* javaop.h (jfloat, jdouble): Make them structures mirroring
the bit fields of IEEE float and double respectively.
(JFLOAT_FINITE, JFLOAT_QNAN_MASK, JFLOAT_EXP_BIAS,
JDOUBLE_FINITE, JDOUBLE_QNAN_MASK, JDOUBLE_EXP_BIAS): New.
(union Word, union DWord): Delete.
(WORD_TO_FLOAT, WORDS_TO_DOUBLE): Update to match.
* gjavah.c (java_float_finite, java_double_finite, F_NAN_MASK,
D_NAN_MASK): Delete.
(jni_print_float, jni_print_double): New. Generate
hexadecimal floating constants.
(print_field_info): Use jni_print_float/double.
* jcf-dump.c: Include math.h. Use ldexp/frexp to assemble
finite floating point numbers for output; special case
non-finite floats.
From-SVN: r64671
2000-01-31 Scott Bambrough <scottb@netwinder.org>
* gcc/java/javaop.h (WORDS_TO_DOUBLE): Allow WORDS_TO_DOUBLE to
assemble doubles correctly when HOST_FLOAT_WORDS_BIG_ENDIAN is
defined to be 1.
From-SVN: r31777
* Makefile.in (gjavah.o): Depend on $(CONFIG_H) and system.h.
* gjavah.c: Include config.h and system.h.
* javaop.h (inline): Don't define, its handled by system.h.
(WORD_TO_FLOAT, WORDS_TO_LONG, WORDS_TO_DOUBLE): Change these
from `inline' to `static inline'.
* jcf.h (inline): Don't define, its handled by system.h.
* lex.c (inline): Likewise.
From-SVN: r24952
