Re: Libraries and repositories

[Tony Kimball <address@hidden>]

Quoth Paul Prescod on Thursday, 20 December:
: Gosling's language specifies a large library that you MUST implement.
: That's what allows you to assemble a large library of third-party
: utilies and systems on top of the platform....

Exactly.

: There are two major and a few more minor implementations of Python but
: they are highly similar because compatibility is a primary goal. The
: same goes for Java. 

Oh but the same does *not* go for Java.  Not at all.  Not one bit.
GCJ is incompatible with Kaffe which is incompatible with Chai which
is incompatible with J++.  Sometimes this is quite intentional,
sometimes it's resource starvation, sometimes it's human incompetence.  
Fortunately Sun's ploy has been successful enough so that all but the
most malicious parties at least vaguely aim for compatibility.
I seriously doubt that it has been successful enough to support a CJAN.

Now Jython is cute, and does nice tricks, but I seriously doubt that
it would play any significant role in any hypothetical CPyAN, which
does not exist, but I suspect more for lack of an interested and
motivated user community than because of the amusing footnote of
Jython.

: If R5RS Scheme had a sister standard that specified as large and rich a
: standard library as Java's API, it would not matter that there were a
: hundred implementations. They would be alike in every way that mattered.

That really wouldn't be necessary to make Scheme wildly successful in the
marketplace of ideas.  Just a practical basis set of interfaces for
the common devices of most platforms would suffice.  Maybe even just a 
portable FFI.

And parens wouldn't make one whit of difference.