I asked Trevor to take a look at the server for me.
I didn't see any problems server-side. I scanned through the error_logs and nothing jumped out at me, although I think the client would've produced an error regarding a corrupted JAR file if the checksums didn't match after downloading...
The JNLP that is generated looks correct as well. (Jay, you probably know this by the link for the WebStart URL can have the jnlp
replaced with txt
and you'll get a text version of the JNLP in the browser window. Just a shortcut for looking at the file contents easily.)
It would seem to be some kind of local issue where the java.lang.zip packages aren't reading the JAR files correctly? Or maybe the ClassLoader hasn't been initialized properly when the first lookup attempt happens? I know that MapTool has a cp://
URL handler installed early in the startup process that searches the classpath -- does InitTool do anything like that? (Note that I have no idea why
MT does that as I've never seen any use of that URL, although it does make for some interesting possibilities.