Thanks everybody, Dicts are working fine now, for my tests with UInt64 and
UInt128s as keys... (I haven't benchmarked it against my old stuff, I'll
have to do that at some point ;-) )
What I'm doing right now isn't really performance critical (I, of course,
am obsessively interested in any performance topic!), I am testing building
different data structures for handling
the tasks that utf8proc is used for right now... that won't take 1.5MB, and
will be much faster... (in the database world, a lot of what I'm doing is
changing it from a row store to a column store...).
I don't know how much time in compiling Julia is spent on seeing if a
character is a start identifier character or a plain identifier
character... but this should speed those checks up quite a bit.
BTW: Jacob & Steven, you are both very evil, pointing me at another need
performance thing to get obsessed about! Now I'll start thinking about
Judy as well as Julia... (Judy is LGPL, which is OK
for making a wrapper, but I don't think that would allow me to rewrite it
in Julia, would it?)
