frexpl bugs remain on OS X 10.5 intel

[Gary V. Vaughan]

Hi Bruno,




$ VERBOSE=1 make check TESTS='test-frexpl test-printf-frexpl'
[[...]]
make  check-TESTS
i:-16382 exp:-16384
../../../tests/gnu/test-frexpl.c:136: assertion failed

FAIL: test-frexpl

../../../tests/gnu/test-printf-frexpl.c:95: assertion failed

FAIL: test-printf-frexpl
===============================
2 of 2 tests failed
Please report to [EMAIL PROTECTED]
===============================



tests/gnu/test-frexpl.c:135:fprintf(stderr, "i:%d exp:%d\n", i, exp);




Cheers,
        Gary
--
  ())_.              Email me: [EMAIL PROTECTED]
  ( '/           Read my blog: http://blog.azazil.net
  / )=         ...and my book: http://sources.redhat.com/autobook
`(_~)_

In preparation for some libtool patches against Eric's reports, I decided to upgrade my gnulib checkout and rebuild m4 HEAD against it. I'm on the opposite side of the planet to my ppc machine for the next few months, but there are /\(still\|new\)/ frexpl bugs in gnulib on my intel MacBook: /bin/sh: line 1: 49522 Abort trap EXEEXT='' EXEEXT='' EXEEXT='' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' EXEEXT='' LOCALE_FR='none' LOCALE_FR_UTF8='fr_FR.UTF-8' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' ${dir}$tst exp=0XF.FFFC002P-16417 mantissa=$0X0P+0 my_ldexp<=0X8.FE01A800000000CP-16386 /bin/sh: line 1: 49528 Abort trap EXEEXT='' EXEEXT='' EXEEXT='' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' EXEEXT='' LOCALE_FR='none' LOCALE_FR_UTF8='fr_FR.UTF-8' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' EXEEXT='' srcdir='../../../tests/gnu' ${dir}$tst I added these printfs to the failing test sources to generate the extra output: tests/gnu/test-printf-frexpl.c:94:fprintf(stderr, "exp=%LA mantissa=$ %LA my_ldexp<=%LA\n", exp, mantissa, my_ldexp(1.0L, i - LDBL_MIN_EXP)); The mysteries of double precision representation in memory continue to confound me, but if I can provide you with anything else to help fix this, please let me know.

PGP.sig
Description: This is a digitally signed message part