Mail Archives: djgpp/2009/02/03/18:30:04
Hi,
On Feb 3, 2:37=A0am, "Rod Pemberton" <do_not_h DOT DOT DOT AT nohavenot DOT cmm> wrote:
> "Rugxulo" <rugx DOT DOT DOT AT gmail DOT com> wrote in message
>
> news:5a8fd366-52e8-4389-adde-8ea41684279c AT t39g2000prh DOT googlegroups DOT com...
>
> >http://gnuwin32.sourceforge.net/packages/libgw32c.htm
>
> > "This is an implementation of a small part of GLibC, just enough to
> > compile most Unix programs on MS Windows.
>
> It's good to know what that minimum set is. =A0As you may recall, I've ke=
eping
> track of similar info. =A0Although from what they listed, that minimum se=
t
> seems quite large to me... =A0and has way too much POSIX. =A0It's amazing=
to
> think about the fact that they didn't have the forsight to restrict codin=
g
> their utilities using only ANSI C - except where impossible. =A0It's a hu=
ge
> waste of the potential of their codebase, i.e., not portable except to PO=
SIX
> environments.
Windows isn't POSIX either. They've removed their POSIX subsystem in
latest versions, Interix is non-free and seems frozen (from what I've
heard), and people always whine about how they lack many functions. So
I dunno. Of course, the main point of me mentioning this is that "Hey,
somebody ported part of GLIBC, so I guess DOSEMU ain't too impossible
to host on Windows now." Then again, DOSBox CVS is supposedly improved
a fair bit now (P166 ??), but I'm not really interested in that until
final release.
> > It contains functions for
> > passwords, process id's groups, and strings. Most are interfaces to
> > the MS-Windows Win32 API. Some are just dummy functions that do
> > nothing. "
>
> The conversion to Win32 API makes it useless for non-Windows use...
> although it probably edited out much useless *nix code. =A0It's likely no=
where
> near ANSI C. =A0I just can't win. =A0Although, it could be a good startin=
g point
> for an ANSI C compileable libc, if you wanted to deal with the GPL.
There are a lot of projects that aren't as wide-reaching as GNU. Of
course, even GNU seems to hinge on "GNU/Linux" these days (thus at
cost of less support for others, if at all).
> > Hmmm, actually all I had otherwise was just a bunch of links to GCC /
> > EMX / FPC cross-compiler stuff (e.g. BinUtils Win32 cross-tools from
> > FreePascal for various targets, various EMX DOS+OS/2 compilers +
> > updates, various RSX + RSXNT + RSXNTDJ + NT09D files which are
> > apparently very hard to find, as well as the old FPC 1.0.10 DOS+OS/2
> > EMX version). I guess that's not hugely relevant here, but at least I
> > found it interesting.
>
> I've heard of EMX, but since I don't have OS/2... useless (?). =A0I think=
I
> looked at that RSX stuff. =A0I don't recall what I thought about it. =A0I=
'm
> unfamiliar with the other combinations.
I've never had OS/2 either, but EMX isn't useless (although OpenWatcom
now exists and works well too), esp. since it works in both DOS and OS/
2. RSX is just the extender to work under DPMI instead of VCPI (which
most Windows hate). RSXNTDJ 1.51 lets you use DJGPP to make
"bound" .EXEs that run in either DOS (w/ RSX.EXE) or Win32 (w/
RSXNT.DLL). Actually, his RSXWIN + RSHELL lets you run EMX or RSX apps
in a Win16 GUI shell (uses less memory than standard DOS prompt).
Seems that he dropped bound support in 1.6 beta2 (which I can only
find truncated at 99% from WayBack), preferring separate DOS and
Win32 .EXEs (and using CRTDLL instead, which loosens the license a
bit). But at least 1.6 beta2 has a more recent version of RSX.EXE
(5.24) and RSXNT.DLL:
rsx exe 111,900 Feb,09,2000 06:59:32pm A...
rsxnt dll 60,416 Jan,06,2002 07:56:50pm A...
- Raw text -