Re: [DotGNU]Changing pnetlib license to LGPL

[BioChem333]

On Thu, 2002-07-11 at 22:23, Mario D. Santana wrote:
> If I read you right, the loophole you mention involves front-ending the
> modified LGPL'd object code with, eg, SOAP. By presupposing that offering
> a program this way counts as internal usage (as opposed to distribution,)
> this loophole applies to GPL'd code as well, linking exception or no.
> AFAIK, there's no free license that explicitly restricts this sort of
> thing. In fact, I don't see how a license could be free, if it dictates
> what a user can do with the program, absent distribution.

The loophole I refer to involves back-ending the code, so that
derivative works aren't covered. If you add or replace a function in the
library, but do so using native code, then you only have to declare the
native function in the library. There's nothing stopping a proprietary
software vendor from making improvements to the library, which would by
common sense be considered derivative, but by law would not. All such a
vendor would have to do is release the native implementations in object
code only, along with the 'improved' library's source; the modified
source itself would be useless to us if the actual implementation of the
'improved' or added routines were not available. The linking exception
explicitly prohibits a library extender from doing this, while otherwise
providing linking permissions which are essentially the same as those
found in the LGPL.