Again: "derived work" is meaningless here.
Posted Feb 18, 2011 6:07 UTC (Fri) by khim
In reply to: PostgreSQL, OpenSSL, and the GPL
Parent article: PostgreSQL, OpenSSL, and the GPL
While the FSF unofficially advises an expansive definition ("anything that links"), here LD_PRELOAD can allow completely decoupled distribution of the two pieces of code without any plausible license-cross-infection claim.
Yes, of course. Unless you'll include such script in your distribution and make it available by default. You see, GPL is pretty expansive: if forces you to distribute everything derived from GPLed sources to be distributed under GPL terms. Including whole distributions! But there are couple of exceptions: system libraries ("unless that component itself accompanies the executable") and "mere aggregation". Now, of course libedit-linked postgresql is not infringing. But what about distribution which forces usage of readline injected in postgresql client? This very much a grey area. If it's unconditionally injects the readline then it can probably viewed as a clever way to circumvent the GPL requirement. But if it's configurable and user can select one of two choices... then it's more like mere aggregation - but only court can say for sure... and two different courts may decide differently too.
We can argue till we blue in face but because law is squishy there are no resolution till actual court proceeding will occur. We can only estimate probabilities of different outcomes, really.
to post comments)