Re: STiNG, CAB, and Multitasking (strikes back) (fwd)

[Konrad Kokoszkiewicz <draco@mi.com.pl>]

> Then, tell me, why did you invite me in the first place to report on what
> I think to be a bug in MiNT ? If you're too busy to discuss this, ok,
> tell me and I'm gone. You've chosen to flame me instead, which will yield
> the same result, but will certainly not lead to me speaking about you very
> friendly on Usenet. It's been your choice, and I accept what you've chosen.

I have invited you because I think that a discuss between more people than
two is better to catch every aspect of the subject. The MiNT list doesn't
consist of me alone and I hoped that MiNT list members will either
convince you that there's no bug in MiNT or you convince them (incl. me)
that there is and it may be safely fixed. Which hasn't happened so far.

BTW. please stop frightening me with USENET coz it is childish ("if you
won't be a good child, I'll complain of you on the USENET"). I couldn't
care less.

> > other hand they know/realize/think, that this may ruin compatibility
> > between MiNT and some/most existing TSR programs they use. At least, they
> > may have some experience with MiNT, because this is the system they keep
> > using. Right?
> 
> Sure. And a sensible, productive, discussion had probably emerged if these
> things had been stated in a calm, discussion-like way. I've stayed calm,
> which Jo took as a chance to discuss in a nice exciting way.

The other thing is to be calm, and the other - to be arrogant. I've
pointed you out your lines those you should rather avoid to keep the
discuss at a reasonable level. But you didn't. And you for sure don't help
to keep the discuss at such a level by not reading mails to the end or not
replying to them or to parts of them.

> I have
> presented my arguments, you had your chance to reason against them, but
> dogmas like "exception vectors belong to the OS, don't touch them" do not
> help. I do not consider MiNT an OS, perhaps an OS extension, but I consider
> STinG an OS extension too, so why should I acknowledge that MiNT may touch
> vectors while STinG may not ? You've failed so badly on precisely what
> you're asking from me here : Consider the other side's position ! You've
> ignored mine, but ask me to know and take into account (even though I
> never subscribed to your list) that you're so busy with so much more
> important things. I go with a german saying : I'm washing my hands in
> innocence. Ya know what that means ?

Well, no. I disagree. You haven't presented any reasonable argument for
proposed fix. Additionally, you're apparently messing my mails up with 
Howard's mails. The point is not that it is forbidden to use exception
vectors, because many programs do and nobody seems to feel bad due to
that. Furthermore, it is not forbidden to use fault handlers. But it is
rather not reasonable to insist to enforce a fix what is considered
dangerous and not so advantageous.

Let's follow your argumentation, as I have remembered it:

1) first of all, you proposed to move the privilege violation vector
   in MiNT to the end of XBRA chain (with an unfounded supposition
   included in two subsequent mails that I don't know what XBRA is).

2) when it has been put into doubt due to inconsistency, you have
   proposed moving all vectors to the end of the chain. What, and
   everybody seems to agree, will be bad for compatibility with
   existing software (while you insist that it _improves_ compatibility)

3) when your usage of privilege violation vector has been called dirty
   hack, while you probably prefer to call it a clever trick, apparently 
   having no arguments to defense it you made an attack on MiNT structure,
   putting into doubt everything, even obvious things like why MiNT
   unliks all TSRs while terminating itself.

4) you have written that MiNT uses vectors inappropriately according
   to Atari documentation, while your usage of privilege violation
   fault isn't documented at all as approved method of programming and
   this concept is as "original" as several people independently but
   una voce called it a hack. And it doesn't matter if MiNT is an OS
   or not or if STiNG is an OS extension or not. No clean program
   should use methods like this, regardless of its status. And certainly,
   no system software regardless of its status should accomodate to
   such a hack. For now, as I understood, nobody agreed to it, but
   at the other hand several people (programmers) refused to agree.

5) you have ignored all messages saying that your method is inappropriate
   f.e. just because there's no way to recover when a fault will occur
   inside STiNGs handler by a strange accident, coz it is threatened
   by a processor halt. That's why it _may_ (it doesn't mean _does_)
   affect stability.

6) for test purposes (I agree that a test would be an ultima ratio
   for everybody) you offerred to write the code to be included into
   MiNT and when I asked you for the code, you didn't send it.

7) and you keep insisting that your fix is not STiNG oriented while
   that's the STiNG usage of privilege violation vector only causes
   the problem.

Is it _really_ a reasonable style of a constructive discuss?

--
Konrad M.Kokoszkiewicz
mail: draco@mi.com.pl
http://www.orient.uw.edu.pl/~conradus/
http://www.obta.uw.edu.pl/~draco/

** Ea natura multitudinis est,
** aut servit humiliter, aut superbe dominatur (Liv. XXIV,25)
*************************************************************
** U pospolstwa normalne jest, ze albo sluzy ono unizenie,
** albo bezczelnie sie panoszy.