🚀 go-pugleaf

>>>>> On 7 Sep 90 02:30:39 GMT, rpw3@rigden.wpd.sgi.com (Rob Warnock) said:

RW> In article <NELSON.90Sep2223753@image.clarkson.edu>
RW> nelson@sun.soe.clarkson.edu (Russ Nelson) writes:
RW> +---------------
RW> | In article <68378@sgi.sgi.com> karsh@trifolium.esd.sgi.com writes:
RW> | > FSF hasn't really invented anything nor authored a new user
RW> | > interface as far as I know.
---

new interface: Gnu G++ library.

This is the an important ``interface'' in the current Gnu software. If
I, Joe Programmer, want to sell bundled Gnu software, I can invoke the
user-does-the-link for e.g., gnu libc, because all libc's need to look
basicly alike, or they're useless. Thus, you could claim that the user
could link with *any* libc, not Gnu libc.

However, when using Gnu libg++, you're depending on an encoded
interface to libg++ (the typesafe linking convention mangles names)
and you're admitting (because libg++ is very different from most other
C++ libraries) that you're depending on libg++.