Licenses and subterfuge

Alex Rousskov rousskov at
Thu Feb 26 23:43:47 UTC 2004

On Thu, 26 Feb 2004, Arnoud Engelfriet wrote:

> The issue I see is that your software requires that particular
> GPL-licensed library to run. Normally, you would supply your
> software together with that library (dynamically linked). Now, for
> the sole reason to avoid having to comply with the GPL, you omit
> that library. That sounds like a subterfuge.

The software does not require any particular implementation of an API.
The fact that only one implementation is known to software writers
today should be irrelevant from copyright point of view, IMHO.

> >   "Our software can be linked with any library supporting Foo
> >   API. Users report success with FooLib on Linux. Other Foo API
> >   libraries may be available in your environment. Known compatibility
> >   problems with Foo libraries are available by searching our bug
> >   database at ..."
> Can you really say that seriously if the _only_ implementation
> available is GPL-licensed FooLib? And more importantly, will a
> judge believe you?

IMO, you can say that seriously if your program can accept any
implementation of a published API. Availability and licensing of
libraries should be irrelevant.

> No such thing as a "viral header" because header files are usually
> purely functional and so not copyrighted. And even then, does any
> code from the header file end up in the compiled binary of your
> code?

This seems to be no longer true (if it ever was). There are entire
libraries written via headers (no .c files or equivalent) and most
headers I look at are copyrighted. So, yes, header code does end up in
binaries and can be viral.


license-discuss archive is at

More information about the License-discuss mailing list