RFC: Signed JAR Packaging Policy

[Andrew Haley]

Nicolas Mailhot writes:
 > Le mardi 13 mars 2007 à 14:01 -0400, Matthew Miller a écrit :
 > > On Tue, Mar 13, 2007 at 01:54:56PM -0400, Rob Crittenden wrote:
 > > > But any random JVM that a user downloads from Sun or IBM directly 
 > > > wouldn't know about this extra endorsement, right? So any 
 > > > non-rpm-installed JVMs would still not work?
 > > 
 > > They would work fine for whatever they were installed for. They just
 > > wouldn't work for this. Presuming that "this" can be made to work just fine
 > > with gcj and/or future-free-java, then it should just stay that way. Am I
 > > not getting something? 
 > > 
 > > I mean, do we consider "the kernel doesn't build with Microsoft C" a
 > > problem?
 > 
 > The kernel is mature and the main implementation. If
 > fedora-packaged java apps fail when people try to use them with
 > proprietary jvms that's a bigger problem. Till gcj is mature,
 > efficient and recognized in the marketplace being able to mix and
 > replace components (including the jvm) is a huge plus.
 > 
 > (now on this particular point, I don't see how we can accommodate
 > SUN & Fedora requirements, and Fedora goals come first)

It's just a dependency, no different from any other package
dependency.  A proprietary Java runtime environment that needs signed
crypto JARs has a dependency on such JARs.  When such a proprietary
JRE is installed, the signed crypto JARs it needs should be installed
with it.

Andrew.