* sysdeps/ieee754/ldbl-128ibm/s_nextafterl.c (nextafterl): Remove

* sysdeps/ieee754/ldbl-128ibm/s_nextafterl.c (nextafterl): Remove

	unused ily variable.  Fix nextafterl on +-__LDBL_MAX__ and +-Inf.
	Remove unreachable code at the end.

2007-06-01  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/ieee754/ldbl-128ibm/s_fpclassifyl.c: Correct description of
	ldbl-128ibm in comment.
	(fpclassifyl): Correct classification of denormals.
	* sysdeps/ieee754/ldbl-128ibm/s_nextafterl.c (nextafterl): Correct
	return value for MIN denormal. Rewrite using long double math too
	correctly handle denormals and canonicalize the results.

2007-06-05  Jakub Jelinek  <jakub@redhat.com>

	* sysdeps/ieee754/ldbl-128ibm/mpn2ldbl.c
	(__mpn_construct_long_double): Fix conversion where result ought
	to be smaller than __LDBL_MIN__, or the low double should be
	denormal.  Fix decision where to negate low double - honor round
	to even rules.
	* stdio-common/tst-sprintf2.c: Include string.h.
	(COMPARE_LDBL): Define.
	(TEST): Also test whether a string hexadecimal float representation
	can be parsed back to the number.
	(main): Add a couple of further tests.

2007-06-04  Jakub Jelinek  <jakub@redhat.com>