[rc5] Next Crack

Eric Gindrup gindrup at okway.okstate.edu
Mon Aug 25 12:38:49 EDT 1997


     The quoted material reminds me...  Jave clients would probably rotate 
     just as poorly on any machine.  There's no built-in support for 
     describing rotates, so you'd have to do the masking, shifting and 
     combining in the high-level code.  Have a JIT compiler recognize that 
     as a rotate would be a miracle.
     So we'd probably end up with a widely distributable client that ran 
     *way* slower because it was impossible to represent architecturally 
     dependent optimizations in the language.  I'm not saying it's bad 
     because it's high level.  It's bad (in this case) because it's not 
     general enough.  It doesn't let you state what you want (rotate) and 
     then do that the best way on the running platform...  The language 
     lacks a necessary expressiveness for highly-tunable code.
            -- Eric Gindrup ! gindrup at Okway.okstate.edu


______________________________ Reply Separator _________________________________
[stuff elided]
  The native client could have more specific instructions (since it isn't 
cross-platform you don't need generic instructions on how to run a program 
or things like that), and could tap features of individual computers (like 
running as an NT service, or hidden in win95).  
     


----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list