>LZW can be really slower than huffman. Currently my routines >on the TI92 are very very slow, because of the dictionnary. >that's why i plan to program a "hashing table" (can we say that in english >?it is "Table de >haschage" in french) Yeah, it's "hashing table" or "hash table" in English. That's usually the best way to speed up an LZ-variant algorithm...good luck. :-) --Cliff Biffle