Characters in Squeak (was Re: The Mosner bit)

Yoshiki Ohshima ohshima at
Tue Sep 5 23:39:08 UTC 2000


> >   Anyway, I'm trying to find the time to incorporate those
> > changes into 2.8 release and re-design the framework.
> I hope I havn't made you feel compelled to do that, and that you work because
> you enjoy it.

  Oh, no no, I'm trying to find time because I want to do
so.  Recently constructive discussions have been made in the
Japanese Squeakers and couple of other people have shown the

# By the way, the question about "collecting dust" was not
# rhetorical one.  I didn't know what it means:-)

> No, I meant the fact that it isn't easily accessible one way or other.
> It isn't enough that a framework like yours in fact exist, it should be
> well known to exist and easily accessible. This casts no shadow on your
> work.

  Ok.  So, your argument [it should be well known to exist
and easily accessible] is not only for the multilingual
stuff, but also Squeak itself:-)

> The question of how the two lowest bits are interpreted is a different
> issue, although related which lead it into this direction.

  Sorry, I should improve my English.

> If you have read the rest of this thread I would appreciate your
> comments on the immediate object idea. I would guess it fits well with
> Asian (= large) character sets and your framework?

  Umm, I second the Andreas' opinion here.  Using
"flyweight" pattern for 64k (or more) characters is really
space (and speed) efficient than creating instances, storing
the code point in its field and comparing the value?

  And if what you need is the fixed table of instances of
(large) character set, you can do it without using tag bit.

  By the way, 8 bits of a 32 bit word (or 7 bits of 31 bit
word) is used for the "tag" to identify the character

  -- Yoshiki

More information about the Squeak-dev mailing list