Re: NB 8 (Trial version)

["Robert Holmgren" holmgren@xxxxxxxx]

** Reply to message from tebrake@xxxxxxxx on Mon, 08 May 2006 09:46:56 -0400


> But isn't the problem with these specific
> dictionaries that they can't handle ANSI coding, so unless Steve
> could get permission from the owners to change them, they never will
> be much use to him? Might it not be better to try to license
> thesauri that were made for ANSI in the first place?

Well, NB-NYC hasn't made ANY effort with respect to dictionaries OR thesauri.
They use the same 1987 Microlytics dictionary that Xy3 and up use. Same
thesaurus too. I was pointing out that the IBM spellers now work -- which is
to say, that NB merely *fixed* the broken "hooks" in the engine that *already
existed* from Signature up (they only partially worked in NBWin v5-v6, and I
never tested in v7).

In short, we're seeing here a "path of least resistance" kind of "effort". The
US-UK dictionaries are also ASCII, not ANSI -- but it doesn't matter with
English, because English scarcely ever uses accented (=8bit, Ascii 128-255)
characters. So they "get away with it" in English.

> Or are you
> suggesting that it should be possible for NB to make these
> dictionaries work as is by creating a shell to make the necessary
> substitutions?

Precisely -- simply a filter, which substitutes X for Y. For example, a pound
sign £ is Ascii-156 and ANSI-163. So you substitute ASCII for ANSI during
lookup, and then substitute ANSI for ASCII during replacement. 

The problem, really, may be deeply buried in the code of the engine. Both
spelling and the thesaurus are controlled by rather opaque commands: SPELL,
and function SY (SYnonym). Users have no way of intercepting either of those
commands (if we could, I'd just write the filter in a jiffy). Well... I
suppose, it _could_ be done by using the SO SpellOne function, and moving from
word to word individually; but it would *probably* be slow. Hmmm... Might be
worth trying anyway...

-----------------------------
Robert Holmgren
holmgren@xxxxxxxx
-----------------------------