Re^2: MS-DOS PD lex and yacc

[Charles Marslett]

In article <9288 at chinet.chi.il.us>, ignatz at chinet.chi.il.us (Dave Ihnat) writes:
> In article <201802wv4bX.01 at amdahl.uts.amdahl.com> dacseg at uts.amdahl.com (Scott E. Garfinkle) writes:
> >
> >I'll buy that for flex, but you must know that not only is bison decidedly
> >*not* public domain (though it is free), it is almost worthless for
> >most commercial endeavors.  I do *not* wish to start yet another running
> >debate on FSF's copyleft -- personally I use bison quite a lot, but always
> >for software to be distributed within the terms of the Gnu General License.
> >I only write this followup to ensure that someone does not naively misuse
> >the Gnu software. 
> 
> The gist of this is that, because the GNU copyleft is included with the BISON
> parser files, your application then must be redistributable under the same
> terms.  This is not true.  You might, at most, be responsible for providing
> the GNU parser include files as source, on request--in which case, you would
> most likely provide the entire GNU package.  But RMS and company cannot require
> you to release your code on the basis of their include file.

NO!  The LICENSE, not the copyright, that you must agree to asserts that the
entirity of the product containing GNU code is covered by the copyleft.  This
has been validated often in US courts (for a big dollar example, the enforcability
of licenses was a significant part of IBM's case against the Japanese mainframe
makers -- Fujitsu, et al.).

Secondly, if a significant part of a copyrighted work is included in your
work (anthology), without the permission of the author, that is copyright
infringement.  The author may extract any concessions he can from you (money
is a very common one in both publishing and software industries) and your
only option is to agree or not include his work.

> If this were the case, then the object code compiled by many commercial micro
> compilers would belong to the vendor providing the compiler, since many include
> a copyright notice in the object code.

It would in fact be the case, except that the commercial software vendors have
not demanded such a concession (and they would lose a significant part of their
business if they did!).  Most (Borland and Microsoft, for example) explicitly
grant you the right to do what you will with the result of merging your
object code with their run time libraries.

BTW, a discussion ran too long last year on the requirement Borland makes that
a binary containing Borland's runtime code must be copyrighted.  You can put
your name to it, or Borland's, but you cannot make the executable public
domain [This is to prevent the run time library from being made public
domain, I understand].  So such requirements, or distant relatives of such,
are not totally unheard of in comercial circles.

Charles Marslett
chasm at attctc.dallas.tx.us