[comp.sys.atari.st] dLibs for GCC

dal@syntel.UUCP (Dale Schumacher) (07/16/89)

[apratt@atari.UUCP (Allan Pratt) writes...]
> I expect I'll have some trouble figuring out what parts of GCC I can
> use in software for sale -- the library for the ST is a hodgepodge of
> PD stuff (derived from dlibs), GCC-distributed stuff (under copyleft),
> and the PML (portable math library) which is free but kinda slow, being
> written in C.

This brings up an interesting piece of news.  I've recently had some
discussions with some people who work with the Free Software Foundation.
The gist of our discussion was that there is a major outstanding problem
with using GCC for commercial work.  The libraries are covered by copyleft.
My response was basically, "So why doesn't someone just write a truly
public domain library?  I did."  That lead to the suggestion that dLibs
could possibly be the basis for such a library.

I've brought this up with Jawahr Bammi, since he is largely the force
behind GCC under TOS, and the maintainence of the corresponding library.
As I understand it, there is very little copyleft code (if ANY) left
in the TOS library for GCC.  The Minix library, on the other hand,
has quite a bit of code which is actually copyRIGHTed as part of Minix.

What I'd like to begin doing is the following.  Integrate Bammi's dLibs
changes into the current dLibs baseline library for TOS.  Expand the
functionality of dLibs (under TOS) to conform to dpANS X3J11 and as
much as possible, Posix.  Port the code to Minix and giving full Posix
conformance (if possible).  The resulting combined library should be
usable with little or no porting effort under many other *nix platforms,
and could conceivably be port to MS-DOS.  This library should also be
usable under Alcyon C, Sozobon C and GCC with TOS as well as Sozobon C,
ACK and GCC under Minix (at least).

> I'm glad I posted what I did, because it started this discussion and
> has started to set me straight about what you can sell and what you
> can't.
> 
> However, I'm still looking for definitive statements, not examples and
> guidelines.  What are the rules?  I'm not talking about distributing
> the compiler; I'm talking about distributing code compiled with the
> compiler. Pretend I didn't post anything earlier, just answer the
> question: under what circumstances can I use GCC to create a program
> without having to make anything but the executable available, under
> copyleft?

Code generated with the GCC compiler is *NOT* covered by copyleft.
Code generated which includes portions of the GCC libc *IS* covered
by copyleft.  Linking with other libc's (such as, perhaps, the one
which came with your *nix os) allows you to avoid the copyleft.
It should now be apparent why the above porting effort may have some
value to commercial users of GCC.

\\   /  Dale Schumacher                         399 Beacon Ave.
 \\ /   (alias: Dalnefre')                      St. Paul, MN  55104-3527
  ><    ...umn-cs!midgard.mn.org!syntel!dal     United States of America
 / \\   "What is wanted is not the will to believe, but the will to find out,
/   \\  which is the exact opposite." -Bertrand Russell