This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Aw: Re: TR1 Special Math
- From: Joseph Myers <joseph at codesourcery dot com>
- To: Ed Smith-Rowland <3dw4rd at verizon dot net>
- Cc: Szabolcs Nagy <szabolcs dot nagy at arm dot com>, Florian Goth <CaptainSifff at gmx dot de>, Jonathan Wakely <jwakely dot gcc at gmail dot com>, libstdc++ <libstdc++ at gcc dot gnu dot org>, gcc-patches <gcc-patches at gcc dot gnu dot org>
- Date: Thu, 19 Nov 2015 22:07:30 +0000
- Subject: Re: Aw: Re: TR1 Special Math
- Authentication-results: sourceware.org; auth=none
- References: <CAH6eHdSK80f08Fr+2gAiX_+QpmfHOgv+-bTtM0DxtVPmRwTykw at mail dot gmail dot com> <554CC2B4 dot 80401 at verizon dot net> <CAH6eHdQ+OkErGe_fXh1k4pzr3vaKUi=XteBu+AymeHCO-iQfDA at mail dot gmail dot com> <562D1571 dot 2070207 at verizon dot net> <CAH6eHdRjG8T9nXkguUzhgjtXGYm9xj+vUnWU+J1yc79OHCqn8A at mail dot gmail dot com> <CAH6eHdQ7gMAzU_OBY5fRXujDQXMqKbSt95QRbFN6UaNjvtZHqw at mail dot gmail dot com> <56460DE2 dot 3050905 at verizon dot net> <56479952 dot 50804 at verizon dot net> <56479C44 dot 4010501 at verizon dot net> <trinity-2724aee0-1bf0-49a2-9a5e-165a8bb4eae9-1447720111375 at 3capp-gmx-bs12> <564A8A31 dot 2070107 at verizon dot net> <564B20A4 dot 1000104 at arm dot com> <564E1608 dot 8040401 at verizon dot net>
On Thu, 19 Nov 2015, Ed Smith-Rowland wrote:
> Testing coverage is growing.
> I starting to build check_inf tests for everyone (but they aren't in the
> patch.
> I'll look at musl to see how you do check_denorm, etc.
FWIW, in glibc I've systematically put special cases and a representative
range of input values in auto-libm-test-in (where the test inputs and
mathematical results are finite, but possibly overflowing / underflowing)
/ libm-test.inc (where not all finite). Random test generation has also
been done with both mpcheck and ad hoc scripts that put random inputs in
auto-libm-test-in before regenerating auto-libm-test-out and running the
testsuite, in order to find cases with large errors or spurious or missing
exceptions (those cases then being added to auto-libm-test-in when the
bugs are fixed). This all relies heavily on having MPFR / MPC support for
the functions in question in order to compute the ideal correctly rounded
results for all rounding modes (so if new functions were added to glibc,
it would be a very good idea to add them to MPFR / MPC first).
--
Joseph S. Myers
joseph@codesourcery.com