From: Mark Everett (everettm@merlin.think.com)
Date: 07/07/93


From: everettm@merlin.think.com (Mark Everett)
Subject: Re: Linux vs. RSAREF
Date: 7 Jul 1993 08:47:53


   |-In article <Jul.6.23.35.21.1993.10891@geneva.rutgers.edu> hedrick@geneva.rutgers.edu (Charles Hedrick) writes:

   markh@vanbc.wimsey.com (Mark C. Henderson) writes:

   }>Questions:
   }>1. Can one build RIPEM/RSAREF for Linux?
   }>2. Can one use RIPEM/RSAREF on a Linux system?
   }>3. Can one distribute a Linux ripem binary?
   }>4. Can one distribute a makefile entry for RIPEM under Linux?
   }>5. Is selling Motif for Linux a violation of the GPL?
   }>6. Can one, "in general" distribute and use, software the FSF considers
   }> not "free" for a Linux system.
   }
   }I think the answer to 1-4 and 6 is "yes", and the answer to 5 is "no.
   }I would use as a precedent the fact that FSF blessed the Lucid "Energize"...

   |- While I agree with you, my recent correspondence with rms suggests
   |- that he considers the answer to be 5 and 6 to be "yes" and "no". He
   |- appears to take the position that a binary that requires a GPL'ed
   |- library is a derivative work, even if integration isn't done until
   |- runtime.

I totally agree with rms' sentiments. However, I don't think you're
interpretting "requires a GPL'ed library" correctly. There are plenty
of implementations of libc, for instance, and if a program requires
a libc it doesn't matter that the libc being used is from FSF. It
*does* matter if FSF provides a particular interface in their libc
that isn't generally available and, for that reason, the program will
*only* work with FSF's *implementation* of libc. I think if you could
get rms to agree on this, and I don't see why he wouldn't, then there
should be no problem sticking with the existing libc.

-everettm