[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: STiNG, CAB, and Multitasking (strikes back)
Hello,
after my mail to the MiNT list about problem with STiNG being loaded
before MiNT I've got private mails from few persons with their opinions
about it. Getting those I've composed a comprehensive response for Peter.
Because the previous mail from Peter, describing the problem, was also
referred by me, not forwarded, to the MiNT list, it came to my mind that
perhaps it might be a source of misunderstanding. Therefore I would like
to bring the problem into discuss again, this time however I would like
also that Peter could personally demonstrate his arguments. Generally I
think by the way, that you, Peter, should send the problem description
initially to the MiNT list avoiding sending MiNT blames to the USENET, as
it also could be considered emotional (esp. lines about your willing to
support N.AES and unwilling to support MiNT, while N.AES works only with
MiNT).
Since this is intended to start the discuss again, I would like to avoid
giving my further suggestions about it. I have only two notices:
1) MiNT is not "my software" coz I didn't write it. MiNT is free,
available in source form. However, in fact, if MiNT and MiNT development
belogns to anyone, it is the MiNT list.
2) Atari declarations about "legal" and "illegal" ways of doing things:
a) are 5 years old now
b) are stated by a company that has been not interested in ST platform
since 4 years now
c) are stated by a company that is really dead since two years now.
d) refer mainly to TOS, not multitasking systems
e) when it comes to MiNT, may be changed by us if really necessary
considering new subjects of computing (networking, multiuser
systems, security etc).
Gtx,
K.
On Mon, 9 Mar 1998, Peter Rottengatter wrote:
> On Sun, 8 Mar 1998, Konrad Kokoszkiewicz wrote:
>
> > I've discussed the problem with some people involved in MiNT and Fenix
> > developments. The answer is: MiNT will not be "fixed" because the way
> > it is dealing with privilege violation handler is correct. MiNT is not
> > a TSR or an application program. MiNT is a loadable part of the operating
> > system and, being that, it is strictly integrated with the ROM kernel
> > and once loaded becomes an operating system. Due to that, its exception
> > handlers are and have to be primary ones. That's the first point.
>
> It is of course your right to define the exact position of your software.
> However, using this for a reason to refuse this change is quite silly and
> inappropriate here IMHO :
> - The change is not a big one.
> - It will not cause any compatibility problems with programs.
> - It will enhance compatibility with programs that hang in system vectors
> using legal, Atari approved (XBRA) methods.
> I see only reasons in favour of the change. Any argueing against it I can
> only assume to be emotionally loaded, which is bad for the subject.
>
>
> > The other point is that we won't fix MiNT for STiNG, because it would
>
> I've never asked to fix MiNT *for STinG* !!! I asked to fix it for all
> programs using legal methods to hang in system vectors.
>
> > signal other developers that your usage of exception handler is legal
> > and supported by operating system developers. Which is not. It might endup
>
> Sorry, but this is bullshit. The method I have described has been actively
> supported by Atari's Moshe Braner, and declared legal. There is nothing
> the MiNT community can do about that. Any attempt by it to "take it back"
> and simply declare it illegal must fail, and is in fact badly counterpro-
> ductive for the endangered platform Atari.
>
> > with a situation that we would have to include exception vectors to
> > context switches, and that is the thing we would want to avoid, because
> > of general system performance.
>
> I do not see in any way what my suggestion of a change has to do with
> context switches. I can only deduce that either you did not read my
> suggestion very carefully, or you reported it to your fellow MiNT
> developers rather inaccurately, or the developers do not know what
> they talk about.
>
>
> > The third point is, that any TSR, if it wants to install own handler
> > anywhere, is allowed to patch the original operating system handler.
> > But TSR like this has to be loaded after the operating system, not before.
> > Since MiNT is an operating system, you have to load STiNG after MiNT to
> > get it working. As you already solved it.
>
> But then it's bullshit to still recommend to run other software *before*
> MiNT, which is still done in the MiNT docs. If you really want to keep
> this, then you're doing things in a pretty inconsistent way. It is this
> that has caused the trouble in the first place, as people followed the
> MiNT advice and loaded STinG before MiNT.
>
> I do not think that providing an entirely correct AUTO folder sequence
> for any given set of programs is an easy task, when it comes to 8 or
> 10 or even more programs. For a newbie it is *very* difficult to achieve
> even if only two or three programs are involved. I therefore try to avoid
> all actions that require a particular sequence of programs. It cannot
> always be done. But in this case of MiNT, it is so easy to remove this
> contraint, see above. I cannot comprehend why you still should refrain
> from doing it. Are you keeping the knowledge requirements for installing
> MiNT artificially high ?
>
>
> I'm sure that my suggestions have not been really discussed. Maybe *some-
> thing* was discussed, but it had little to do with my suggestion then,
> otherwise your comment had not missed by point so badly (which is f.i.
> clearly indicated by the mentioning of context switches). I'm very sorry
> about this, as I had thought this could have contributed a little to
> easing the general problems MiNT always creates when system software is
> involved. I accept your reply, but you can be sure that I'll proceed
> blaming MiNT and calling it buggy when it comes to such problems.
>
> I intend to openly discuss this letter of yours in the newsgroups. If you
> object to that, please say so within a week.
>
>
> Cheers Peter
>
> ---------------------------------------------------------------------
> Peter Rottengatter perot@pallas.amp.uni-hannover.de
> http://www.stud.uni-hannover.de/~perot
> ---------------------------------------------------------------------
>
>
Konrad M.Kokoszkiewicz
mail:draco@bl.pg.gda.pl
http://www.orient.uw.edu.pl/~conradus/
** Quem Iuppiter vult perdere, dementat prius.
*******************************************************
** Kogo Jowisz chce zgubic, temu wpierw rozum odbiera.