| From: | Sam Jordan |
| Date: | 08 Dec 99 at 17:50:00 |
| Subject: | Re: [warpup] 68K emulation |
Am 08-Dez-99 schrieb David LAROCHETTE:
>Sam Jordan wrote:
>> No WarpOS hasn't. That was basically nothing more than a little experiment.
>> I did never hear of any software allocating protected memory (right now
>> I also wouldn't recommend doing it) and even if a few would do it,
>> most of the memory still wouldn't be protected. It's simply of no use.
>> Sorry, this has nothing in common with *real* memory protection, which
>> I anyway don't expect to see in AOS, because of some fundamental
>> conceptual problems with AMIGA software programming (like passing
>> data inside messages by reference instead of by value).
>So if you ever develop a PPC AmigaOS, do you intend to implement some
>sweeties like Memory Protection, Resource Tracking, Garbage Collectors
>and so on ?..
There are several issues which have to be considered when talking about
a potential AOS PPC project:
a) It must be done in a reasonable amount of time. That's simply because
the project would loose any commercial sense, if development would cost
heavily much time, so that the OS would only appear in years and would
then still cost very much. Therefore the only reasonable approach for
this project is not to try to add too many new concepts at a time,
first it is simply necessary to do the port of the OS (which is still
a real challenge, especially when realizing that AMIGA software companies
can't be compared to software companies on PC/Mac or other
architectures, concerning financial background or manpower).
b) It is absolutely necessary, that source level compatibility is kept
as high as possible, because an AOS PPC project can only be successful,
if the AMIGA developers can be convinced to port their software to
AOSPPC. And they will only do that, if it is possible to do it quite
quickly without too much effort. If now some fundamental concepts are
changed inside the AOS, then the developers would also have to change
fundamental concepts in their software, and this doesn't make sense,
especially for commercial projects. And such things like changing
the message system to make it MP compliant *is* a major fundamental
change of concept.
Because of these two (and probably more) reasons, fundamentally new
concepts have to be shifted backwards. Of course, in future some of
these concepts might be implemented, but I guess that's still far
away.
We simply have to realize, that we can not have all at the same time,
but we should be happy, if we at least make good progresses, and
I consider a PPC port of AOS (and making it running on other
platforms) as a very important progress.
bye