On Thu, Mar 21, 2002 at 09:28:13PM +1100, Roger So wrote: > > I am not too happy that the U+20021..U+2F9D4 to Big5-HKSCS mapping > > takes up so much space (nearly 70 KB for just 1651 entries) because of > > the sparse array. Anyhow, BIG5HKSCS.so grew from 115165 bytes to > > 185240 bytes here. Not a big deal, but even the big GB18030.so is only > > 130356 bytes... :-) I am thinking of using a hash instead for that > > area, but that'll be an experiment for another day. Fun stuff! :-) > > Heh. Still better than the giant "switch ... case ... case ..." block > that I first thought of. :)
Hehe, I simply inherited the "from_ucs4_idx" index table that was originally in big5hkscs.c. Did you wrote that, or Ulrich did? :-) > Or perhaps you can have a sorted mapping table with which you can do a > binary search on. Hey, that's another neat idea! I never thought of that. :-) With 1651 entries, it would take at most 11 tries? (Just a rough guess of 2^10 = 1024, 2^11 = 2048; I'm too sleepy to do the real math... :-) But I do want to try a hash table first. Now, if a hash table of size 2n (3302) or less with a good random function can accommodate these 1651 characters with only a few collisions.... I wonder if a binary search or a hash is faster in this case. Oh well, fun stuff nevertheless. :-) Cheers, Anthony (zzZZZzzzZZZ) -- Anthony Fok Tung-Ling ThizLinux Laboratory <[EMAIL PROTECTED]> http://www.thizlinux.com/ Debian Chinese Project <[EMAIL PROTECTED]> http://www.debian.org/intl/zh/ Come visit Our Lady of Victory Camp! http://www.olvc.ab.ca/ _______________________________________________ I18n mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/i18n