[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#709382: Built-Using, libgcc, and libc_nonshared



Russ Allbery dixit:

>of the GPLv2, the GPLv2 itself requires that all of the *source* for the
>binary be distributed under the GPLv2.  And the libgcc *source* is only
>available under the GPLv3, and the runtime exception doesn't allow one to
>distribute the *source* under different terms, only the resulting binary.

Ouch!

>Clearly no one else in the world is worrying about this; there's lots of
>GPLv2-only software out there and all the distributions are happily

The same thing happens with software labelled as “Public Domain”.
There are several cases (the authors aggressively sell licences
for those who don’t trust their say-so it’s PD (sqlite) and which
are to be avoided at any cost; these where the authors say it’s
PD (and subsequently don’t care nor sue) but it isn’t (e.g. if
you’re in a country where you can’t dedicate something into PD
by yourself, or if some conditions are met, even for USA), or
legitimate PD in *one* country (often stuff done by government)),
but there is no such thing as global PD as it’s by its definition
absence of copyright protection, and the Berne Convention only
harmonises copyright protection, so something PD in one country
is proprietary in all others in virtually all cases… but this
issue only has popped up on the OSI mailing list, and so far,
almost nobody cares (I try to get “fallback” clauses from authors
myself, succeeded for some, failed for most).

>I'm not sure to what extent we can use that as an excuse, though.

I guess until you come in front of a court.

But there’s the other thing: if a distro/OS promises to use only
code with (known) good licences, so that their users can rely on
it, like http://www.openbsd.org/goals.html (second list item),
one probably does not _want_ to have “grey zone” in their distri‐
bution, so it may be good to be proactive.


There’s something else about Built-Using:

Are those source packages (that would not otherwise be kept in
the archive) released along with “stable”, despite having no
binary packages?

If not… well, since snapshot.d.o is an official service now,
I’d say, point there for the “legal” side, and only require
Built-Using to be used for the cases where it’s desireable
to have this information present more explicitly, that is,
the original toolchain and embedding stuff, possibly static
linking and Haskell. Pragmatic but probably doable, and since
quite some time, sbuild records (in the build log) the versions
of the toolchain packages installed already anyway, so we just
need to put build logs into the snapshots archive as well; I
believe they’re deleted when an architecture moves off the main
archive (to d-ports, or because it’s no longer even in oldstable)
normally.

bye,
//mirabilos
-- 
  "Using Lynx is like wearing a really good pair of shades: cuts out
   the glare and harmful UV (ultra-vanity), and you feel so-o-o COOL."
                                         -- Henry Nelson, March 1999


Reply to: