[BioPython] licensing (was PDBParser & Structure class)

[Andrew Dalke]

Christopher Lee:
>I realize this may be off topic, but I'm worried that you may be spreading
a
>misconception about the (L)GPL.  Copyright law specifically *doesn't* apply
to
>ideas.  The LGPL would prevent you from copy/pasting (or or translating)
code
>into your proprietary code, but you would be free to take ideas in LGPL
code and
>use it in proprietary work.

First, IANAL so you may be right.

There's always the possibility of accidental copying of code, where
since I know how one chunk of code works, I may 'rewrite' it but where
the new code is essentially the same.  If I'm consciously aware of
the similarities, I may have to decide to use another approach.  If
I don't know how the code works, then I can point out that there is
no copyright infringement.

This is true for any copywrited work.

Your statement at its face seems incorrect.  You say "Copyright law
specifically *doesn't* apply to ideas."  But it does.  If I write a
story and you take the idea of the story (and don't modify it enough) to
make a new one, then you are violating my copyright.  There might not
be a single word shared between the two, as with a translation to a
foreign language, but it is still a copyright violation.  Yet all
that's shared is an idea.

> The best analogy might be with scientific
>publications.  The copyright is usually owned by the authors or journal
>publisher, so you can't copy/republish figures or text without permission
>(leaving out fair use), but the ideas and results may be used freely.

Arguing by analogy is complicated.  I've read of cases where an author
has included a chunk of text from elsewhere in a book and not attributed
it.  It was completely accidental as the author has claimed not to have
remembered even reading that original source.  Sorry, don't have a
pointer to such a case.  I tried looking for links under a search for
"accidental plagiarism" but the top hits were articles on how to prevent
that from happening.  (Which implies that it is a frequent worry.)

Are you suggesting that it's impossible to plagiarize from scientific
publications?  It does seem there are different types of "ideas and results"
in such a work, and that some can be copied and others cannot.  Again,
IANAL and can't characterize the difference between them.

>Your worry about "tainting" is valid if you frequently work with both free
and
>non-free software projects simultaneously and if you copy code back and
forth.
>However, for any code you write yourself, you own it and can control the
>licensing---issue it both under the biopython license and any proprietary
>license you'd like.

I disagree.  The concern is to keep from being sued or otherwise blamed
for copyright infringement.  Your point is sufficient to justify things
to myself.  My point is needed to justify things to others.

Suppose I do not copy code back and forth and still the code looks similar
to work you did.  Similar enough that you claim there is copyright
violation.
If I have looked at your code, it may indeed be a case of accidental
plagiarism.  If I have not looked at your code, then it's much less of a
problem, as I can say with authority that there was not even accidental
infringement.

In a related realm, shows like Star Trek got many unsolicited submissions.
The policy there is that those submissions can be a accepted but are
not reviewed by the writers of the show.  Otherwise a storyline or character
may accidentally be used which is owned by someone else.  That's why
story submissions must be accompanied with a signed release form, which
includes ( http://members.aol.com/pinkworld/stscript.htm )
] 2. You agree that you will not use the Material unless you shall first
] negotiate with me compensation for such use, but I understand and agree
] that your use of material containing features and elements similar to or
] identical with those contained in the Material shall not obligate you to
] negotiate with me nor entitle me to any compensation if you determine
] that you have an independent legal right to use such other material which
] is not derived from me (either because such features and elements were
] not new or novel, or were not originated by me, or because other persons
] (including your employees) have submitted or may hereafter submit material
] containing similar or identical features and elements which you have the
] right to use.)

This issue (tainting) is important enough that there are legal ways to
ensure it doesn't happen, like using a so-called "clean room" approach
http://www.mlg.com/news_events/attorney_articles/cleanroom.html

I cannot claim full clean room because I have looked at other source code,
but in nearly all cases I do so after I have implemented my version of
the code, so I know what's written is clean.

And finally, this issue of taint/clean doesn't address my concern that
people who want to develop software which is incompatible with the [L]GPL
 -- and that includes some free/open source licenses -- are unable to
do so and instead waste time rewriting the code from scratch, without
even the ability to scavange for parts.

                    Andrew
                    dalke@dalkescientific.com