pouët.net

@gasman asked "does ZX Spectrum platform needs to be subdivided?" (http://pouet.net/topic.php?which=6761&page=1&x=11&y=10)

This got me thinking a bit about another category : DOS
As of today, pouet has only 2 sometimes unfitting categories : DOS and Windows, while none of them seem satisfactory imho if you look at objectivity and mostly user-habits.

1) DOS is dead and doesn't exist since win98 ?
If you select DOS only platform, startup values for NTVDM are different than pure legacy DOS wich is unattainable without VM or rebooting (for some hardware ports).
For example, none of the acclaimed DOS Future Crew demos work on my XP PC (read xp NTVDM) when clicking on them.
The samecan be verified with all DPMI and XMS based DOS demos (rougly 70 % of 1990 demos).

2) NTVDM doesn't exist as platform on Pouet
To most DOS on PC means double-clicking on .com and .exe in downloaded zip file from the download link (while surfing in 64-bit on 2nd monitor) : it's not DOS it's NTVDM.
If you're ever subtle and click both platforms to indicate it's ntvdm compatible, there will still always be DosBox or other Win32 platforms saying it don't work (omitting "on my system"), or verifying which version it is meant for by clicking on the .nfo, if you cared to indicate it.
The worse you can get is "Not DOS, not Windows" - which is not senseless in its own right !

3) NTVDM differs on each Windows operating systems as well as x86 architectures
NTVDM is different on each version of windows , and most 16-bit binaries don't work on Intel's 64bit ripped architecture from AMD : although you make no call gate privileged calls and the AMD design reverting to 32bit boasted full backward compatibility.
But it gets worse, NTVDM startup values are different from one Window system to another, for example Windows Me and Windows XP (see the troubles Rudi had recently with his 64 byte intro : http://pouet.net/prod.php?which=61108).
So unlike fixed "defunct" systems, your cherrished optimized lz77 routines and other snippets collected over the last decade are not sure to run correctly in next competition.

4) no DOS, no NTVDM and no BIOS platform.
The only true DOS platform in people's mind seems not even to be DOS, but the "black screen you get at the beginning" like what the OSDev guys have with the Master Boot intro competitions really @7c00h. Which is IBM PC BIOS, and as freeze-framed by Ralf Brown's legacy interrupts in fact.

5) Efforts to normalize "DOS" startup context have been disregarded ?
The decade old Hugi size-coding competitions (http://retrolandia.net/hugi/showthread.php?tid=58) organizers have made a systematic effort to
set once and for all what one can expect in order to differenciate a software bug versus a compatibility issue (and problem between chair and computer) :
http://retrolandia.net/hugi/showthread.php?tid=4 but to no avail except techies.

6) Other problems affecting lack of coherency in categories (otherwise called rants)
-128 byte intros on Atari are 160 bytes ?
-midi is never clearly indicated "on" whilst it ought to be and always present (but depends on the person devising the virtual sound card driver installation at one moment). SBlaster is incorrectly emulated also.
-vga standard mode 6ah (cf. wikipedia) was incorrectly zapped without anyone noticing during last years, either to force DOS users to call Windows or same non-interest as Yamha OPLx as previous point.
-people accept Windows API thunk to call SAPI, but not when you call the SAPI from DOS exec API through vbs dumping (where is the logic ?). An API is an API.
-even Windows programs are more compatible with it's hugely overseen and hidden backward compatibility WinSxS folder. Compo organisers claim to
delete the media directories from win7 for fairness but do they delete the WinSxS folder, which are like virtual machines enabling a non-running IRL
windows demo for xp to run in windows7 ?
-BIOS is not hardware (it's cheating - read in #asm IRC) Amiga is more difficult you do everything yourself.BIOS is on a hardware chip and available through all boot stages (if you call a chip on Amiga, nobody says it's cheating).

So you see gasman, you are not alone :) Yes some categories need fixing, but no don't expect them soon.

So did I read you right? Since some random emulator does not run them properly on some architectures or whatever, the category should be split somehow? Should we split C64 prods according to the ones that run properly with CCS64 and the ones that do not?

Get a real computer that is running a MS-DOS compatible operating system. There is your platform right there. If your emulator refuses to run programs that run on that computer, then the emulator is at fault and not any platform definition. If the program refuses to run on real hardware, then the program is buggy (or the hardware lacking). If there was any need for splits (which I don't think so), the sensible ones would be to look at memory requirements (EMS compatibility), graphics modes (low res 32/24-bit mode requirement) and sound cards (but we already have DOS/GUS category). That's the stuff that caused the most trouble back then.

Yes I hear you. You know your stuff, but did you read :



Also, what about the missing category of BIOS platform (a platform in its own right if there is one that exists since before MS-DOS) ? It doesn't classify as MS-DOS (edi=7c00h and you can acesss 4GB flat real mode through A20 gate enabling 640x480 in 256 colors without calling bank switching).

No.

Fair enough, though I at least don't belong in this category (I've always used DOSBox). Something like a list of NTDVM compatbile productions would definitely be nice if someone wants to spend an effort to compile such a thing. Also, some kind of huge torrent that contains DOSBox, all the classics and necessary settings for running them without a hassle would be really nice as well. I don't think neither are a reason for splitting the historical category, though.



Are there actually any productions like this?

I'm not aware of the poll you used as a source for this, and is thus quite sceptical to the conclusion. For me at least, "DOS" means "ugh, dusty old computer".

I don't think anyone expects [random DOS demo] to work flawlessly in NTVDM on their windows boxes. As far as I'm concerned, DOS compatibility went to hell with Windows XP, a bunch of 256b intros being the exception.

If NTVDM was sometimes targeted as a platform in its own right, I could see the use of such a category. But is it really?

NTVDM was never meant to run demos and games.
Dosbox or PCEm are much better for watching DOS demos on modern PCs.

Granted, there's a handful of intros that exploit NTVDM by calling Windows-specific features from a 16-bit MZ binary, but hey. Why not just list them as Windows prods? No need for separate NTVDM category.

if a demo relies on startup values to be set by a specific version or emulator of msdos, then that should be mentioned in the readme, at least.

NTVDM is not available on Win x64... and it is meant for providing DOS compatibility since Win NT.

No.

Now, a separate Windows x86/x64 platform division might be interesting.
Current Windows platform will be for x86, and if there are/will be demos that have an x64-version (or ar even x64-only), they can get Windows x64 added.
Then NTVDM falls under Windows x86 only.
Problem solved?

While we're at it, let's remove all words currently considered "offensive" from republished versions of literature originally written over a century ago.

also shut the god damn fuck up if you've never coded a real dos prod in times where dos was the standard

JUST
SHUT
THE
FUCK
UP

fucking hipsters

I couldn't read much of it because your enter is broken.

Leave the DOS alone! It's fine as it is :)

The DOS platform needs to be consolidated, if anything. Don't really need the separate MS-DOS/GUS category.

I agree with Phoenix.

gloom: why not drop all platforms in favour of yt instead? maximum compatibility ftw!

Yep, forgot about the dos/gus category. Doesn't make much sense when I want to search best dos demos and miss some good stuff in the other list. Not a big distinction for PC too.

the older the intro/demo the more you get the chance that it won't run on a newer system. so where is the problem? :D



nope.

To me real DOS would be :

8086.
286.
386 max.

Then :
Hercules.
CGA.
EGA.
VGA.

Then
Real sound (beeper)
Adslib.
soundblaster.

I dream of a demo wdesigned especially for the Amstrad PC1512 or 1640.

To say "Dos" computer without explaininga bit some of the Hardware configuration is somewhat stupid.
Demos are supposed to work the Hardware out of its own ass.

A demo for GEM, you mean? ;)

Then so they would add GEM as a category and later some Atari dudes will come and say that GEM should be subdivided too.

:D

Hey everybody, let's consolidate Amiga ECS/OCS, AGA and PowerPC because it's all the same to me I can't give two shits about things that are over my head k thx.

That won't happen.

You see: There are demos that require MMX, some require a GUS or even a specific GUS model, some need an Adlib OPL2/3 card to function, some require specific CPU+FPU combinations etc. The list goes on and on.

The DOS/GUS category is a half-decent compromise that mainly works because of the popularity and (often) exclusiveness of that hard-/software combination, making it a de-facto platform ("No Gus, no demo!").

But the general solution here, would be some requirements management, either by user tagging (assisted by nfo-file parsing upfront) or some other req-list system, that would likely need constant maintenance. I'm sure Gasman and the demozoo people have that kind of stuff on their long to-do list, but pouet with all its internal and external inconsistencies is probably the wrong place to ask for it. :)

yes, it would be helpful if there would be a demoscene db that would actually have proper info for what kind of hardware/emulation would be good to run the demos in realtime.
i find it sad that pouet is not even trying to do that. but i guess the pouet admins want that people only watch demos from youtube.