Path: utzoo!attcan!uunet!jarthur!elroy.jpl.nasa.gov!jpl-devvax!lwall
From: lwall@jpl-devvax.JPL.NASA.GOV (Larry Wall)
Newsgroups: comp.lang.perl
Subject: Re: variable is numeric?
Message-ID: <7443@jpl-devvax.JPL.NASA.GOV>
Date: 16 Mar 90 19:12:31 GMT
References: <16918@orstcs.CS.ORST.EDU>
Reply-To: lwall@jpl-devvax.JPL.NASA.GOV (Larry Wall)
Organization: Jet Propulsion Laboratory, Pasadena, CA
Lines: 24

In article <16918@orstcs.CS.ORST.EDU> pvo@sapphire.OCE.ORST.EDU (Paul O'Neill) writes:
: What is the Canonical Test of whether a variable is numeric?

I'm not sure what you mean by that.  All scalar variables are potentially
numeric, and any scalar that has been evaluated in a numeric context has
a numeric value.  That value will be whatever atof() returned.  So non-numeric
strings simply have the numeric value 0, along with their string value (a
variable actually can have a string value and a numeric value simultaneously--
hopefully these always stay in sync, but a recent patch fixed substr() because
it allowed the values to get out of sync (one variable is purposefully
forced to be out of sync, and that's $!)).

There is no way that I know of (apart from the magical behavior of
autoincrement) to tell whether a variable has been referenced in a numeric
context.

If you're asking about how Perl decides whether to use $# in printing out
a number, it does it if it already has a non-zero numeric value, or if
the string part matches /^\s*[+-]?\d*(\.?\d*([eE][+-]?\d*)?)?\s*$/ and
atof() says it has a non-zero value.  (Actually, that regexp is a translation
of the C code of looks_like_number(), which is a heap more efficient than
that regexp would be.)

Larry