Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!bbn!mit-eddie!uw-beaver!uw-uw-june!pardo
From: pardo@uw-june.cs.washington.edu (David Keppel)
Newsgroups: gnu.misc.discuss
Subject: Re: Query: Status of Mach cleanout?
Message-ID: <9061@uw-june.cs.washington.edu>
Date: 23 Aug 89 16:09:39 GMT
References: <12720001@eecs.nwu.edu> <1311000001@upba> <9056@uw-june.cs.washington.edu> <407@eedsp.gatech.edu>
Reply-To: pardo@june.cs.washington.edu.cs.washington.edu (David Keppel)
Organization: University of Washington, Computer Science, Seattle
Lines: 34

>pardo@june.cs.washington.edu (David Keppel) writes:
>>...  A program that uses part of gcc (for
>>example is linked with the compiler) *is* derived from gcc and IF it
>>is distributed at all, it must be freely distributable.

baud@eedsp.UUCP (Kurt Baudendistel) writes:
>Please define ``freely'' as used in this paragraph.


In the FSF sense of `freely'.  That is, IF you distribute a `thingy'
that is or needs to be linked e.g., with gcc, then you must agree to
provide all recipients with sources for a `reasonable' copying fee.
You must also agree to let all recipients of the sources give away
those sources.

Case in point: NeXT created an Objective-C compiler that links with
GNU CC.  They were distributing .o files that the user linked with
`gcc' .o files.  FSF said ``under the copyleft, you must either not
distribute your Objective-C compiler, or you must agree to provide
sources for a nominal copying charege.''  NeXT will be giving their
Objective-C compiler to the FSF, to be incorporated in future releases
of `gcc', much like `g++'.

Note that LINKING something with `gcc' is different from compiling
something with `gcc'.  If you COMPILE something with `gcc', the result
is not considered a derived work.  You can do with it what you will.

(a) I hope this is clear.
(b) I represent neither the FSF or the lawyer's guild.

	;-D on  ( Or the liawyer's guild )  Pardo
-- 
		    pardo@cs.washington.edu
    {rutgers,cornell,ucsd,ubc-cs,tektronix}!uw-beaver!june!pardo