From f977124d05964e9ab3139e44c1d1c3dda8e746af Mon Sep 17 00:00:00 2001 From: "Andreas J. Koenig" Date: Sat, 20 Dec 2008 18:00:01 +0100 Subject: [PATCH] recount a little about the first findings --- lib/Acme/Study/Perl.pm | 21 ++++++++++++++++++++- 1 file changed, 20 insertions(+), 1 deletion(-) diff --git a/lib/Acme/Study/Perl.pm b/lib/Acme/Study/Perl.pm index 864f326..cf6ab9a 100644 --- a/lib/Acme/Study/Perl.pm +++ b/lib/Acme/Study/Perl.pm @@ -33,6 +33,26 @@ multiple platforms. The single function C only does some trivial formatting which will help deciphering the output of the testsuite. +=head1 EXAMPLE + +My first descriptive table overview after I first released +Acme::Study::Perl was generated with the following command: + + ctgetreports --q meta:perl -q conf:use64bitint -q conf:use64bitall --q qr:"(native big math float/int.*)" Acme-Study-Perl + +The result showed me six perls that have a bug in their conversion of +strings to numbers and 28 perls that did it right. To find out more +about the parameters that might have an influence, I called +ctgetreports with --solve and --ycb (example in the ctgetreports +manpage). + +The calculation provided a hint that conf:ivtype is a candidate. +Looking closer helped confirm the finding that all perls with ivtype +C have this bug. And only one architecture with ivtype +C, namely C. All others either provided the +correct result, C<-1> because they were native 64 bit processors or +the expected wrong-by-rounding result C<0>. + =head1 EXPORT =head1 FUNCTIONS @@ -105,7 +125,6 @@ L =back - =head1 ACKNOWLEDGEMENTS Thanks to RJBS for Module::Starter. -- 2.11.4.GIT