On the Java licensing front, two observations:

1. Gosling says that the Java specification grants rights for any complete implementation of the specification, and that right extends to patents. See bullet 4, http://today.java.net/jag/page7.html#59.

2. The Sun JCP licensing explicitly grants a "perpetual, non-exclusive, worldwide, fully paid-up, royalty free, limited license (without the right to sublicense) under any applicable copyrights or patent rights" to *any* implementation of the specification (which they further define as fully implementing the spec in its entirety, not extending the spec, and passing the relevant TCK for the specification). See http://jcp.org/aboutJava/communityprocess/final/jsr129/j2me_pb-1_0-fr-spec-license.html for an example.

So I think Miguel's claim that there are no guarantees that Sun won't sue an open implementation of APIs are somewhat overblown relative to the .NET problem. Plus, Sun has licensed the TCK to ObjectWeb and Apache for free.

The JCP has a bunch of problems. Still, it's not a terrible start, and it's improving.

I'm attempting to recompile the tip of the Mutt CVS tree with the IMAP headers caching patch:

http://wwwcip.informatik.uni-erlangen.de/~sithglan/mutt/