DOS ain't dead

I don't think that Linux is any kind of priority for GNU. :) Their projects on Linux yes. It's kind of a test case for Hurd.

> Though the GNU Coding Standards
> doc does say
> this:
>
> "In particular, don?t reject a new feature, or remove an old one, merely
> because a standard says it is 'forbidden' or 'deprecated'."

Well, it depends on the project. GCC is known to conform to the latest standards. That means the old code (especially true of 1990's C++) might not compile with a new compiler. Also, if you want to compile the pre-standard C (k&r), you have to use the specialized tools (or convert them manually).

> But, face it, when the heck was the last time a GNU maintainer (or any *nix
> or Windows user) gave a crap about anyone else?

Me? :) I have a wide experience with different OSes, including the current "big-three" (Windows, OS X and Linux) and BSD derivatives. :) Only area I'm lacking is CP/M. Was to young at the time and never got hooked while trying under CP/M emulators.

And DOS was the first, so it has "the special place". 1992. 386sx with 1mb ram and HGC. Ah... The good times. :)

> Automake requires Autoconf,
> which assumes a POSIX shell plus lots of POSIX utils. Usually it "barely"
> works in other places, e.g. DJGPP with ancient Bash 2.x (under DOSEMU or
> NTVDM).

Well, it wasn't originally concieved to be run on anything else than *nix. On some level it really is cross-platform. It makes porting the software between different Unices easy. That was a big time problem in the 80ies. Before POSIX there were much greater differences between different Unices (for example: BSD, AT&T, OSF).

> None of the *nix nerds even bother testing with DJGPP (though it's
> freely available), so I'm not really surprised. Plus most projects have
> long ago dropped working (much less non-working) DJGPP support.

That's also true for most of the Windows based projects, isn't it? Not fair to point the finger only in one direction. ;) Microsoft was the first to kill DOS by promoting other solutions and not improving it since the first half of the nineties. Only support stem from the need they had for it, since both Win 3.x and Win 9.x used it underneath.

> So yeah, we know DOS is cool, but most people want shinier, newer bullcraps
> to mess with (OpenGL? HTML5? Unicode? SMP? x64?).

Not all of that is bad. The problem is, bringing them to DOS would make DOS being something else. It won't be "DOS" anymore. :)

Well, it's a simple 16-bit OS. Still usefull for RT development and embedded uses where Linux and/or *BSD are overkills. It's also very usefull for the retro-computing buffs. Either for old computer collectors, either for the new generation of emulation buffs.

I'm somewhere in the second category and would like to see DOS last for the ages. I'm sad to see anything DOS related die out.

> > How active is DJGPP still? To have influence you must first be active,
> no?
> >

DJGPP still has an officially-maintained website and newsgroup and (old but mostly relevant) FAQ.

> > If I go to their site, and click release, all I see are archives from
> the
> > early 2000s.

"Archives" of what exactly? The two mailing lists (one of which mirrors the newsgroup) show plenty of recent activity.

The public development .ZIPs (libc, headers, etc.) for "current" (2.03p2) and "beta" (2.04) are approximately from 2002 and 2004, but they still work fine (more or less). Other .ZIPs for other utils have been updated more recently, esp. GNU software (GCC, et al.). Most, if not all of these, are announced on the newsgroup.

> They are still alive. Andros, for example, provides Fedora crosscompiler
> regulary. He also has GCC 4.7.2 available. Native (DJGPP) versions are
> usually also available on DJGPP ftp mirrors too.

Some work has apparently been done for prereleases of 4.8.0 (see Changes), so we may get packages for that in the next month or two.

> While I haven't (yet) had luck in getting it to work on the distribution I
> am using, I follow it from time to time. But, I'm close. The biggest
> problem is that the patches are scattered all around. I have to decipher
> Red Hat spec file to turn it into something I can use. And, I also don't
> have much experience with crosscompilers ATM. That's a bit of problem too.
> :)

There are various workarounds for that (Ozkan's older cross compilers, native under DOSEMU, CDEpack), but I also admit to not having tried using Red Hat cross compilers (and am skeptical, also due to my inexperience).

> LibC is unfortunatelly only somewhat active in CVS. But Rugxulo is
> following that much more closely than I am and he might know more.

I don't follow it that much, but I am subscribed to djgpp-workers (for whatever reason, I'm not really a contributor). It works as designed, so there is little need to add much except for rare bugfixes or minor tweaks / additions to aid in porting other (usually GNU/Linux) software.

P.S. A year or two, I signed up so I could edit the FreePascal Wiki, specifically the Go32v2 section of the helper tools page. But I've been distracted, as usual, and probably forgot the password. Though I don't think marcov (or maybe anybody??) reads that page, there is no Go32v2 maintainer. (I really don't know why marcov frequents here. He's always welcome, but he mostly focuses on FreeBSD, and I'm not sure he's ever directly used DJGPP at all. I know he's not that dumb re: DOS, but certainly it's of very very very low interest to him.)