Re: JFORK: Or a reasonable response to the Sun SCSL
Ean R . Schuessler writes:
> On Tue, Sep 14, 1999 at 10:05:19PM -0700, Cris J. Holdorph wrote:
> > This has been discussed before. I have pointed out that you do NOT need to
> > sign the SCSL to purchase a book that describes the specification of the JVM
> > and the Java 2 class libraries. It *IS* possible to fully implement these
> > specs without agreeing to the SCSL.
>
> Ok, to some extend you are correct
According to Baratz, Chris is wrong. Along with the SCSL, Sun
has publicly taken a different stance on spec based clean-room
implementation, in a high profile court case. Sun is adopting
a notion that the specifications and APIs themselves are
intellectual property.
If there is the remote chance that a US court follows this
interpretation, then there are no clean-room implementations
of Java2. Period.
Then there is omission from and obfuscation in the specs, which
has already been an issue for Classpath and Japhar, IIRC.
Leveraging patents in some core component of Java is another
possibility to consider.
Finally, there are areas in which Sun's specs simply do not
address issues at all because they themselves don't know (with
respect to a Japhar-based Echidna-like Java service for Linux,
the issue of multi-VM/mixed-version invoked from a single process
comes to mind. JNI does not really tell you how this is supposed
to work).
> The SCSL will cause a widespread proliferation of sources
> that would otherwise be distributed to a very narrow audience.
Definitely. I have been lured into the JDK source licensing in 1995,
which is bad enough, I will definitely stay away from SCSL.
> Well, there is a big difference between "not free" and "almost kinda sorta
> free but you better get a lawyer, then again get three". Given the choice
> I would rather have just plain old "not free", since I can actually figure
> out what is going on there.
If cloning the API is illegal, then all free software ends there.
See OpenGL for comparison: there is a proprietary API, and a
registered trademark. There is also an ARB, there is a free
implementation (Mesa) which was supported by SGI in many ways
over several years, and there is full specification including
the state machine at the most detailed level, with no SCSL
licensing restriction enforced. This is a situation in which free
software can thrive. Sun has never made a similar offer to the
open source communities.
b.
Reply to: