[openbeos] Re: bedevtalk digest: August 17, 2001
- From: "Cedric Degea" <cdegea@xxxxxxxxxxxxxxxx>
- To: openbeos@xxxxxxxxxxxxx
- Date: Sat, 18 Aug 2001 23:36:22 +0200
>Subject: Re: Developers - which direction?
>From: "Michael Phipps" <mphipps1@xxxxxxxxxxxxxxxx>
>This is an interesting idea (rewriting BeOS).
>Open Source has proven its ability to take an existing API and
re-implement
>it.
>
>It would be a very interesting (and fun) project to work on.
>The modular design of BeOS would allow for this, in terms of replacing
>one piece at a time, much as we (the community, not me personally)
>have done with OpenTracker, OpenDesktop and mail_server.
this idea (working on genuine-BeOS first much like
BGA/Nathan did for the mail-daemon2) has interesting points
too; maybe as a complement to the "rewrite from top of Linux kernel"?
As a validation approach maybe: we know we can label a
piece of code "final" in the OpenBe source tree when the
same code compiled on genuire-Be passes a test-suite..
>app_server seems like one large piece that would be challenging enough
>to be interesting but small enough to be (fairly) quickly do-able.
hmmm... let's be realistic it would be a huge huge piece
of work. I remember the Be Newsletter back in R3 daysmentionning
a "rewrite of the app_server that had" made it shrink by
10.000 lines, so how long it is is anyone's guess but it coul
d be 10x or 30x that many lines or more. According to JLG
BeOS was 5 millions l.o.code, back then...
Plus,
>we could implement enough new features to make it worthwhile
>(multiple monitors, new widgets, etc).
that's one very encouraging aspect of this silly-sounding
project after all: by starting from an existing kernel
we benefit from mmap() ..etc that was lacking for all those
days to BeOS in order to port some POSIX apps, we can "improve"
kits as we re-implement them while still being source-compatible
with everything on bebits.com..
>print_server is a place where BeOS has always been weak, due to lack
.
>BSD (X N) , NewOS, Darwin, Minix and other kernels for X86 exist in the
O.S. world.
>
>I dunno. This sounds kind of like fun. Anyone else really interested?
Talking about the only one existing project I actually know about,
there are 4 people in the blueos team yet.. semi-experienced
people including the author of a sound card (whose name escapes
me) driver, and the lead dev of one of the top 5 GFX drawing
apps on bebits (IMHO) so it's a good start but it'd obviously
be in need of gurus/hackers in varius areas and have its "staff"
multiplied by ~10 or so.
>>> AtheOS, even fun?
>>
>>maybe if all else fails.. and I get semi-blind after an accident..
>>who knows. The API is the closest, though I dislike how the
>>author both ripped it and at the same time claims he has nothing
>>at all to do with this ~"BeOS thinggy people keep mentionning to him".
>>Ideology should not be tangled with OS technology like some
>>other OS is more a religion than a piece of code, but there's
>>a threshold not to go over :-)
>
>I can't speak to how much was ... ripped and how much was coincidence.
>To be honest, before I saw BeOS, I sketched out an OS design (on paper,
>and I still have it as proof!), and someone who was skeptical would say
>that I ... ripped... some of Be's design. Awful hard to design a
modular, OO
>OS too many different ways.
ok I see your point.. and my statement above is contradictory
to my "leitmotiv" that the BeOS API (at least the core parts)
just is how an API has to be, period, and any OS that would
start from scratch would have to look similar for me to even
look at it blah-blah.. So i'll stop contradicting myself here,
maybe Kurt did not do a clean-room re-implementation by just
looking at the headers but actually came up with the same
ideas (and class names) on his own simply because in this day
and age you have to apply design patterns the correct way (I
was stunned to see in a web M.I.T course that the BHandler thing
is actually a "delegation" design pattern, didn't know about
this one).. Thinking about it
maybe I get pissed off only because of his not willing to
be like BeOS, reacting like if it was an insult that people
suggets so... but then again i'm getting emotional and
it's not a profesional attitude <g>
Ok enough rambling from me on this topic.. I'm being counter-productive.
cedric.
>>Also (disclaimer: more personal opinion),
>>I wish I could say that the OpenSauce ideology is the big
>>winner of this sad ending, but I fear it's not even partially
>>true and MS is really a bit stronger today: Linux has actually
>>lost LokiGames, VA L., Corel, Dell support, and others in the
>>last few months, for a start.
>
>I agree - if you are in software for money, OpenSource (esp GPL, where
the users
>seem to be far more ... radical) is not the place to be.
--
http://cdegea.free.fr/ | BeDev E-16870
"God exists, and she loves Bill" -- BMessage
Other related posts:
