Re: Case change bug workaround

["..." adpf@xxxxxxxx]

≪	In v3 xpl, where the UD buffer can strain memory, this routine
that clears it can make or break a long program:
 	GT XD DF DF RD XD
≫ --me

≪	Annie, did the UD buffer share the 64K that v3 XPL also used? ≫ --Carl

Hey, Carl, that's the kind of question I'm supposed to ask *you*.
My relationship with xyWrite is purely pragmatic. I don't know why
it does anything it does. I just study the effects and go with what works.
So I certainly can't answer authoritatively, but ...

≪	Wouldn't that have to be true for this concept to work?
Otherwise a big XPL should break the memory bank regardless of UD. ≫

... xpl *acts* as though they share a buffer. The impressive v4 UD and
xpl capacity--even in a system like mine, with no memory configured as
expanded--indicates that v4 changed their structure bigtime.

I should say that I'm using "big" as shorthand. This applies as well
to programs that aren't necessarily overlong but that put heavy demands
on resources. I've found again and again that adding that routine cures
flaky v3 xpl behavior, which is how low memory seems to manifest
itself--e.g., after shelling out, a pgm treats a s/g sv'd before
as existent but empty. When running xpl I don't remember ever seeing
an error msg directly related to memory other than msgs caused
by a programming error.

≪	Or is it that a UD overflow would render Xy generally unstable? ≫

Well, when UD is overtaxed, v3 is draconian--"out of memory" and empty
buffer--but that seems to be that. General instability is something
I never experience in xyW 3. 			--a

============================= adpFisher  nyc