Re: Paging Beth Stone


Richard Cooper <@win***> wrote:

| > Why don't you/we all make up a list of what an OS should support
| > in Low-Level ASM-terms and create an ALA-OS which fits all the demands.

| Damn, I had to hit Google to figure out the ALA in ALA-OS meant...

:) sorry for I assumed it a known abbreviation.

| Some time ago I discovered MenuetOS. I thought "this is awesome" and
| then I looked at the API and thought "what the fucking hell!"
| I would have been content to leave it at that were it not that every
| page on the site says "MenuetOS - The Assembly Lanugage Programmer's OS"
| and I figured that anyone making a statement like that about such an
| awful OS design needed a good talking to.

| So I went to their forum and posted this:
| http://board.flatassembler.net/topic.php?t=2814

I would hear many similar critics on my KESYS-API,
as it is designed to fit just 'my' targets.

| In the end I had a little discussion on OS design going on on my web
| site. It was going pretty good for a couple of weeks until someone
| started discussing how the GUI should handle hotkeys, and at that point I
| guess everyone lost interest, and I went on to work on some Z80 stuff.

It seems very easy to loose the 'order of importance' and most folks
will give up early when it comes to define details.
Many good ideas/projects died on this already.

| I've thought about making my own OS since then, but I'd have to make my
| own assembler or compiler or something to program under it with, and I
| have a phobia when it comes to that sort of thing.

| Even so, I did manage to make a pretty nice assembly parser in NASM,
| but then I got to looking at the effective address encodings and that
| about made me sick. I hate parsing user input.

I don't think it needs new tools, most of the assemblers around
can be modified to work for any API.
Even my API would not support the stack-polluter methods,
so all the INVOKE-API-macros will become redundant.

| Even so, reading the Intel manual, it's not entirely obvious to me how
| Intel ever intended someone to make a multitasking OS with the features
| they put into the 386. The best idea I've come up with is to keep the
| nested task bit set in the TSS of all tasks (except interrupt tasks), and
| make an interrupt task that, when called, modifies the return task field
| in it's TSS so that when it does it's iret a different task will run.
| However, I can't find any indication that that's how Intel expected you
| to do it.

Me not either, looks like Intel designed it for Billy's HLL-crap only.
I found only two reasons for using the slow hardware task-switches:
Stack-faults and Debugger. Software multitasking is much faster than TSS.

| It all seems to be a few inches over my head.

It is almost impossible for a single persons to code a complete
general purpose OS alone.
And as non-M$ OS-developers are very individual folks,
I'm afraid the idea of an ALA-OS died before it even started.

__
wolfgang


.