Back to home page

DOS ain't dead

Forum index page

Log in | Register

Search:  
Back to the forum
Board view  Mix view

NASM version 2.09 available | A86 and kernels (Announce)

posted by ecm Homepage E-mail, Düsseldorf, Germany, 08.09.2010, 20:30

> > > But yeah, it hasn't been developed since 2000, [...]
> >
> > That really doesn't bother me.
>
> Nor me, but it just means he hasn't kept it up.
Which wouldn't matter to me if it was as sophisticated as NASM or more so to begin with.

> Granted, I personally can't
> help but wonder if we've "jumped the shark" on x86, adding so many things
> (SSSE3, SSE4.1, XOP, AVX) in addition to obvious things like FPU, MMX,
> SSE2, AMD64.
It's what works with the market right now. Ask the next Itanium-based PC you see.

> Yes, obviously I didn't mean to imply legalities.
I don't see why the term "derivative" mentioned by you would apply otherwise in this context, but oh well. Probably just because that's what's on my mind right now.

> Looking at the past ten years, we would've
> never guessed computers would change so much.
But what about their evolution from 1990 to 2000? The decade before that? Computers always changed much. I'm used to it. The only thing that might have been surprising was the internet's development, although that took off in the 90ies too.

> Yes, of course, I just hate that it has a license which apparently "sucks",
> but nobody is around to change it or even discuss it officially.
Better than with DRDOS, Inc where they have refused to provide EDR-DOS with a freer license. (That they tried to sell trash made of stolen software is an entirely different story.)

> 99% of Linux distros aren't even FSF-labeled "free".
FSF isn't even what I would label "free". I don't want to primarily produce free software for its own purpose (at least, not currently). Strategically making available certain projects only with Copyleft is acceptable, and this Copylefted part of a popular distribution of course needs to be massive enough to avoid it being "stolen" for closed-source software (with the Copylefted stuff replaced). But I don't think Copyleft really is the way to go for all software, at least not in its current implementation. (The GPL is too strict as well; I would prefer LGPL (for everything). If I were to choose a license from the FSF.) I want people to copy parts of my program if it helps make their software better - no matter what license they (have to) use.

> Some of the kernel devs probably knows more here than me what would need to
> be done.
On gcc? Add 16-bit support and OMF object/.EXE/.COM output, he. On TFK? Probably not much, I assume it's portable enough that adding support for another compiler (gcc or not) wouldn't be too hard.

> I don't honestly even remember what memory model the kernel uses.
Memory models are enframing. They don't exist. Seriously: That doesn't matter much. Unless your compiler doesn't support far pointers at all you can do everything. I don't know of any compiler that does all of the (sophisticated) relocation required for a kernel, so you need to do some things on your own anyway.

> In theory, we could probably easily convert it from (non-portable) C to
> (non-portable) assembly for NASM, I guess. WDIS is our friend. ;-)
Shush, interest in kernel development is little enough as it is.

> No, I think they removed it after the TomTom incident. Last I heard, they
> only write the LFN, not the SFN now, but I haven't tested.
I think that was left as an option that defaults to off, for now.

> I didn't mean it was "better", just smaller, [...]
As the pedant I am, I have to note that you are right: TC is smaller. Its output is not :-P

(I meant "produces smaller output" with "better" here.)

> [...] and originally LOADFIX for FreeDOS was a separate util anyways, [...]
LOADFIX is easier to implement in the shell because you don't have to hook the program loading or termination explicitly.

> 0.84 had "too
> much added" without review, says Eric, including an obscure bug/hang, and
> even Blair never updated it "yet again" like he originally claimed
> (swapping, oh well).
So yeah...

> But yeah, we could use a bit better organization
What organization?

> I still say we should (also) focus on DOS emulation,
> which is "teh futurez!"
I say someone should make a serious fork of DOSBox which makes its CPU, BIOS and DOS compatible properly, or just replaces the latter with a host file system redirector driver done right. An interface to load and access host modules (.DLL or equivalent dynamic libraries) should be provided too. They should also care for bugs and issues that don't affect any specific DOS games.

dosemu might be a better base looking at its features, but I don't think it would be easy to port that to operating systems other than Linux and/or architectures other than x86 or AMD64. (Not a big problem now, but might become one. (That is, if PCs switch to another architecture before they're powerful enough to fully emulate a current x86 machine.))

---
l

Post reply
 

Complete thread:

Back to the forum
Board view  Mix view
22649 Postings in 2111 Threads, 402 registered users, 896 users online (1 registered, 895 guests)
DOS ain't dead | Admin contact
RSS Feed
powered by my little forum