中国DOS联盟

Today I saw that UIDE has been updated again. After admiring it, I couldn't help but jump out to speak. First, I declare that I am not an expert, and I am also a long-term lurker learning in the forum...

0. When I see that authors of first-class programs like UIDE, UMBPCI, etc., write programs, they all mark how many bytes of memory are needed. Although it is limited by the system, it also reflects the difficulty of development from the side.

In addition, from the development of these great people, I also have some ideas about DOS development:

1. Modifying the kernel is basically not reliable (not that I'm pouring cold water). The biggest problem is that DOS is owned by Microsoft, closed source, and there are many restrictions (let's forget about the MSDOS 6.0 source code); personally, I feel that Wengier's work can only stay at the HEX EDIT level; and compatibility issues are even more of a mess. DOS has its own mission and limitations. It is impossible to require it to be 32-bit and multi-tasking, etc. Even if you do it, you have to find programs to run on it. This is not just technology. Then I think FreeBSD, Linux may be more suitable for you. Of course, there is still prospects for doing industrial control, embedding, etc., which are not in this category.

2. DOS ain't dead, but it won't become very glorious either. The reason is as in 1. In addition, it is said that those motherboard dealers have agreed that they will not provide compatible BIOS in the future, and only need to be able to enter DOS GHOST. Oh my...

3. The practical approach. I think the most practical approach is still like those people such as Jack Ellis, japheth, and mplayer for DOS, to see in which places we can enhance the functions and usability of DOS, and do more practical things.

In addition, this forum seems to have a feeling of being flooded. The popularity is very high, but should there be a way to make everyone try to post less water (of course I also occasionally can't help but post water), so that useful information is not hard to find...

Bored and can't sleep in the middle of the night, talking a lot of nonsense, don't take offense if I say something wrong.

Although I also agree that more practical programs should be developed, I still find it difficult to fully agree with some of the owner's viewpoints.
I think that if DOS does not evolve to 32-bit, it will definitely be eliminated. The current 16-bit DOS has too many restrictions and is simply not sufficient for developing modern programs, especially in terms of hardware support.
The 16-bit one will definitely be eliminated in the future.
It is of course impossible to modify MS-DOS, but it is completely possible to support and participate in the development of FreeDOS and FreeDOS-32.

Originally, I thought I had too much saliva and looked very cuo, but I still couldn't help cuo-ing again.

FreeBSD has come from joy's BSD UNIX. This system is old enough and mature enough, right? Now it has a function of GNU/Linux Binary Compatibility. Why is that?

The developers of those heavyweight software (Adobe Reader, Mathematica, Matlab) only develop for GNU/Linux because they are the most popular; FreeBSD? They simply don't know how many people are using it (the above is excerpted from FreeBSD handbook 10.2, let's let go of the rough parts). The efficiency of FreeBSD 7.0 was better than the mainstream LINUX kernel at that time (there are evaluations), but LINUX is still the most popular. This is what I said is not just about technology. Here, I use the example of BSD VS. LINUX just to question the prospects of freedos 32. What if freedos32 can't run those important software? Build a linux/win32/freebsd/svr4 binary compatibility?

In short, my point is that no matter what OS it is, it has its own place. I also like DOS very much, but I can't ignore its proper place because of this. Forcing it to replace others' places, then what is developed is not DOS, but the latest OS.

Of course, I'm not good at development. If you all really do it, I'm also willing to follow and learn with my butt blacked.

In addition, I would like to add that as everyone has seen, FreeDOS has been inactive for such a long time (FreeDOS is currently at version 1.0, released on September 3, 2006. Wikipedia), this is a fact. Even if it is under development, how many people take it as the main system?

UNIX was originally 16-bit and written in assembly. Later it switched to C and evolved into 32-bit. Can you say it's not UNIX? When UNIX first came out, the CPU didn't have protection mode, so it was actually just something similar to DOS.

Win9x/ME have 16/32-bit kernels (with part of it being DOS), while the NT-based 2k/XP/Vista/7 are completely different from them. Can you say they're not Windows?

DOS isn't a very bad operating system. It's compact, simple, and has good real-time performance, with many advantages that other systems don't have. Even if there are many aspects that are not good, there's still a lot of room for improvement. The key reason why DOS has stagnated so seriously is Microsoft's monopoly and lack of ambition to not make DOS 32-bit. Its interest is completely in NT. Also, UNIX was originally a commercial product, and Linux, as an open-source unix-like system, has developed rapidly. By contrast, FreeDOS, as an open-source DOS, has developed extremely slowly. UNIX has never been the mainstream on the desktop, while Linux has developed like this; DOS was once so popular, but FreeDOS has developed so slowly. There must be a reason for this.

Also, please note that what I'm saying is exactly continuing the topic of "developing software" mentioned by the original poster. I certainly agree with developing practical programs, but I think we must also pay attention to kernel development because the reality is that the 16-bit DOS limitations make software development extremely difficult, and even some software can't be developed in theory. The AC97 driver that was very concerned about on the forum is an example. The problem now is that it can only exist as part of a 32-bit program and can't be written as a real DOS driver because DOS drivers must become (equivalent to) part of the kernel in the form of TSR, and 32-bit TSR can't be implemented. With new hardware and technologies emerging one after another, DOS can't be practical at all if it doesn't evolve to 32-bit. Except for embedded and some very low-level work, it's useless elsewhere.

To take a step back, now 64-bit CPUs are starting to be popular, and it's estimated that 64-bit operating systems will soon be popular on the desktop. Not to mention that DOS's 16-bit is seriously outdated, it's completely possible that future CPUs will completely abandon support for 16-bit software. At that time, if DOS doesn't develop, it will completely become an exhibit in the museum.

Regarding the support of 32-bit DOS for 16-bit applications, it can completely be implemented using virtual machines, just like Windows does, and it can be done much better.

The original poster may think that developing the kernel is not realistic at present, and actually I also think so; and at the same time, I think that developing software without developing the kernel can temporarily solve some problems, but in the long run, it's not of much significance; the reality is so cruel. If no one does anything, DOS will only face extinction. I'm relatively pessimistic about the future of DOS, even though I personally like DOS very much.

Last edited by brglng on 2009-5-19 at 11:30 ]

The hard disk is approaching the limit of the traditional partition table of nearly 2T, and desktop operating systems are facing a turning point.

At present, the situation is that many capture cards have turned to Linux and Windows, with drivers provided by manufacturers, and development is反而 simpler; embedded systems and machine tools based on Linux and Windows are everywhere; high-speed devices use high-precision timers which are more accurate than the 8253 and 8254 compatible with the 8086; the streamlined version of Linux is only a few M, and PE is only more than 40 M, and it can run smoothly by discarding most unnecessary modules. DOS may be at the time when it's time to say goodbye.

The only possible reason for DOS to exist for a long time is: BIOS is not completely eliminated. Even after EFI becomes mainstream, it is backward compatible with BIOS. Another hardware basis is that the X-generation speed of SATA reaches a qualitative leap, and the factory BIOS settings are no longer compatible with the IDE mode (the environment that int13 must rely on). Only then will DOS come to an end. So for now, DOS still has its place in hard disk-level maintenance. In addition, the boot and versatility of LINUX or WINPE are not yet mature. At this time, the statements that deny DOS are unfounded.

Last edited by gmy on 2009-5-24 at 03:28 ]

After reading gmy's introduction, so I think DOS needs to be developed more, with the goal of adapting to new hardware environments as the development target.

Originally posted by brglng at 2009-5-9 07:00 AM:
I think that if DOS does not evolve to 32-bit, it will inevitably be phased out. The current 16-bit DOS has too many limitations and is simply insufficient...

You can modify FreeDOS, but I have looked at the Kernel's CVS before, and there are too many details!

Just organizing that source code would take a lot of effort. No one has been willing to do it for several years, let alone talk about doing it. Few people are even willing to talk about it...

Recently, there are some new test versions of the Kernel, but they are still not 100% compatible. You can tell by the difference in CONFIG.SYS between the two.


Last edited by johnsonlam on 2009-5-29 at 13:50 ]

After reading everyone's discussions, I can't help but want to express some opinions. What I want to say is why do we always talk about the kernel-level development of DOS? Personally, the feeling is that if we have the ability to modify the kernel, why not redevelop an OS kernel? Why must we be entangled with DOS? (Just because DOS was once glorious and very classic?) Why do we always stay away from modification or secondary development activities? Just because DOS is useful (temporarily or forever), just because modifying someone else's kernel is easier than redeveloping it (is it definitely easier)?

Maybe someone will say, do you think developing an OS is easy? Yes, developing an OS is not easy. It is at least a large-scale system engineering, and maybe even a key project of a country. But then again, if we don't dare to go in this direction, we won't have any opportunities. Now the 64-bit architecture has come out, and 64-bit is definitely the future trend. So we can also consider the 64-bit direction and make a 64-bit kernel. Even if it may not be useful for a long time now or in the future, but as long as our kernel is well-made, the architecture is flexible, and the efficiency is high enough, one day someone will use it and improve it. Wasn't Linux popular like this back then? And we don't need to argue too much about which is better between DOS or FreeBSD/FreeDOS, Linux, these OSs can all be used for reference and for our own use. After all, the future is a diversified world, and each different OS has its own suitable positioning. Well, after talking so much. Actually, I just hope that one day we will have an OS that we can be proud of, hehe