Help Please

Michael Sweet mike at
Mon May 21 14:56:56 GMT 2001

Kenichi Okuyama wrote:
> ...
> If you really know about Unicode, you must not have selected UTF-8
> as internal code. Not UCS2 itself too. You should be using 32bit
> fixed variable which each word points one character, to treat
> multi-word characters.

Um, you *do* realize that doing that will quadruple memory use and
make SAMBA incompatible with every app & OS on the planet overnight,
don't you?

UTF-8 is the only accepted way (currently) to support Unicode
characters properly under UNIX (that is, for Unicode filenames,
etc.)  It is also what is being used to support Unicode in
dozens (hundreds?) of network protocols, simply because it
works best.

I understand that you might not want the overhead of UTF-8
encoding for a 32-bit character, but how many Unicode glyphs
actually need all 32-bits?  IIRC most Asian glyphs can be represented
by 3-5 UTF-8 bytes, so on average there is no overhead (memory-wise)
to use UTF-8 instead of UCS4.  In the worst case you'll need a
7 byte sequence to represent a 28-bit Unicode glyph, something
that is not defined at this time and somehow I don't think it
will be unless the Unicode committee poorly allocates the
Unicode values.  (we're talking about 0.5 billion glyphs for
a 28-bit value...)

On the other hand, using UCS4 introduces a guaranteed 300%
overhead for non-Asian languages, and also introduces an
extra layer of conversions, functions, etc. just to support

> ...
> AT> I think that you have very little understanding of how free software
> AT> is developed.
> I think YOU don't, Andrew.

So where is your imaginary "Kenichi-ware" software that does all
these things, hmm???

The strength of the free software development model comes from the
contributions of developers and users.  If you don't like how SAMBA
is handling your language of choice, SUBMIT A PATCH THAT FIXES IT!
If you are unqualified to do this, write down what needs to be
done to support it, and submit a polite request that it be
implemented.  Don't expect people to respond well to unwarranted
criticism or to jump whenever you say something is wrong.  That's
not how the world works.

Or to use a Western expression: "You can catch more flies with honey
than with vinegar."

> There's no such thing as "free software development model" that you
> think there is.
> Several free software succeeded, because they stick to small
> functionality and kept code small, so small that they didn't need to
> care about multi-lingual.

Um, how about GNOME and KDE, both of which support I18N quite well
(or so I've heard)?

Michael Sweet, Easy Software Products                  mike at
Printing Software for UNIX             

More information about the samba-technical mailing list