#142 Re: [osFree] OSFree and our future
Expand Messages
criguada@libero.it
Feb 21, 2002
Hi Adrian,
first of all, please understand that I understand what you mean, and I
would agree with you if we were not in this particular situation.
Second: please (to all), don't start building walls and fences (I won't
do that, that developer won't join, etc.), the OS/2 community is already
enough divided, and with few people in it this could be deadly.
> >When OS/2 2.0 was designed and a few years ahead the team had one
> >of the largest companies in the world backing them up. Also OS/2 was
> >at thet time extremly important for IBM.
> >I heard quotes that IBM spent almost $1 billion on OS2PPC !
>
> so what? Linux wasn't done like this and Be neither I guess (with money
> but with much less).
I know, and I agree with you. The point BTW is that Linux was built upon
an ultra-known design. I, and every other university student with a
background in informatics, have studied the unix design in the classes
about Operating Systems Design. Every student, given enough time, could
write a simple unix kernel in C. Most have done it for their exams.
Linus did more or less this, perhaps toying a little more with his
creation. There was NO other design after his kernel, and his design
wasn't even really smart (simply because the original unix design was
smart only if compared to the available non-mainframe OSes at that
time... and there were very few).
Designing a good OS today from the ground up is entirely a different matter.
Another consideration: on the FreeOS list, the two "leaders" have been
advocating this kind of approach (design first, coding after) from the
beginning. Look what happened there. The FreeOS list started in 1999,
they have produces a pile of words, and not a single line of code.
Worse, not even a line of DOCUMENTATION!
They were all debating over this or that design, and did ABSOLUTELY NOTHING.
Now, the two cited leaders have this theory: "design first, otherwise
you'll have to face reimplementations".
Well, if you aren't an enterprise you can't afford the design of such a
big application as an OS, simply because people get bored about
designing, and because without a central and strong guide people only
goes on and on debating over the "best" design.
But you can have a good product (and good design) if you're not scared
about reimplementing. That's it: start coding, but keep as open as
possible to change. When you learn enough to think that you didn't take
the right approach, REIMPLEMENT without fear! You'll discover that the
reimplementation will take a tenth of the original implementation, you
will have a better (or "the best") design, and it will be done earlier.
NO AMOUNT OF FORETHOUGHT will give you the necessary mind-clarity about
what you're gonna do.
> Linux was started more than 10 years ago. No doubt that it will take
> long but I prefer to have something that works instead of something
> that works fast but does not perform in the future.
If you have an open source, and you keep sufficiently open while
developing, this is not a problem. Even Linux kernel have changed
radically from the start of its life.
> I personaly doubt that it makes sense to invest time in tools before we
> think about a design. Beside this the JdeBP tools and also GNU utils
> were mentioned already. That's enough about tools IMHO.
The GNU utils don't cut here, and regarding JdeBP, if he doesn't join
the project his work cannot be integrated. Have you ever heard about an
OS that does not have utilities because you have to download another
package from another author?
Sure, Linux uses GNU utilitites, but those are opensource, while JdeBP's
utilties are not.
> Believe me you are wrong. Money definitely counts as well but without
> *very* good relationships to IBM eCS wouldn't exist.
And in fact, eCS isn't what it could be, because IBM is too blind to
give their complete sources to Serenity.
Bye
Cris