Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-10 Thread Stephen J. Turnbull
"Martin v. Löwis" writes: > Note that it is ultimately up to a court to interpret these words of the > GPL, not to the FSF lawyer. True, and in the case of a non-FSF product, any ambiguities would be resolved first by determining the intent of the copyright owner, second (perhaps even overridin

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-10 Thread Joao S. O. Bueno
On Thu, Mar 10, 2011 at 2:41 AM, Stephen J. Turnbull wrote: > Joao S. O. Bueno writes: > >  > Any libraries commonly avaliable with a CPython instalation can be >  > considered as "system libraries" for GPL purposes - and so >  > this would fall in the "system library exception" as described by th

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-10 Thread Martin v. Löwis
Note that your interpretation would allow Python to distribute arbitrarily licensed libraries and GPL programs to link with them. That is surely not the intent of the authors of the GPL, and in the past, the FSF has explicitly restricted the interpretation of "system library". Note that it is ul

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread Stefan Behnel
James Y Knight, 10.03.2011 06:52: But you're also left with not being able to 'import hashlib'. While python has fallback code, those modules (_md5, _sha, _sha256, _sha512) aren't built if openssl was found at build time. So you can't just select at runtime that you didn't want to use openssl.

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread Stephen J. Turnbull
Westley Martínez writes: > Is it legal to distribute GPL programs that use the Win32 API? Yes. Their use of the Win32 API falls under the "essential system library" clause. The criterion for "essential" is that normal, basic use of the system would fail without the library. Windows won't boot

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread James Y Knight
On Mar 9, 2011, at 6:45 PM, Sandro Tosi wrote: > It seems introduced by the patch debian/patches/setup-modules-ssl.diff > with description "# DP: Modules/Setup.dist: patch to build _hashlib > and _ssl extensions statically" Indeed you're right -- out of the box, python still builds _ssl.so as a s

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread Stephen J. Turnbull
Joao S. O. Bueno writes: > Any libraries commonly avaliable with a CPython instalation can be > considered as "system libraries" for GPL purposes - and so > this would fall in the "system library exception" as described by the FAQ: Note that your interpretation would allow Python to distribute

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread Joao S. O. Bueno
On Thu, Mar 10, 2011 at 12:49 AM, Westley Martínez wrote: > On Wed, 2011-03-09 at 18:20 -0500, James Y Knight wrote: >> It's well known that OpenSSL is incompatible with the GPL. [1] Python (from >> 2.6) is *always* linked against openssl, instead of waiting for you to >> "import ssl". >> >> Doe

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread Martin v. Löwis
Am 09.03.11 18:20, schrieb James Y Knight: [1] Python (from 2.6) is *always* linked against openssl Others have already pointed that out, but let me stress it again: this statement is false. It is as possible and easy to build Python without OpenSSL as it always was. > instead of waiting for y

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread Westley Martínez
On Wed, 2011-03-09 at 18:20 -0500, James Y Knight wrote: > It's well known that OpenSSL is incompatible with the GPL. [1] Python (from > 2.6) is *always* linked against openssl, instead of waiting for you to > "import ssl". > > Doesn't this mean it's now impossible (rather, a license violation)

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread Joao S. O. Bueno
On Wed, Mar 9, 2011 at 8:20 PM, James Y Knight wrote: > It's well known that OpenSSL is incompatible with the GPL. [1] Python (from > 2.6) is *always* linked against openssl, instead of waiting for you to > "import ssl". > > Doesn't this mean it's now impossible (rather, a license violation) to

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread Sandro Tosi
On Thu, Mar 10, 2011 at 00:32, Antoine Pitrou wrote: > Also, with the system Python 2.6 shipped on a Mandriva distribution: > > $ ldd /usr/bin/python >        linux-vdso.so.1 =>  (0x7fff8456d000) >        libpython2.6.so.1.0 => /usr/lib64/libpython2.6.so.1.0 > (0x7f1a0b1e2000) libc.so.6 =>

Re: [Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread Antoine Pitrou
On Wed, 9 Mar 2011 18:20:01 -0500 James Y Knight wrote: > It's well known that OpenSSL is incompatible with the GPL. [1] Python (from > 2.6) is *always* linked against openssl, instead of waiting for you to > "import ssl". > > Doesn't this mean it's now impossible (rather, a license violation)

[Python-Dev] GPL'd python code vs Python2.6 linked against OpenSSL

2011-03-09 Thread James Y Knight
It's well known that OpenSSL is incompatible with the GPL. [1] Python (from 2.6) is *always* linked against openssl, instead of waiting for you to "import ssl". Doesn't this mean it's now impossible (rather, a license violation) to distribute a GPL'd python program (or to use a GPL'd library in