Michael Eager eager@eagercon.com
Mon Jul 16 15:11:00 GMT 2007

Richard Kenner wrote:
>>> You are assuming here that the patch ITSELF has some license that's applied
>>> to it irrespective of the file it was derived from and I don't see the
>>> legal basis for such a claim.
>> Your patch, once accepted by the FSF, becomes their property, and the
>> FSF, not you, determines the license the *patch* will be covered by.
>> They can, if they wish, decide that every patch will be covered by
>> GPLv3.  They can, if they wish, decide that some patches will be covered
>> by GPLv3 and others by GPLv2.  They can use whatever scheme that they
>> wish to decide which license to apply to the patch, including schemes
>> which ignore the license of the original source.
> Yes, they certainly *can*.  But my question is whether they *have* and
> if so, who has done it and when.  I've seen no evidence of any assertion
> ever of what license applies to a *patch*.

It is source, covered by the copyright assignment.  The assignment, if I
recall correctly, says that the FSF will distribute the source under license.

>> One person can't develop the same identical patch, as you posit,
>> from similar sources, and claim that they were independently
>> derived.  A derivative work is clearly covered by the same version
>> of GPL which the original was covered by.
> Yes, sure, but the independent claim can clearly be made.  Suppose I
> write a sed script to do an edit to a file.  I do not look at two
> files, one of which is GPLv2 and one is GPLv3, but instead run the
> script on both of them and run diff to get a patch.  There's no
> question that each of the resulting patches is covered by different
> versions of the GPL (by your last sentence).  They are clearly independent
> because there was no possible vehicle for "contamination" (I didn't look
> at the files).  But if they end up as identical, we now have two identical
> patches that have difference licenses.

This is tedious.  The result of a sed script is not a creative work.

>> This is why there are "clean room" implementations of proprietary software
>> -- to prevent just the copyright contamination which you describe.
> Yes, but we're talking about *patches* here, where the underlying license
> derives from the file being patched, not the patch itself.  There's a big
> difference!

I was talking about patches -- copyrightable creative works which may
be assigned and licensed.  You appear to be talking about something else.

Michael Eager	 eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306  650-325-8077

More information about the Gcc mailing list