Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!iuvax!cica!ctrsol!ginosko!uunet!pilchuck!dataio!bright
From: bright@Data-IO.COM (Walter Bright)
Newsgroups: comp.lang.c++
Subject: Re: libg++ and copylefts
Message-ID: <2063@dataio.Data-IO.COM>
Date: 19 Jul 89 19:10:42 GMT
References: <799@redsox.bsw.com> <6590194@hplsla.HP.COM> <318@gt-eedsp.gatech.edu> <2053@dataio.Data-IO.COM> <42888@bbn.COM>
Reply-To: bright@dataio.Data-IO.COM (Walter Bright)
Organization: Data I/O Corporation; Redmond, WA
Lines: 37

In article <42888@bbn.COM> jgrace@porter-square.BBN.COM (Joe Grace) writes:
<In article <2053@dataio.Data-IO.COM< you write:
<<	Redevelop libg++ from scratch. Explicitly make it *PUBLIC DOMAIN*.
<<	This would solve those nasty copyright problems.
<I'm not sure this approach would really work in general since GNU can
<take software from *PUBLIC DOMAIN* and incorporate it under their
<copyleft restrictions.  Someone pointed out how FSF seems to have
<taken some government subsidized code, copylefted it, and taken over
<maintenance of the code --- thereby killing the code's *PUBLIC DOMAIN*
<availability and utility.

You can copyright a work derived from a public domain work. However, when
it's copyrighted, you must state on the copyright registration form what
pd work it comes from. Otherwise, you are lying on the form.

Copyrighting a derived work doesn't change the status of the original
work. What you've actually copyrighted are the diffs.

There's nothing stopping *anyone* from reimplementing libg++ from scratch
and making it pd. Neither GNU nor RMS can stop you. The only thing they
can do is not put it on the tape that they send out. The pd version of
the library can be posted on this net.

Why don't I do it? Well, I'm in this business for fun and profit. I sell
source to C and C++ libraries.

Most library routines for C and C++ are pretty obvious and straightforward
to implement. The exceptions are qsort() and the transcendentals. The
transcendentals can be produced fairly easily following the cookbook
approach described by Cody&Waite. qsort() is a *bear* to do. Note: all
magazine versions I've seen suffer serious problems. My friend Joe Huffman
who wrote my qsort() is able to cause most compiler vendors' qsorts to
crash. (He implemented it based on somebody's PhD thesis on qsort!)

If you're doing a project and want to use GNU C but not their library,
you also can buy library source code from any number of vendors (Turbo C,
MSC, Zortech C, etc), port the library, and link with GNU output.