[mod.computers.vax] Point and buffers problem in TPU

Bakin@MIT-MULTICS.ARPA.UUCP (07/30/86)

Thanks Paul (Paul S. Winalski) for giving the details I left out!

-- Dave

Bakin@MIT-MULTICS.ARPA.UUCP (08/03/86)

More:  Note that SCAN will never return a 0-length match, furthermore
pattern assignment will never assign a 0-length match.  There are
undoubtably other problems as well.

I STRed this (and a problem where in nested pattern matching if the
inner pattern fails the message "Search Failed" is repeatedly issued,
once for each time the pattern is retried, with no way to turn it off)
in December 1985.  I didn't get a response, does anyone know if it was
answered?

The developers clearly had N.I.H.:  They didn't look at Emacs.  Then
when they had finally developed the thing the only acceptance test
performed at Digital was "Will it do everything EDT does?"  Some of the
primitives are clearly designed for EDT emulation, not for general use.

Now that Epsilon is available on the IBM PC and clones, and GNU Emacs
is ported to VMS, I'm not going to bother with TPU.  The confusion of
0-length and 1-length strings due to the misdefinition of the cursor
position is crippling even if it didn't have other problems.

-- Dave (Bakin -at mit-multics.arpa)