[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Scheme-reports] Padding/placeholders (hashes) in numerical syntax

 | Date: Sun, 4 Sep 2011 16:59:18 -0400
 | From: John Cowan <cowan@x>
 | Peter Bex scripsit:
 | > Later I was told that this might be reconsidered if it was shown
 | > to be really difficult to implement.
 | It's on the ballot now, and so far there are 4 votes in favor to 0
 | against to remove it.  We won't be sure for a week or so, though.
 | > The latest version of this test can be found here:
 | I picked up version 24935.
 | > Outputs of other Schemes would be interesting to see as well, and
 | > suggestions for new testcases are welcome too!
 | I ran tests against Bigloo, Chez, Chibi, Ikarus, IronScheme, Kawa,
 | Larceny, Mosh, SCM, SISC, STklos, Ypsilon on a 32-bit Linux
 | system.  I tried Scheme 9, but it fails on the syntax-rules
 | declaration.  The modified scripts and results are at
 | http://www.ccil.org/~cowan/temp/strconv-results.zip .
 | Here's what I got:
 | ...
 | SCM: Inexact complex numbers only.  Test "+nan.0+nan.0i" blows up
 | with "Wrong type passed to make-rectangular: 0/0" error.  When this
 | is commented out, 28 errors.

Making string->number case insensitive for infinities and nans
<http://groups.csail.mit.edu/mac/ftpdir/users/jaffer/scm.zip> reduces
this to 19 "ERROR"s.  Seven of these are "SERIALIZATION ERROR"s due to
SCM using engineering notation for numbers greater than 1000. or less
than -1000, or between -1 and 1.  I believe that R5RS permits

(define (nan? x) (and (number? x) (not (complex? x)))) ; SCM

Scheme-reports mailing list