On Mar 26, 3:01 am, problems@[EMAIL PROTECTED]
wrote:
> I think I read sometime ago that [perhaps unix/C] designed a set/s of
> library/s, intended to be a minimal but sufficient set to do 'any' task
?
> But that it is not widely accepted.
> So eg. for strings, you might have:
> * return position of substring in string, else zero;
> ...
> * insert substring in string at position;
> ... etc.
> I imagine programming then being mostly a matter of
> selecting procedures/function out of the library ?
The thing is, when you have that low level set of libraries, there can
still be a lot of design work to be done to get a set of capabilities
that works smoothly together for a particular task. And when that
design work is done, and put into a library in turn, people will start
borrowing from that higher level library for something where it
*almost* fits, and often add in the hook needed to make it fit better
than it did originally ... and that higher level library starts to
take a life of its own, based on the history of uses to which it was
put.
And somebody comes along and say, "look at this pile of encrustations
and barnacles that have been added to make this thing a jack of all
trade, master of none", and goes about rebuilding a library from
scratch to do one thing and do it well (and of course under their view
of what "well done" counts for in that area), and around it goes.
And of course, maybe they are a firm believer in the canonical library
system and rebuild on top of the canonical library, and maybe they
aren't, and they build from scratch, or on the basis of some other
library that they think is more effective for their area of concern.
However, its clear that effective library browsing technology is going
to be useful in any part of that library life cycle.
Have you looked at how one of the commercial Forth's goes about it?
Both of the commercial Forth houses have been working with the task of
providing effective access to their bundled libraries for quite a
while now, so they have certainly learned some lessons on usability
along the way.
|
