Mail Archives: djgpp/1999/04/08/22:48:13
[Note: this is long. Please prune it to the relevent parts if you
reply. DJ]
> Sigh, you're missing my point. DJ invited for a reflexion on the
> `avenir', future, leap forward, what other cools thing to do, to
> bodly go where DOS
Not DOS - DJGPP. I didn't specifically say anything about DOS. Use
your imagination. DJGPP is more than a "dos" compiler, more than a
development tool. It's an environment, a community, and much more.
If we limit yourself to DOS, we'll die with DOS. Let's not limit
ourselves, but let's not re-invent the wheel either.
> That's what I'm trying to get out with one more comment if DOS/DJGPP
> wants to survive as a usefull system it must keep up with its time.
My rebuttal to that is simple - why not switch to Linux or NT?
There has to be a reason people use (more importantly, continue to
use) DOS/DJGPP over Linux/GCC or NT/Cygwin. What's the reason? How
can we keep people using DJGPP when other free packages are available?
*Do* we want to push it, or should we put our collective energy into
something with a brighter future? Preferably, something that we, as
developers/users of djgpp, would have an advantage over others in
doing.
What is it that DJGPP (and the djgpp community) has that other
projects (linux, cygwin, msvc) don't - *won't* - CAN'T have? And, how
can we take advantage of that?
Note: I'm not saying we should drop DJGPP. I expect it to be around
forever (well, as long as *I* have a machine, at least :) There's no
reason to let it fall by the wayside as long as people want to keep
using it. However, if it's basically done, does it make sense to keep
putting effort into *growing* it, or should we so something else?
Think big. Use your imagination. That's what I said, and there have
been a few big ideas (FreeDOS, X). Cool. Think bigger. DJGPP is ten
years old. What will the next ten years bring?
Example: One idea I had was to package up DJGPP, FreeDOS, an enhanced
RHIDE, and a HUGE information/tutorial library into a "Everything You
Need To Learn Programming - No Matter What" CDROM. Do a big marketing
campaign to sell it to schools - package deals of CDROM, installation,
what little training is required, and support. We can each do support
contracts with our local schools and make some money. Eventually,
make the package support Linux and NT also (using their compilers, of
course - it shouldn't matter, right?) and become *the* way to learn
programming. We could even write some tutorial-specific software,
like integrated html viewer, BASIC (or whatnot) interpreter, graphics
screen, IDE, and smart tutor/agent. Do a bundle of asm-like
simulators designed not for debugging, but for teaching. Combine
Allegro, Perl/BASIC/whatever, and RHIDE to make a "Learn Programming
by Writing Arcade Games" bundle.
Example: Do a "DJGPP/Linux" distribution, which is a Linux bundle
specifically designed for users who will run DJGPP programs in DOSEMU
(although *I* don't see the point in that ;). Include a cross
compiler, native djgpp and linux tools, but toss all the multi
user/server stuff (i.e. make a dos-like machine with a linux kernel).
Prepackage it with all the usual DJGPP goodies installed by default.
Add a point-n-click admin tool that really covers 100% of all the
stuff it comes with (you can do this if you limit what it comes with).
Start with a raw Linux kernel and build it up from there, dos-style
(config.sys, autoexec.bat anyone? (ick)).
Example: The DJGPP community is, IMNSHO (in my not-so-humble opinion)
a *great* community. Everyone works together *very* well (kudos to
you all) and we've done some big projects together with little hassle.
This is not normal for the Internet. Some problems I've seen others
notice: matching projects with developers, putting the finishing
touches on projects, coordination, infighting, loss of interest. I
think our community works together better most, even better than some
companies. Why not act like a company? We could be the first
net-based moneyless real virtual company. We could have a real org
chart (something most net development groups don't have), real
schedules and commitments (ditto), standards for "finishing and
polishing" projects, and maybe real marketing/sales/admin groups. It
would be an interesting experiment to see *if* it works - we'd have to
avoid dealing with money at first, but that's what we do now. If it
did work, we could leverage our success to pull in development
contracts and maybe make some *real* money, or at least get around to
some of those projects that are languishing on the net for lack of
interest ;-)
Example: Port XFree86 to Windows. Write a DJGPP X library that can
talk to it. Make sure cygwin and TCP can talk to it too, of course.
- Raw text -