Re: Microlytics and London, etc

[Robert Holmgren holmgren@xxxxxxxx]

Paul:

OK, just kidding about ability to spell. (But, actually, I
do proofread, and when done, there aren't any spelling
errors.)

I have access now to a Win95 OSR2 machine...

Well, this is most interesting, and certainly the crux of
the problem: it's choking on abbreviations (you may have
"trained" it to accept them, via a personal spelling list,
but by default they're all going to be rejected).
"Cutlerian" aside (whatever that means... I choke on that
too), they're almost all alphanumerics.

Not to rub it in, but I gotta ask, why the hell didn't you
mention before this transparently obvious pattern of
crashing only on a particular type of $tring????? Grrrrh.
This is why I've stopped participating in the mail list;
it's a huge time drain, and for what? Somebody says
"thanks", and you're out hours of time... (for which, by the
bye, I charge my clients US$250/hour).

What does your machine return on "VA/NV CK", the num
checker? If it's zero, try changing it to 1 ("d ck=1"),
see if that makes a difference. Under Win2K, with CK=0, all
these $trings are flagged as errors by the Microlytics UK
speller: 25mm IEC 6500MW 11th 280gm 22nd. But NONE of them
crash the machine. And when I alter them to proper format
("magazine style" notwithstanding), e.g. "25 mm, 6500
megawatt, eleventh, 280 gm, twenty-second", none of them are
flagged, period. When I change to CK=1, none of the
alphanumerics in this paragraph get flagged -- only abbrevs
like "IEC" "VA" "NV" "CK". So that modifies the behavior by
masking part of the problem, but it doesn't actually solve
it. No effect on Cutlerian, I'd suppose...

When I go over to Win95, and SPELL this present message to
you (written in XyWrite), it flags Cutlerian (which I F1-
"Ignore"), then crashes Editor on "25mm" (the very first of
the $trings that crash for you too). It flags "IEC" but
does NOT crash (you know, I'd bet money that you're wrong
that IEC and Cutlerian always crash your machine: I very
much doubt that they "always" do that -- at least, not until
some internal instability in W9x has been triggered; I
suspect that the only 100% consistent crash occurs on
alphanumerics). With CK=1, most -- maybe all -- of these
errors are circumvented, and the error (which has got to be
buried in the DOS subsystem, probably a MALLOC [memory
allocation] error) is masked. I think that the reported
Overflow File "error" is irrelevant, probably a coincidental
result of fiddling with memory settings but entirely
unrelated to this problem.

Anyway, this is DEFINITELY a Win9x issue -- neither more nor
less. We are truly beating a doornail-dead horse here.