modular -> monolithic

[Alan W. Irwin]

On 2008-01-22 20:55+0700 BuraphaLinux Server wrote:

> As an adventurous user (yes I build everything on my system but java),
> I remember downloading one big file, setting a few environment
> variables, doing a make world and then using the software.  With
> enough time and disk space, anybody could do it.  Even a fool like me.
> Obviously I still have to use 6.9 since nothing after that ever built
> according to directions.
>
> Now you have to be some super-guru to get the thing to build since in
> this thread even the ordinary gurus admit the build instructions don't
> work AND THEY DON'T CARE.  Telling us to use a distro version pretty
> much is admitting that an unpatched Xorg that works is a myth.  Yeah,
> the pace of development is fast.  So is diving out a plan with no
> parachute, but the end result in both cases sucks for most people.
>
> You guys need to get a build script that works, and have ONE place you
> download everything you need in one big file (a distro-agnostic iso
> image or something and not having to search dozens of directories) and
> then it just works for people with enough time and disk space.  Then
> we don't care if it is modular or monolithic.  If it works, we are
> happy, you get more testers, and you can spend time fixing bugs
> instead of having these silly flame wars.  Make a new ISO once a month
> that has been build-tested and run-tested with say real VGA and VESA
> mode.  That's enough for most people to have a 2-D fallback, and they
> can test your 3D and special drivers and if it breaks they stay 2D
> VESA for a month until the next ISO.  People get to build from source,
> they can use that stuff that has all dependancies worked out  for them
> so they KNOW they have compatible versions that will build in the
> right order, and life will be better.  Heck, the kernel is every 2
> months - even a guaranteed working build every 2 months is better than
> what you have now (total chaos which you appear to be proud of).
>
> Just my $0.02

I woke up this morning with the determination to pound some sense into the
subset of X developers who don't seem to care whether its a straightforward
process to build X or not, and then somebody else said it for me!

I haven't yet built X myself, but I do have lots of experience building
various software packages, and I naturally assumed I would be building X in
the future.  However, I was simply appalled when reading some posts in this
thread that indicated (what I hope is) a small subset of X developers don't
care about buildability issues.

It's time for the giant foam clue bat!

Today's amateur builder is an immediately important addition to your testing
pool and a potential X developer for tomorrow so you piss those guys off at
your peril.  So as top priority do what you have to do to make it
straightforward (notice I didn't say easy) to build X _again_. I emphasize
"again" because I well remember my office mate (who was no computer guru)
building X for all the different kinds of sparcstations our department had
in the early 90's.  It took him a couple days (because early sparc stations
were no speed demons), but it was a straightforward process which just
worked.  Make it so again!

My $0.02.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of
Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________