develooper Front page | perl.perl5.porters | Postings from October 2013

Re: [perl #120363] perl-5.18.0 apparently miscalculates the NV 1e-298

Thread Previous
From:
Zefram
Date:
October 28, 2013 11:09
Subject:
Re: [perl #120363] perl-5.18.0 apparently miscalculates the NV 1e-298
Message ID:
20131028110909.GH21945@fysh.org
sisyphus1@optusnet.com.au wrote:
>And this would be a bug in the constant folding, I presume.

Well, it's a bug that the constant folding doesn't match the runtime
behaviour, but the constant folding was more accurate.  So the main bug
is in libc, having an inaccurate powl().

I reported it as a libc bug via Debian
<http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=727841>, and it turned
out it's been fixed already.  glibc 2.13 has the inaccurate powl(); in
glibc 2.17 powl() produces the same (correct) result as gcc's constant
folding.

-zefram

Thread Previous


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About