Talk:DOS extender

Modified much of the original content

Modified much of the original content, added too much information about stuff that's now irrelevent. Hopefully appropriate and up to Wikipedia's standards, this is my first major contribution. --X3J11 20:57, 27 June 2006 (UTC)

Incoprehensible sentence.

I've been trying to improve the first few paragraphs of this article, because they sound rather clumsy. They must have been writen by someone who doesn't natively speak English. Anyway, I removed the sentence "Most XP users know nothing about them, but are occasionally happy to have DOS based partitioning, recovery or similar products, partially faking the "Win" look, and based on DOS extenders of course." because I can't understand it at all, and I was therefore unable to improve it. If anyone can figure out this cryptic gibberish, please improve it and put it back in the article. Nick Warren 17:00, 24 September 2006 (UTC)

Someone seems to have been kind enough to put it back in for you. :/ There's a lot in this article that (I think) should be changed/removed, but I'm afraid to do it since other folks have done it to what I wrote (which is kinda disheartening). But there's lots of extraneous junk that's been added. X3J11 21:23, 24 December 2006 (UTC)

Cleaning It Up Some

I've taken the liberty of trimming a lot of the extra "junk" from this article. Specifically, the chunk:

...and many older extenders had limitations of 16, 32 or 64 MB (even Windows 3.0, which used a DOS extender, were limited to 16 MB memory, and relied on the Windows 3.0 version of himem.sys to enforce this limitation in standard mode. Windows 3.1 increased the limit to 256 MB memory and that limitation were because of the limited number of descripter table entries, which allowed Windows 95, because of it's flat address space, to increase the limitations to 480 MB, then a limitation of the VMM were hit for which increasing allowed Windows to support more memory, but increased the memory required to initialize the VMM, so obviously it became a tradeoff. Windows 98 required more memory anyway, so Microsoft decided in increase the limitation, allowing it to support 1 GB of memory as long as VCACHE's address space were limited to 512 MB by adding this setting to the [vcache] section: maxfilecache=524288. See [1] for details.) New extenders however, have higher or no limits (all 4 GB supported). Some DOS extenders do not provide access to virtual memory, limiting the application to memory installed in the machine, some of them do.

The article is about DOS extenders in general, not about a particular version of the drivers, or about specific features of particular versions of Windows. The above quoted paragraph is also too dense and difficult to decipher. Also removed several entries from the "Notable" section... the very idea of a notable program is that it left some mark on the PC application/gaming scene. The section should not be used for a dumping ground of DOS extended applications, although defining what is and is not a notable application is somewhat difficult to do while remaining neutral. X3J11 00:50, 25 March 2007 (UTC)

Well, even though I didn't initiate the idea, I did think adding at least one example for most every DOS extender was a good idea of showing who used it for their apps. If Wikipedia doesn't mention the existence of every DOS extenders, who will? They need to be chronicled somewhere, especially because some have bugs (Zurenava -> FreeDOS) or can be easily replaced (DOS4GW -> DOS/32A) or are still being updated (HX) or are abandoned but stable (PMODE/W or WDOS/X) or have their own quirks, advantages, and disadvantages regarding DPMI compatibility, virtual memory, DLLs, .EXE types supported, etc. Granted, a lot of that kind of info wouldn't necessarily belong in this particular article, but it should go somewhere (I hope).
Armslurp 04:56, 25 March 2007 (UTC)

I still don't think this is the place to dump every extended application known to man (figuratively speaking of course). By definition, the word "notable" means prominent, important, or distinguished. Applications that drove the business world, such as Lotus Notes and AutoCad are notable. Games that made a major impact such as DOOM and Warcraft are notable. Small programming languages and assemblers (Sphinx C-- and Lazy Assembler) are hardly notable, for example.

There should probably be a article for comparison of DOS extenders, then. I don't care about specific apps, but I did want to show what extenders are "out there", so to speak. Oh well, whatever. Armslurp (talk) 14:42, 11 December 2007 (UTC)

http://www.tippach.net/wdosx/ It is annoying since I was specifically looking for this not-exactly-rare one. I had to look in the comments here to find the name I couldn't remember. We need a separate article for a list, so as not to clutter this one up.  :( 67.190.106.206 (talk) 22:14, 31 March 2016 (UTC)

I also feel that the section on extender operation has once again been made too dense. While the information is applicable to the topic, it contains a lot of extraneous "junk" that is not essential and could be removed or moved to a more relevant article (the information on the limitations of Windows, for example, could be in the Windows article if it's not already).

However, I don't want to play an Undo wargame with the other people contributing to this article, and it's a bit disheartening to see my attempts at clarifying this end up for naught as someone comes along and reinserts all the stuff that I think shouldn't be there. :) X3J11 16:32, 22 September 2007 (UTC)

Some remarks

As far as the facts:

Wasn't Quake I compiled using DJGPP, usind CWSDMI as the DPMI host?

Did DOS/32A actually provide a Win32 emulation? I thought it was a DOS4G/W replacement? WDOSX did, though. I should know, because in a past life I did this WDOSX thing mentioned in the article. ;-) —Preceding unsigned comment added by 84.155.250.63 (talk) 22:35, 8 September 2007 (UTC)

Yes, Quake 1 was originally built with DJGPP using CWSDPMI. As for WDOSX, it was one of the best extenders I ever used, and if I were still trying to write DOS software, I'd probably still be using it. (So I guess that makes you Michael T, huh? =) X3J11 16:02, 22 September 2007 (UTC)

AFAIK, DOS/32A doesn't provide any Win32 emulation (or even most of the DOS/4GW API ??). It's just a smaller (faster?) replacement for the bloated DOS/4GW without swapping support (but supports more total memory). Armslurp (talk) 14:42, 11 December 2007 (UTC)

Windows extender?

Does anybody know how feasible it would be to make an extender to run 64 bit windows programs on 32 bit windows? Of course I could just buy Vista x64, but I don't want to deal with any unsupported drivers.

The main issues I'm thinking of are:

1. Still need separate 64 bit shared libraries

2. Registers not preserved on context switch (not a problem for DOS extender)

3. Need to switch to 32 bit mode before system calls (like switching to real mode in DOS extender)

Probably it would be same doomed short-life half-done stub like Win32s was for 16-bit Windows and Win-OS/2 —Preceding unsigned comment added by 77.91.196.70 (talk) 17:48, 27 February 2010 (UTC)

This is an old discussion, but still. I do know that Mac and Linux do utilize the method you describe on older versions of the OS, though a 64-bit processor is still required. In fact, in Apple's case all their version of OS X/macOS were 32-bit with 64-bit extensions until 10.6 "Snow Leopard", and even then the true 64 mode was optional. It's certainly possible, but given NT's source isn't available unlike Mac or Linux it will be a lot more difficult to accomplish. 2600:8801:2180:2E17:C186:73A5:91F2:2867 (talk) 19:38, 6 October 2017 (UTC)

Borland C++ and Pascal

Since Pascal 7.0 Borland passed 3 IDE builds: turbo.exe - real-mode IDE, very slow because of exnesive Overlays used, bpw.exe - IDE for Windows 3.x, and DOS-extender based BP.exe

Later they did the same for Borland C++ package.

Both packages included redistributable 16-bit DPMI server/DOS extender —Preceding unsigned comment added by 77.91.196.70 (talk) 17:51, 27 February 2010 (UTC)

80286, later 80386 – wasn't it the other way around?

About the second sentence:

DOS extenders were initially developed in the 1980s following the introduction of the Intel 80286 processor (and later expanded upon with the Intel 80386), to cope with the memory limitations of DOS.

Is this really the case? I read that VCPI came first, and VCPI was 386-only. Later Microsoft came along and developed DPMI, which worked on the 80286 Protected Mode as well. It was released into the open and VCPI was dropped in favor of DPMI, now also supporting the 286.

Am I right?

‣Andreas•⚖ 13:56, 25 January 2022 (UTC)

Long list not particularly useful

For someone mainly interested in the history it doesn't provide information on dates or historical developments, whereas for someone looking for a DOS extender to use it doesn't make it easy to compare them.

The line about Windows is also questionable, it presents the NT branch as exceptional even though in today's world NT essentially is Windows, and as I understand it the 9x branch basically used DOS only as a bootloader and 16-bit compatibility layer, doing the operating system nitty-gritty inside the 32-bit kernel instead of calling the DOS services. Depending on your definition of DOS extender they might still technically qualify, but it leaves a wrong impression with the reader if you aren't explicit about it. — Preceding unsigned comment added by 77.61.180.106 (talk) 11:48, 16 April 2022 (UTC)

Tyrian 2000 - Possible notable program (game)

According to the creator of the "Ancient DOS Games" series on YouTube, Tyrian 2000 not only was written in Pascal but also used a 16-bit DOS extender when the combination "alone is really weird. Most games at the time were using 32-bit DOS extenders."[2]

I would place it myself if I was completely certain that it would be a worthwhile game to note, but since I'm not, I'm leaving this on this Talk page instead of actually editing it in.

CtrlAltCuteness (talk) 17:13, 20 April 2023 (UTC)