[theora-dev] Consistency regarding compiled Cortado 0.6.0 source and the official binary

Timothy B. Terriberry tterribe at xiph.org
Mon Nov 29 18:01:34 PST 2010

> The locally-compiled instance of the Cortado applet seemed to be functional and seemed to work (at least,
 > when tested on the Mac OSX platform.) The question is: When providing 
copies of the officially-released

You should try testing it on very old JVMs (for which Cortado is 
intended as a fallback). I can almost guarantee you that your version 
will fail to work on any 1.1-era JVM, because Sun changed the way their 
compiler emits code for monitors in a way that triggers a bug in such 
JVMs. It's been like this since at least JDK 1.4. They insist that this 
is required for language compliance (though AFAICT it is not) and refuse 
to fix it.

 > Cortado applet jar file (the cortado-ov-stripped-0.6.0.jar that has 
such internal file names as a.class
 > and b.class) to others, my assumption has been that the 
cortado-0.6.0.tar.gz file can be treated as the
 > corresponding source code for the purpose of convience and license 
compliance. Is this correct? (For all

That is correct. The official binaries have simply been run through 
Proguard. One of the optimizations we got added to Proguard was to have 
it rewrite the instructions for these monitors in a way that works with 
the old JVMs. See 
for details.

More information about the theora-dev mailing list