[EMAIL PROTECTED] wrote:
> [EMAIL PROTECTED] wrote:
> > Ah, but see, if the API docs were released / reviewed / evolved with
> comments from the developer community prior to release of the
> implementation, then even this approach could work; an open source /
> non-Command implementation could conform to the same API as it evolves.
>
> That sounds like a cunning excuse to _not_ do it! :-)
Well, at least it's a cunning excuse. ;-) Really, I'm just trying to point
out the tension introduced into the whole process by trying to propagate a
new platform but trying to tightly control development of said platform in
a closed fashion at the same time. The inevitable tension is that the
vendor will have to prioritize, and inevitably that's going to leave some
people out in the cold. The two approaches that seem to work are: (a) let
the market drive the priorities, rather than some quasi-religious
technology vision (not saying Rebol's doing that, it's a trap I myself have
fallen into) or (b) let the market help implement the pieces they want
through some open or semi-open development effort. The alternative ---
leave the market to invent its own workarounds then trump them with
"official" solutions later on --- just isn't a reasonable alternative, IMO.
One last thought: it's all about the market, not about technology.
Technology --- tools --- are the enabler. The litany of companies that
have had great tech that ultimately washed out in the marketplace is
endless: Inmos with the Transputer, Atari's ATW, Commodore / Amiga, NeXT,
BeOS, the Smalltalk vendors, Apple's Newton, Scriptics with Tcl (which I
believe Sun originally bought in order to stifle Ousterhout while Java
gained popularity) even Sun with Java to some extent.
On the other hand many of the great marketshare-taking conquer-the-world
success stories of late --- DNS (the largest distributed system on the
planet,) Perl (the duct tape of the Internet,) Apache (the world's most
common Web server,) Linux, etc. all have in common an open approach to
development. Aside from a few high-profile niche successes --- Mirabilis
with ICQ, NullSoft with WinAmp and ShoutCast --- there are few examples of
closed boutique development actually igniting revolutions. (And those, by
the way, are apps --- a different beast entirely from tools and platform
tech.) (Netscape doesn't count because they really launched on the
shoulders of another, more open effort: Mosaic.)
I guess my summary point here is this: we're all interested in a
revolution. The strategic weakness of a closed development effort is
already being pointed out by the discussion on this list about what
priorities for Rebol ought to be. If we want our revolution to succeed, we
need to find ways to be responsive to the market without undermining
efforts (i.e., by making somebody hack together workarounds which become
obsolete in short order.) An open specification process, some internals
documentation, and an open extension mechanism in the core interpreter
would be one way to accomodate this.
>
> Andrew Martin
> ICQ: 26227169
> [EMAIL PROTECTED]
> http://members.xoom.com/AndrewMartin/
> -><-
