[openbeos] Compatibility Redux

  • From: "Paul \"Bo\" Peaslee" <h2oman@xxxxxxxxxxxx>
  • To: <openbeos@xxxxxxxxxxxxx>
  • Date: Mon, 23 Sep 2002 10:43:06 -0400

Axel, et al:

Been following the recent debate about OBOS R1 compatibility with BEOS R5.
A point that several folk seem to miss is that an R5 compatible OBOS allows
coders like me to develop apps for R5 now, with the knowledge that we won't
have to rewrite an app for OBOS R1.  It enables us to continue to enhance
BEOS, making it a more desirable desktop, perhaps even adding new BEOS users
before OBOS is even released.  Think of what it's doing for kit development.
It'll make 'selling' OBOS that much easier, too.  OBOS will be a familiar
entity that will not have to endure the pains of introduction. (More
appropriately, OBOS developers won't have to endure the pains of
introduction)

The goal of binary compatibility enforces a disciplined design approach.   I
think binary compatibilty forces OS developers to stick to their 'knitting',
ie code to a known and accepted design.  It [the goal] avoids a long
fractious design period, that would likely end in nothing being delivered.

Additionally, development of a binary compatible R1 will buy some time for
the design of R2.  R2 designers will be reined in too, because a large
deviation from the original design will not be easily tolerated.  Delivery
of R2 will be quicker for it.
Makes a great case for participation in the Glass Elevator project.

I wonder what success Linus would have had if he hadn't developed a UNIX
like MINIX derivative kernal.  Or, how far he'd have gotten if he had
ignored the contribution of the GNU utilities could make.

Great debate.  It's wonderful that everyone's so tolerant and open to
discussion.

Bo



Other related posts:

  • » [openbeos] Compatibility Redux