Wireshark · Ethereal-dev: Re: [Ethereal-dev] RADIUS's "Message Authenticator"

[David Frascone <dave@xxxxxxxxxxxx>]

On Monday, 25 Mar 2002, Hamish Moffatt wrote:
> On Sun, Mar 24, 2002 at 11:03:21AM -0600, David Frascone wrote:
> > I don't think so.  If we don't *require* openssl, or ever even point to how
> > to download it, then we're not recommending it's usage either.  But, if it
> > happens to be distributed with the OS, we will detect and use it.  I *do*
> > think that will keep the licenses free from question.
> 
> Just consider the intent of the license (GPL). The intention is that you can
> have the source code to the entire program, including the C library.
> (On proprietary systems that isn't feasible, so there's a specific
> exemption for libraries supplied with the system.) And then you must
> be able to modify the program for your own use and distribute your
> changes, also licensed with the GPL, as you see fit.
> 
> If you consider run-time linking of libraries as acceptable, then
> anyone is free to write plug-ins or new code for Ethereal for which
> they won't distribute the source - which violates the intent of the
> license.

Ok.  Good example.  But, what about libc under Solaris.  It's just 
runtime linking.  I think using an external library for our (the authors')
benefit is not in violation of the intent of the license.  (writing a 
closed source plugin definately would be)

> As free software authors we should not be trying to evade a free 
> software license - especially the license of the product we are 
> working on!

I don't think we're evading a license.  If OpenSSL was GPL, then there
would be no discussion going on.  We're trying to use another free
library, with a *less* restrictive license than our own.  We seem to
have no problem linking with libraries with *more* restrictive licenses
(i.e. libc under HP/UX, Solaris, and, God forbid, MS Windows), but are
having many issues linking to a less restrictive one.  And, we're even
(If we take my runtime only linking approach) completly avoiding any
source code contamination by linking only at run time.

I say let's do it.  I seriously doubt anyone will complain.  And, if
*anyone* does (even one author), then we just throw out the changes. 

Comments?

-Dave

-- 
David Frascone

   A bad random number generator: 1, 1, 1, 4.33e+67, 1, 1, 1