Copyright on own implementation of Standard C Library

Question

I am busy writing a bare-bones Standard C Library for my hobby OS. I am only including the functions I really need for now but the library is still referenced as 'libc' in my source and I plan to expand it to a fully featured C Library. My OS is going to be released as open-source so I would like to know what kind of copyright notice I should put on the source files for my 'C Library' because the coding is mine but it is still based on the source of other C libraries such as GNU C.

I have seen this copyright notice on a C Library implementation for AVR

00001 /* Copyright (c) 2002, Marek Michalkiewicz
00002    Copyright (c) 2004,2007 Joerg Wunsch
00003 
00004    Portions of documentation Copyright (c) 1990, 1991, 1993, 1994
00005    The Regents of the University of California.
00006 
00007    All rights reserved.
00008 
00009    Redistribution and use in source and binary forms, with or without
00010    modification, are permitted provided that the following conditions are met:
00011 
00012    * Redistributions of source code must retain the above copyright
00013      notice, this list of conditions and the following disclaimer.
00014 
00015    * Redistributions in binary form must reproduce the above copyright
00016      notice, this list of conditions and the following disclaimer in
00017      the documentation and/or other materials provided with the
00018      distribution.
00019 
00020    * Neither the name of the copyright holders nor the names of
00021      contributors may be used to endorse or promote products derived
00022      from this software without specific prior written permission.
00023 
00024   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
00025   AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
00026   IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
00027   ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
00028   LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
00029   CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
00030   SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
00031   INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
00032   CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
00033   ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
00034   POSSIBILITY OF SUCH DAMAGE.
00035 
00036   $Id: stdlib_8h_source.html,v 1.1.1.4 2012/01/03 16:04:22 joerg_wunsch Exp $
00037 */

Answer 1

If you've based your source on the source of glibc in any way (including looking at the glibc source, roughly figuring out how it works, and then writing your own similar code), your library may already count as a derived work. Since glibc is licensed under the GPL with linking exception, that means your code must be under the GPL as well. The copyright should then say something like, "Based on glibc sources. Additional parts (c) ."

Deriving works indirectly from others is a tricky legal issue, and you may want to consult an expert about it. In some companies, engineers aren't allowed to look at GPL code that is remotely related to what they're doing, for fear that they could be "poisoned" by the code and produce derived works. (I remember one engineer who works on a compiler for some big company who watched a talk I gave about Clang - he told me he had to close his eyes when one of my slides contained a bit of Clang source code. And Clang isn't even GPLed.)

On the other hand, to the best of my knowledge such a loose connection has never been tested in a court, and in the case of a clearly defined programming interfaces as the C standard library, the obviousness defense may well apply in a lot of cases. In other words, if you write a simple strcpy:

char* strcpy(char* dest, const char* src) {
  for (; *src; ++src, ++dest) {
    *dest = *src;
  }
  return dest;
}

then it will be hard to say that you copied that from someone else because that is just the simple, obvious way to implement it. But if the code gets more complex - say, it includes SIMD optimizations, then things get more fuzzy.