Mail Archives: djgpp/2012/05/19/19:45:37.1
Hi, :-)
On May 19, 1:13 pm, DJ Delorie <d DOT DOT DOT AT delorie DOT com> wrote:
> > I was glossing over it ... DJGPP needs a purpose.
>
> To me, DJGPP was never the goal. DJGPP existed to *fulfill* a
> purpose, not *as* a purpose.
Yeah, you wanted to write a 32-bit OS, right?
> Originally, it was the only way to make
> 32-bit DOS programs *at all*.
I'm sure other DOS extenders existed, but of course DJGPP was probably
the first free one.
> Then it was the only 32-bit Windows way.
You mean 32-bit apps under 16-bit Win 3.x?
> Then it was a way to get direct hardware access.
That's one of DOS' "features", but it's not necessarily popular.
Though DJGPP made that better with its rock-hard stability.
> These days, MinGW and Linux cover most of the purposes DJGPP was
> originally created for - free software, high quality code, 32- and
> 64-bit environments, etc.
Not really, MinGW only (properly) runs under Windows. Linux binaries
only (properly) run under Linux. Workarounds exist (WINE or *BSD's
Linuxemu), but they're not meant to be portable binaries.
It's kinda a shame. If Java wasn't such a bloated pain (with obvious
bad mojo), perhaps something similar would be more universal. Perhaps
we really need a better VM (though it seems there are dozens but none
totally stable, easy, popular, etc). If anything, you could almost
call DOS emulation such a VM. So I don't see how we're any worse than
Java people (though I know that may be stretching it a bit).
> DJGPP still serves some needs, like direct hardware access and DOS
> boot disks, but let's not make more work for outselves when others
> have already done it.
We know you're always busy. Same with CWS. We're lucky it's still
updated at all (thanks Juan, Andris, Eli!). It's just not necessarily
the end of the road. There's still plenty more that could be done. But
I guess talk is cheap, so if someone (like me) wants to do something,
he better roll up his sleeves and get to it, show the code, submit
patches, etc. (Easier said than done.)
Oops, almost forgot: "but let's not make more work for outselves when
others have already done it" ... sorry, but Linux has 300+ distros!
How many *BSDs are there? Was Clang or PCC or TCC necessary? How many
GUI toolkits are there? Different kinds of Unicode? C-based languages
supported?
Sorry, DJ, but apparently the world disagrees with you: the more the
merrier. ;-)
P.S. Though seriously, I think building DOSEMU atop minimal Ttylinux
wouldn't be a bad thing, would it? One of these days I'll have to
try ....
http://ttylinux.net/
- Raw text -