incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roy T. Fielding" <>
Subject Re: [policy] bring in full code history on incubated project?
Date Thu, 05 Jan 2006 03:09:02 GMT
On Jan 4, 2006, at 3:21 PM, Noel J. Bergman wrote:
> Roy, what happens if I have a codebase, and I have one file in it that
> imports or otherwise references third party GPL code, and I then  
> delete my
> infected file?  Do we have a concern about having that GPL infected  
> code in
> SVN, even though it is dead?

No, "infection" by GPL only applies to the act of redistribution
under terms that are not GPL compliant -- it is not a communicable
disease.  Placing source code on a public subversion instance
is compliant with the GPL, so no worries.

There is a separate question about whether "imports or otherwise
references" by name is sufficient to cause a work to be derivative
(without distribution of the referenced work), which is not something
I want to get into again.  FSF and Sun claim that it is sufficient.
Everyone else I've talked to says "no way".  I don't know of any
precedent that has decided either way.  IANAL, ETIDALOWFT.

[some folks are going to hate this analogy -- I don't care.]

GPL is not really a virus. GPL'd code is more like a genetic mutation
that is applied to cells, basically wherever the copyright owner says
the mutation applies.  Copied cells retain the mutation if the original
had the mutation.  The mutation isn't really malignant -- it doesn't
spread itself to other cells -- it simply prevents the entire organism
from reproducing until the entire organism has mutated.

The mutation can only be removed from a cell by its copyright owner.
That's why keeping track of the copyright owner of each cell is
very important.  A "cell" is some amorphous quantity of work that
qualifies for copyright under international law.

The ASF can do anything GPL-compatible with GPL cells.  The only thing
we can't do, really, is distribute them under a non-GPL license.  That
means we can't produce released products containing those cells unless
the mutated cells are in source form by nature (e.g., shell scripts),
since doing so would confuse our users that expect a clean bill of
health from the products we release.  They expect to be able to safely
incorporate our released products within their own closed-source
products without worrying about mutated cells slipping in by accident.

Similar mutations apply to all copyleft licenses.  The real difference
among copyleft licenses is what qualifies as the "entire organism"
for the purpose of reproduction.  MPL and CDDL define the organism
to be each individual source file.  GPL defines the organism to be
anything linked to the mutation.  LGPL defines it in multiple
incompatible ways that are just too confusing to mention.

Whether you consider this mutation to be a disease, or simply some
better form of life (a la X-Men), is largely dependent upon one's
personal PoV and ethical principles.

Regardless, you can eliminate the infection by carving out those
cells that have been mutated, assuming you can identify them, or
by getting a new license from the copyright owners.


Roy T. Fielding                            <>
Chief Scientist, Day Software              <>

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message