If you do things old school yeah it's
more work. That's the whole point of my soapbox rants about the
desktop. I hope to see a better default desktop that will help in
migrations. I spend far too much time helping users with basics and
lose too many on that stage before I can really show them areas where
Linux shines. A better desktop means we get right to the part where
Linux shines and skip that ugly painful middle step.