🚀 go-pugleaf

In article <1992Nov6.175825.2978@ampex.com>, leepeng@ampex.com (Peng Lee)
wrote:
> 	In software, I believe X-window protocol, MS-Window API, DOS
> 	interrupt functions are good examples of some very successful 
> 	reuseable software components.   If we agree on this, maybe 
> 	we should start analysis why these APIs succeed. 

Well, I have to advertise for the Macintosh Toolbox, appeared 1984
and still going strong! It's really staggering actually to think
that people are using the same window creation routines then and
also today. 

Note also that when we made sure that the toolbox was the *one and
only* way to access the system, we gained two things:
a) uniform application look&feel, maybe one of the prime reasons
for our fame in user friendly computing
b) control so developers would not do skanky things behind our backs
(even if they do it now and then), like re-defining the look&feel.

The main idea is to make the use of the toolbox API (or component)
so compelling, that it's not worth the time and energy to rewrite
the functionality. It is true that we failed with some of the 
features, and developers immediately rewrote these.

Another aspect is to look at very clean and elegant ways to override
built-in functionality. In the toolbox case we allow developer to patch
into the toolbox calls, either before the call (preferred) or 
after (should be avoided, because we also patch for fixing problems).

Anyway, we have a long history inside Apple concerning how to make
developers reuse functionality. The interesting observation was that
this strategy worked well, but you just need a very good and compelling
reason. The evolving time-to-market cycle that is slowly decreasing to
> 1 year is IMHO one of the reasons why component based development
work will gain followers.

Cheers,
Kent
-------------------
Kent Sandvik (UUCP: ....!apple!ksand; INTERNET: ksand@apple.com)
DISCLAIMER: Private activities on the Net.