Skip Menu |

This queue is for tickets about the Algorithm-LibLinear CPAN distribution.

Report information
The Basics
Id: 84881
Status: resolved
Priority: 0/
Queue: Algorithm-LibLinear
People
Owner: SEKIA [...] cpan.org
Requestors: ANDK [...] cpan.org
Cc:
AdminCc:
Bug Information
Severity: (no value)
Broken in: (no value)
Fixed in: (no value)

History Show all quoted text
  Fri Apr 26 15:11:22 2013 ANDK [...] cpan.org - Ticket created
Subject: Fails on perls compiled with -Duselongdouble due to insufficient rounding
As per subject. Sample fail report: http://www.cpantesters.org/cpan/report/29638929 A bigger sample of fail reports: http://analysis.cpantesters.org/reports_by_field?distv=Algorithm-LibLinear-0.02;field=conf%3Auselongdouble HTH&&Thanks,
  Thu May 02 23:39:15 2013 SEKIA [...] cpan.org - Correspondence added
On 2013-4月-26 金 15:11:22, ANDK wrote: Show quoted text
> As per subject. Sample fail report: > > http://www.cpantesters.org/cpan/report/29638929 > > A bigger sample of fail reports: > > http://analysis.cpantesters.org/reports_by_field?distv=Algorithm- > LibLinear-0.02;field=conf%3Auselongdouble > > HTH&&Thanks,
Thank you for reporting the issue. I think the problem is in test script, not in module itself. Since this module depends on BLAS internally and it uses 64bit floating point number, it seeems not avoidable to occure small precision error on long double-enabled environment. Therefore I modified number equality tests in the test script to allow small error: https://github.com/sekia/Algorithm-LibLinear/commit/4e3370486ae2ed562e606f2d58b369ed722492d3 The version 0.03 will be availeble on CPAN soon. Yours sincerely,
  Thu May 02 23:39:15 2013 The RT System itself - Status changed from 'new' to 'open'
  Thu May 02 23:39:16 2013 SEKIA [...] cpan.org - Status changed from 'open' to 'resolved'
  Thu May 02 23:39:16 2013 SEKIA [...] cpan.org - Taken