I've just compiled XEmacs for the sparc v9 architecture with the CC 5 beta, and it appears to work. On that build, most-positive-fixnum is 4611686018427387903. Yay! On a less happy note, preliminary benchmarks show it to be quite a bit slower than ILP32 XEmacs. Specifically, `xemacs -batch -eval "(let ((i (* 1024 1024 10))) (while (> i 0) (setq i (1- i))))"' takes 39.49 seconds of CPU time on a v8 build, while it takes 60.02 seconds on a v9 build. Apparently, the price of 64-bit registers is high. :-( (For the Sun-uninitiated, ILP32 and LP64 refer to the bit size of integer, long, and pointer types in "C". ILP32, or v8, behaves like traditional 32-bit Unices. LP64 behaves like Digital Unix.)