On Thu, Feb 17, 2011 at 12:39 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> In particular, getting rid of use of OpenSSL would not be sufficient
> to satisfy the most rabid GPL partisans that we were in compliance.
Huh?
In what way would we not be in compliance? Or rather, what part of the
GPL would we be unable to comply with for distributing binaries?
I think what you're getting at is that distributing source which can
optionally link against GPL code might itself be a derivative work of
the GPL code and need to be distributed under the GPL even if it's not
built against it. I think that's just a straw man though, even the
most ardent GPL partisan isn't going to claim that the Postgres source
is a derivative work of readline because it has the option to link
against readline for additional incidental functionality.
To give context the case where this comes up are things like Gimp
plugins *which are useless with thout the GIMP*. They're entirely
dependent on the Gimp for their functionality. Claiming they're
derivative works of the Gimp is a lot easier than claiming that
Postgres is a derivative work of readline. A more borderline case was
programs based on GMP. However even there it's hard to picture a
useful program which needs GMP being able to do anything useful
without GMP. Even then just providing a (much poorer) alternative
implementation makes the case fall apart.
> Whereas, if we get rid of readline, it no longer matters whether we
> depend on OpenSSL.
Not really, people still need to abide by the OpenSSL license rules
which make our product less useful and less flexible. People might
want to include Postgres in a product which uses other GPL'd code or
which they don't want to alter their advertising.
--
greg