www-legal-discuss mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ceki Gulcu <c...@qos.ch>
Subject Re: LGPL software behind an isolation layer
Date Thu, 27 Aug 2009 10:11:27 GMT
Santiago Gala wrote:
> El mar, 25-08-2009 a las 07:52 -0700, Ralph Goers escribió:
>> On Aug 24, 2009, at 10:40 PM, Santiago Gala wrote:
>>> El lun, 24-08-2009 a las 19:43 +0200, Ceki Gulcu escribió:
>>> (...)
>>>> Assuming the linking was static (that's a big if), then you can't  
>>>> just
>>>> sweep the question under the rug by saying that "it's a system
>>>> requirement". Glibc is a mandatory dependency on Linux and when APR  
>>>> is
>>>> built on that platform, it always links with it.
>>> IANAL, etc.
>>> IIRC, the reasoning we follow is that a libc dependency does not make
>>> products using it into derivative works because the API that glibc,
>>> ulibc or the BSD libc offer is publicly available (~POSIX) and not  
>>> under
>>> a "viral" license.
>>> So, the mandatory dependency that you mention is the POSIX API, as
>>> exemplified by the .h files, and not the copyright-eable part of glibc
>>> (its expression->implementation).
>> How does the user "choose" the implementation of the Posix API. Is it  
>> just a manifestation of what they already have installed on their  
>> machine?
> It is not the user, but the one that builds a binary for a system. As
> the source distribution is able to be compiled under several platforms
> (BSD, linux, cygwin, ...), the Posix API implementation chosen will be
> one amongst several. Things like using ulibc instead of glibc under
> linux are conceivable, though not commonly seen.
> Different libc implementations will produce different binaries, even
> when linked dynamically.
> If glibc was GPL, our source would still not be virally converted into
> GPL, as
> - we don't program against the GPL interface, but against a "free" API
> which they happen to implement
> - you could produce binaries, at least in some platforms or using some
> auxiliary libc implementations, that are not linked against it, hence
> our work is not their derivative in the copyright sense.
> Again, IANAL and the analogies work only while they work...

Distinguishing between source and binary distributions is in my
opinion archaic. Not only such a distinction is not based on copyright
law (AFAIK), it also constitutes an admission (or could be
interpreted as an admission) that something fishy is going on.

The big question is whether coding against a "standard" interface
distributed under a non-copyleft oss license, but where the
implementation is licensed under (L)GPL, neutralizes the copyleft
properties of the latter.  As I share Santiago's sentiment that work
done against a non-copyleft API is not derivative in the copyright
sense.  I seems pretty logical and obvious to me, but nothing is
logical and obvious in matters of love and copyright law.

By the way, if the "isolation" hypothesis listed above were true, it
could create a whole new oss eco-system with bridges to copylefted

> Regards
> Santiago
>> Ralph

Ceki Gülcü
Logback: The reliable, generic, fast and flexible logging framework for Java.

To unsubscribe, e-mail: legal-discuss-unsubscribe@apache.org
For additional commands, e-mail: legal-discuss-help@apache.org

View raw message