Jan Vroonhof <vroonhof(a)math.ethz.ch&gt; writes: I see two potential problems with lrecord characters. First is efficiency -- examining characters from Lisp would become even slower than it is now because it would involve the overhead of "creating" the character objects. This could be partially solved by implementing the small/big character thing, but it still wouldn't help with, say, large Japanese texts. The second problem is with GC -- any Lisp object storing a character will then need to be GCPRO'ed. I don't know if this is a big problem or not, but GCPRO troubles /are/ hard to debug so we should consider them carefully.