On Sat, 2004-09-25 at 10:16 -0400, Tres Seaver wrote:
> Henrik Nordstrom wrote:
> 
> > Note: Squid is GPL and as a result you are only allowed to use GPL 
> > modules with Squid. If your filter implementation is not GPL then 
> > dynamic linking is not an option.
> 
> IANAL, bu wouldn't it be truer to say that the GPL does not allow 
> *distribution* of Squid linked with software under non-GPL-compatible 
> licenses?
> 
> See:
>    http://www.gnu.org/licenses/gpl-faq.html#TOCGPLRequireSourcePostedPublic
> 
> and:
>    http://www.gnu.org/licenses/gpl-faq.html#InternalDistribution
> 
> And I am not sure that dynamic linking triggers the "derivative work" 
> provisions, particularly if Squid continues to function without the 
> presence of the library.

This is incredibly tricky to get right: If you take the position that
dynamic plugins are derived works (of the thing they link into - which
here would be the squid routines), you end up with interface
copyrighting (that is, something that has no code, may only be a couple
of lines, and is trivially the same for similar requirements - bleuch).
if you take the position that dynamic plugins are not derived works,
then someone can write a proprietary gzip compression module for squid,
ship it at whatever cost, leveraging /all the squid internals/. Which is
also quite unattractive. AFAICT there is no standard that one can apply
neutrally and externally to instances of dynamic linking to decide
whether it would count as a derived work. 

Currently it is a very messy situation :(

Rob

> Note particulary:
> 
>    http://www.gnu.org/licenses/gpl-faq.html#GPLAndPlugins
> 
> and the use of the word "believe" in that language.  The GPL itself does 
> not mention dynamic linking at all.
> 
> Tres.
-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to