Re: Ghostscript licensing changed to AGPL
On Wed, 7 May 2014, Bálint Réczey wrote:
> 2014-05-07 14:37 GMT+02:00 Thorsten Glaser <email@example.com>:
> > Which you may want to do, in order to patch a security issue
> > you just found, locally, before filing it upstream.
> In my interpretation in this case I would have some reasonable time to
> comply, i.e. I don't have to publish all 0days on my site if I run
> AGPL-covered software..
No. The licence does not provide for such a delay.
> > Or because you’re a user of Debian and used to be able to do
> > just that.
> Hmm, as a Debian user I'm used to respecting the license of any
> software being it BSD, GPL or AGPL...
Right, but the AGPL (and, to some extent, the GFDL… I smell a
pattern here) are unique in that they restrict usage without
reproducing the software itself.
On Wed, 7 May 2014, Clint Byrum wrote:
> The things that link to ghostscript as a library will now need to be
> evaluated. If they are contacted via network ports, they'll need to
> have source download capabilities added.
This is impractical; upstreams will probably just answer to use an
older (or the commercial) Ghostscript version.
> Where the AGPL fail to be clear, is when we ask what to do with a program
> which listens over the network, and executes an AGPL licensed program. The
AIUI the AGPL triggers if the program is accessed over the network;
local execution does not normally use “the network”. If the program
executing it is called over the network, it does not directly expose
the AGPL software, but to draw the line is very hard.
Also, see below.
On Thu, 8 May 2014, Riley Baird wrote:
> > Should it then communicate that it is Debian version X and that the
> > source can be found on any Debian mirror near you?
> What if the network in question is not the internet?
Right, the AGPL is not technology-neutral.
I have to agree with Francesco Poli here that the AGPL has
no place in main, not just because of its freeness but also
because of its impact at reuse and maintainability (think
security updates; we had this in the Berkeley DB 6.x thread
15:41⎜<Lo-lan-do:#fusionforge> Somebody write a testsuite for helloworld :-)