New OS

[M-Saunders]

Lo,

It is a leap, but not a jumping-over-K2 type leap. Basically, when you switch to protected mode you lose access to the BIOS. You can't do 'int 10h'-type calls to print stuff to the screen, grab keyboard input or read sectors from a floppy disk. You need to write device drivers for these things yourself!

That's a lot of work, and you can get bogged-down in tedious technical gubbins when you'd rather be doing something cool like making a GUI. This is why, in MikeOS, I've stuck to 16-bit: it makes the OS pretty limited but means the codebase is quite clean, and not full of complicated drivers for floppy drive controllers etc.

If you look around the osdev.org forums, you'll occasionally come across a keyboard, mouse or floppy driver that someone has written for the bkerndev kernel. That'll help you get on with the cool bits and bobs.

M

[Dark]

I played around with Mike-OS on QEMU and I have to say it was a good OS. Better then any Microsoft program (Which just about any open source thing is) The only thing is the game was like in German or something O_o

[M-Saunders]

Yes, I think it's Czech. Whatever the case, it's a sub-32K DOS COM game, the kind that should run on MikeOS with its simple DOS compatibility routines...

M

[kilikopele]

I think I'm starting to grab it now. It's a bit like studying fractals--the more you zoom in on them, the more you realize there is to take in.

I think I'll keep my comfortable seat in USER-land for a while longer.

[M-Saunders]

Yes, well put. It's like climbing a mountain for the first time -- you think you're near the top, but then more obstacles appear in your view as you get closer! That's why I think it's best to head lower: making a full OS requires so much time investment that it becomes boring, but making a mini OS as a pet project is great fun. If you grasp the basics of x86 assembly, it'd be cool if you could chip in with MikeOS!

As you may have seen, assembly language is tough but understandable. With the help of the BIOS and copious use of CALL/RET and PUSHA/POPA, you can write modular code in x86 assembly, and MikeOS uses it to the hilt. There's lots of stuff that you could add with a basic grounding in x86 -- text-based dialog boxes, date conversion routines etc. I'm heading for a 1.0 release soon, with documentation galore...

M

[Dark]

Hey on Mike-Os do you think that you could include a text based (Or Graphical) Web browser or a text based IM client? Because that would be cool (Im not really sure how DOS and similar systems work Nor BIOSes and QEMU so im not sure this is even possible....)

[M-Saunders]

Well, it's technically possible, but the amount of man hours involved would be insane. First we'd need a TCP/IP stack (several months), and then drivers for various Ethernet cards and USB modems (more months). Imagine writing something that meets this specification, in assembly language:

http://tools.ietf.org/html/rfc1122

Pretty heavy going! So unless I can magically tie in some kind of DOS TCP/IP stack, I don't think we'll have Internet support on MikeOS. With a terminal emulator and serial modem, you may be able to dial up an ISP and access a remote shell -- but I don't know if any ISPs offer that nowadays. I remember Amstrad PCW fans getting online in a similar fashion...

M

[Dark]

Yah I guess that WOULD be unlikely. TC/IP stacks must be a pain to code in assembly. And drivers probably would be nearly impossible (Due to the lack of some in Linux which is coded in non-assembly)

[Rhakios]

I always used to be amazed by the QNX single floppy demo, a GUI with a web browser, serial modem dialler, a notepad-like app and a game. All in 1.4MB - so how hard can it be Mike?

[M-Saunders]

Give me a 30-strong development team and a few million quid, and I'll tell you

Contiki's TCP/IP stack may be a candidate for porting one day, as it's supposedly very compact and well-documented. But that'd still leave the dialup-modem-only problem...

M

[Rhakios]

Well, as long as you can implement a pcnet32 driver I'll be happy.