Re: Patch for DateTime::Format::W3CDTF to respekt second fragment part

2007-11-08 Thread Zefram
Jonathan Leffler wrote:
Or, what happens if there are 10 digits after the decimal point.

That's what the substr() is for.

Yeah, mostly academic, except I'm working towards a time type (not for 
Perl per se) that extends down to picoseconds (and up to 10^12 years, 
too).

At least 105 bits, then.  TAI64 covers nearly 10^12 years, of course,
and the extension TAI64NA goes down to attoseconds in a 128-bit format.
What's the concept behind your type?

Personally I like to use bignum rational arithmetic, for unlimited
resolution.  Performance is atrocious with the Perl bignum libraries,
unfortunately.

-zefram


Re: Patch for DateTime::Format::W3CDTF to respekt second fragment part

2007-11-08 Thread Jonathan Leffler
Zefram [EMAIL PROTECTED] wrote:
 Re: Patch for DateTime::Format::W3CDTF to respekt second fragment part
 
 Julian Haupt wrote:
 +if ($date =~ s/\.(\d+)$// )
 +{
 +   my $fraction = $1;
 +   $p{'nanosecond'} = (1 / $fraction) * 10**9;
 +}
 
 That inversion can't be right.  Surely you mathematically want
 
$p{'nanosecond'} = 0.$fraction * 10**9;
 
 but actually the nanoseconds member is supposed to be an integer, and
 it would be better to avoid floating-point arithmetic entirely:
 
$fraction = substr($fraction, 0, 9);
$fraction .= 0 x (9 - length($fraction);
$p{'nanosecond'} = 0 + $fraction;

Beware the picosecond!

Or, what happens if there are 10 digits after the decimal point.  I've no 
idea what -3 zeroes looks like, but the answer is unlikely to be correct.

Yeah, mostly academic, except I'm working towards a time type (not for 
Perl per se) that extends down to picoseconds (and up to 10^12 years, 
too).

-- 
Jonathan Leffler ([EMAIL PROTECTED])
STSM, Informix Database Engineering, IBM Information Management Division
4100 Bohannon Drive, Menlo Park, CA 94025-1013
Tel: +1 650-926-6921Tie-Line: 630-6921
I don't suffer from insanity; I enjoy every minute of it!

NB: with effect from 2007-11-28, my address and phone number change to:
4400 N First St, San Jose, CA 95134-1257
Tel: +1 408-956-2436 T/L: 475-2436



smime.p7s
Description: S/MIME Cryptographic Signature


Re: Patch for DateTime::Format::W3CDTF to respekt second fragment part

2007-11-07 Thread Zefram
Julian Haupt wrote:
+if ($date =~ s/\.(\d+)$// )
+{
+  my $fraction = $1;
+  $p{'nanosecond'} = (1 / $fraction) * 10**9;
+}

That inversion can't be right.  Surely you mathematically want

$p{'nanosecond'} = 0.$fraction * 10**9;

but actually the nanoseconds member is supposed to be an integer, and
it would be better to avoid floating-point arithmetic entirely:

$fraction = substr($fraction, 0, 9);
$fraction .= 0 x (9 - length($fraction);
$p{'nanosecond'} = 0 + $fraction;

-zefram