Newsgroups: comp.lang.apl
Path: watmath!watserv1!70530.1226@compuserve.com
From: Mike Kent <70530.1226@CompuServe.COM>
Subject: APL2 external function CTN
Message-ID: <920328054915_70530.1226_CHC16-1@CompuServe.COM>
Sender: root@watserv1.waterloo.edu (Operator)
Organization: University of Waterloo
Date: Sat, 28 Mar 1992 05:49:16 GMT
Lines: 21

In argicle <1992Mar27.133322.10579@watson.ibm.com>, gerth@watson.ibm.com
(John Gerth) writes

 > In and of itself, CTN was a mistake ...

That's a bit stronger than I'd put it.  My timings indicated that it was
faster than Execute for long strings representing lots of numbers (and a
lot faster than the []vi/[]fi emulations), and my fault-finding was
primarily directed at its unsuitability for parsing out interactive user
input.  We used it for converting file data from MVS production programs
(after converting "-" to high-minus as needed), where we were reasonably
sure that the strings we were feeding it would always be valid input.  It
converted this stuff into several gigabytes of data which now reside in
component files on VM (more than once for a lot of it while we were
testing database loads) ... and when I left at the end of last year it
was adding a significant portion of about 500 megabytes a month to the
on-line systems.   

CTN is quite useful, when you don't need the full power (and separate
facilities) of []vi/[]fi.

