At 12:53 PM +0000 on 7/20/99, Mark Rauterkus wrote:
>If you do a subroutine and do NOT release its source in one of a variety of
>ways, -- then in a sense you "FORKED the tree" -- right? You made it such
>that there are now 2-copies of OC out there that are different.
No. The regression tests insure that they are functionally eqivelant. The
purpose of those subroutines is:
Operating system A has a library function. Let's call it "f()".
Operating system B does not.
Programmer C is porting the word of programmer D from operating
system A to operating system B
Programmer C writes a function to emulate "f()"
That's what it covers.
>
>What then? -- We all loose. That is the big NO-NO for an OPEN project. It
>makes for one of the flaws in the get-a-long, and is what really hurt the
>early flavors of UNIX, way back. Those are some mistakes from history that
>should NOT be repeated. Plain and simple.
No. Forking _must_ be allowed. Otherwise, you can get one group of people
dictating the entire project. If someone wishes, he should be able to say
"@#@& your dictatorial control!" and fork. He can call their product
something else, and if he writes a better version, the programmers will
come to his side.
Forking is something that _prevents_ problems. It allow people who have an
idea that is not valued by others proceed with their idea. And, for the
record, as far as I can tell, most good ideas are _not_ popular amongst
others -- until they have proven themselves. People don't recognize good
ideas very often.
Without forking, in order to create a better program, they would have to
start from scratch. And that's not OpenSource, that's OpenStupidity.
As far as I'm concerned, if we wind up with 3 OpenCard-derivatives, fine.
More choice is better. Less choice is Microsoft.
(Yes, I _hate_ the idea of preventing forks -- almost as much as I hate
AppleScript :)
