And I believe we have a contender for my favorite Java error of all time. Recalling the "reallyDoThePaintOrUpdate" escapade from January, we now have the bad magic number error.

I'd actually heard of magic numbers recently when I had some time to work on my Atari 2600 project. Using assembler, it's easy to get locked into a hexidecimal mindset, and I had quite a few things happening in powers of two. Graphics were displayed for sixteen (2^4) scan lines, each player had eight (2^3) frames of animation, etc. When I posted my code to the Stella programming list, I received a useful comment of how to stop using (or to stop limiting myself to) "magic numbers" like 8,16, and other powers of two. To me, however, magic numbers in machine language code make the code easier to follow. Next time I have free time for the Atari, I'll have to see which way I decide to go...

Unfortunately this Java magic number error only seems to be referring to the fact that the return value indicating success for some call is an easy to read number in hexidecimal, and even then (see link) it's just a adolescent-esque joke allowing the programmers to have a return value that uses the letters A-E (which are numbers in hex) to spell "CAFEBABE" (for "java woman", I suppose, which could be an interesting play on words with Java Man).

So I think I'm sticking with "reallyDoThePaintOrUpdate" as my favorite text from a Java exception dump, at least for the time being. Another essential newsflash thanks to myfreakinname!