Skip to content

Latest commit

 

History

History
1173 lines (1051 loc) · 53.7 KB

ChangeLog.old.md

File metadata and controls

1173 lines (1051 loc) · 53.7 KB
Raw

  • Miscellaneous changes

    • Makefile.in (distclean): Add as rule to remove files generated by configure.

  • Miscellaneous changes

    • README.user: Added new section on compliance With ISO/IEC TR 24732, the C++ DFP Draft Technical Report, including instructions on how to use (ios::fixed | ios::scientific) as ios_base::fmtflags in order to get fully encoded precision in output string representation.

  • Miscellaneous changes

    • tests/test-printf.c: Add missing 'D' to a/A conv tests.
    • tests/scaffold.c: Correct comments for _OSC and added a/A support.
    • tests/test-ostream.cpp: Modified expectations such that a/A ignores precision and always uses the num of significant digits in the mantissa when output.
    • tests/test-ostream-g-spec.cpp: Removed unnecessary define that used to be used to get g/G support.
    • dfp/decimal/decimal (FIND_DEC_MANT_DIG): Remove ".*" from default string string. (LIBDFP_META): Make g/G spec default. Enable a/A spec by checking flags for ios::fixed | ios::scientific. Condense uppercase/lowercase check. Don't call printf with a precision specifier if a/A is the desires spec conv.

  • Miscellaneous changes

    • tests/test-negate.c: New testcase.
    • Makefile.in (libdfp_c_tests): Add test-negate.

  • Miscellaneous changes

    • Makefile.in (test-printf.os, test-strtod.os): New rules to depend on printf_dfp.os and strtod[32|64|128].os respectively.
    • strtod128.c (__DEC_MANT_DIG__, __DEC_MAX_EXP__, __DEC_MIN_EXP__): Define to _Decimal128 values.
    • tests/test-strtod.c: Added tests near __DEC*_MAX_EXP__ and __DEC*_MIN_EXP__.
    • strtod64.c (__DEC_MANT_DIG__, __DEC_MAX_EXP__, __DEC_MIN_EXP__): Define to _Decimal64 values.
    • sysdeps/powerpc/dfpu/numdigits.h (left_justify): Fix VSX case by replacing xxlxor with xxlor insn. (getexp): Replace stfwix with stfd to prevent erroneous sign extend.
    • sysdeps/soft-dfp/dpd/numdigits.h (getexp): Make return value congruent with hardware-dfp for NaN.
    • strtod32.c (__DEC_MANT_DIG__, __DEC_MAX_EXP__, __DEC_MIN_EXP__): Define to _Decimal32 values. (FUNCTION_L_INTERNAL): Add left_justify call if proposed exponent is sufficiently close to __DEC*_MAX_EXP__. Add zero return if proposed exponent is smaller than __DEC*_MIN_EXP__.

  • Miscellaneous changes

    • Makefile.in (libdfp_c_tests): Added test-getexp.
    • tests/test-getexp.c: New test.
    • tests/test-decode.c (d64types): Added test for __DEC64_MAX__ and __DEC64_MIN__.
    • sysdeps/powerpc/dfpu/numdigits.h (getexpd[32|64|128]): Fixed in order to prevent erroneous sign extension due to using stfiwx.

  • Miscellaneous changes

    • sysdeps/powerpc/dfpu/numdigits.h: Cleaned up inline asm DEC_TYPE declarations. Removed redundant macros. Merged macros under similar guards.

  • Miscellaneous changes

    • sysdeps/powerpc/dfpu/numdigits.h (numdigits*): Fix inline asm to place result of drrnd[q] into DEC_TYPE instead of double. Update copyright years to range.
    • tests/test-decode.c: Update copyright years to range.
    • tests/test-param.c: Likewise.
    • tests/test-printf.c: Likewise.
    • tests/test-numdigits.c: Likewise.
    • tests/test-istream.cpp: Likewise.
    • tests/test-strtod.c: Likewise.
    • sysdeps/dpd/dpd-private.c: Likewise.
    • printf_dfp.c: Fix erroneous test for _Decimal64. Update copyright years to range.

  • Miscellaneous changes

    • Makefile.in (check): Add printf_dfp.o dependency to check in order to have make check automatically rebuild libdfp is printf_dfp.c has changed.
    • tests/test-decode.c: Add tests for decoding numbers with high exponents near __DEC*_MAX_EXP__.
    • tests/test-param.c (param_test): Remove unnecessary assingments.
    • tests/test-printf.c: Add tests for __DEC64_MAX__.
    • tests/test-numdigits.c: Add tests for __DEC64_MAX__ and numbers approaching __DEC*_MAX_EXP__.
    • tests/test-GCC-PR52140.c (main): Remove unnecessary assignments.
    • tests/test-ostream.cpp: Formatting.
    • tests/debug-test.sh: Add ./ as a path prefix.
    • sysdeps/powerpc/dfpu/numdigits.h (numdigitsd*): Fix for numbers with exponent approaching __DEC*_MAX_EXP__ by biasing the exponent toward zero.
    • sysdeps/dpd/dpd-private.c: Formatting.
    • TODO: New items.
    • printf_dfp.c (__printf_dfp): Fix for g/G specifier that converst to f/F form. Fix for a/A when the user supplied precision is 0, i.e. %.Ha.

  • Miscellaneous changes

    • Makefile.in (libdfp_c_tests): Add test-GCC-PR52140.c.
    • tests/test-istream.cpp: Comment out NaN tests. Two nans can't be compared.
    • tests/test-GCC-PR52140.c: New testcase for ICE.
    • TODO: Add item to upgrade libdecnumber version.

  • Miscellaneous changes

    • Makefile.in (install-readme): New target. (install): Consolidated by making install-headers and install-readme as dependent rules.
    • README.maintainer: Updated info on Versioning.
    • README.user: Added discussion on __STDC_DEC_FP__
    • README.developer: Updated info on Versioning.

  • Miscellaneous changes

    • sysdeps/powerpc/powerpc32/power6x/Implies: Points to power6.
    • sysdeps/powerpc/powerpc32/power6x/fpu/Implies: Points to power6.
    • sysdeps/powerpc/powerpc64/power6x/Implies: Points to power6.
    • sysdeps/powerpc/powerpc64/power6x/fpu/Implies: Points to power6.

  • Miscellaneous changes

    • Makefile.in (libdfp_cxx_tests): Add test-ostream-g-spec test.
    • tests/test-ostream-g-spec.cpp: New file to test decimal/decimal's _LIBDFP_G_CONV_SPEC guard.
    • tests/test-printf.c: Add newly discovered %DDg failures.
    • tests/scaffold.c (_OSC_P): Add comment regarding _LIBDFP_G_CONV_SPEC.
    • tests/test-istream.cpp: Removed unused variable.
    • README.user: Update with information about NOT using #include <dfp/header>.
    • dfp/float.h: Remove commented out decimal class predeclares.
    • dfp/decimal/decimal: Move #include <float.h> to before #include_next <decimal/decimal> to prevent missing _Decimal[32|64|128] definition errors.

  • Miscellaneous changes

    • Makefile.in (libdfp_cxx_tests): Added test-istream.
    • tests/scaffold.c (_ISC _ISC_P): Added new istream compare macros conditional on #ifdef __cplusplus.
    • tests/test-istream.cpp: New test for operator>>.
    • tests/test-strtod.c: Added a new value to test.
    • dfp/float.h: Correct _Decimal[32|64|128] declarations.
    • dfp/decimal/decimal: Added operator>> support for decimal[32|64|128] C++ decimal floating point types.
    • README.user: Updated information on operator<< and operator>> support.

  • Miscellaneous changes

    • tests/test-log10d.c: Added missing #include "decode.h".
    • tests/scaffold.c (_OSC _OSC_P): Added new ostream compare macros conditional on #ifdef __cpluspluc.
    • tests/test-ostream.cpp: Added scaffold.c based tests.

  • Miscellaneous changes

    • dfp/decimal/decimal(operator<< decimal128): Removed now spurious call to ostream_d128() (which has been removed from libdfp) in the decimal128 operator<< overload. Updated comments clarifying use of 'a/A' over 'g/G' for the default.

  • Miscellaneous changes

    • dfp/decimal/decimal: Added back in erroneously removed #include <stdio.h>.

  • Miscellaneous changes

    • dfp/decimal/decimal: Removed ostreamd32, ostreamd64, and ostreamd128 prototypes.

  • Miscellaneous changes

    • decode.h: Added __BEGIN_DECLS and __END_DECLS around decoded[32|64|128] prototypes in order to allow them to be used in C++ compatibility testcases without warnings.
    • scaffold.c: Removed extra #include "decode.h".
    • test-strtod.c: Likewise.

  • Miscellaneous changes

    • Makefile.in: Copied contents of $(system_header_dirs) into $(header_dirs) in order to have -I added automatically to the compiler invocation. This makes sure that dfp/decimal/decimal always comes before the system include/decimal/decimal header.

  • Miscellaneous changes

    • README.user: Add instructions for using <decimal/decimal> to pick up operator<< and operator>> overloads for decimal[32|64|128] types.
    • README.developer: Removed necessity of including -isystem in CPPFLAGS in order to pick up decimal headers.
    • dfp/decimal: Created dfp/decimal/ header include directory.
    • dfp/decimal.tmp: Moved decimal.tmp to decimal/decimal.
    • dfp/decimal/decimal: Moved decimal.tmp to decimal/decimal.

  • Miscellaneous changes

    • Makefile.in: Install dfp/decimal/ directory. Add dfp/decimal/decimal path.
    • tests/test-ostream.cpp: Change to <decimal/decimal>
    • TODO: Added directive to test decimal/ directory installs.
    • dfp/decimal: Moved file "decimal" to "decimal.tmp".
    • dfp/decimal.tmp: Moved file "decimal" to "decimal.tmp".

  • Miscellaneous changes

    • configure: Regenerated.
    • Makefile.in: (@CXX@, @LLD@, @OBDJUMP@, @GDB@, @CXXFLAGS@, and @CPPFLAGS@): New imports. (header_dirs): Move dfp/ to $(system_header_dirs). (CXX_DEFINES CXX_WARNS): New variables. (.c.o): Add $(CPPFLAGS). Add -I$(system_header_dirs). (.c.os): Likewise (.S.o): Add -I$(system_headers). (.cpp.os) Likewise. (GLIBC_LIBS): Remove unnecessary libpthread.so linkage. (libdfp_cxx_tests): New variable. (libdfp_tests): Renamed to (libdfp_c_tests) and replaced with $(libdfp_c_tests) $(libdp_cxx_tests). Added new rule for $(libdfp_cxx_tests).os dependencies on float.h. Added new rule for $(libdfp_tests) to rely on tests/scaffold.c. Added new rule to link $(libdfp_cxx_tests) to their .os files. (.SUFFIXES): Added .cpp.
    • ieee754r/lgammad32.c: Changed <dfp/math.h> to <math.h>.
    • tests/test-printf.c: Discovered a bug in %.Ha. If there is a '.' with no precision specifier it prints a bogus value. This adds this failure to the test suite.
    • tests/test-ostream.cpp: New test to test operator<< header file additions in dfp/decimal.
    • configure.ac: Added AC_PATH_PROG for OBJDUMP, LDD, GDB. Added AC_SUBST for CXXFLAGS, CPPFLAGS.
    • README.user: Added instructions for printf "a,A" Conversion Specifier usage.
    • README.developer: Added instructions for specifying PATH to pick up toolchain default programs. Added instructions for LDD, GDB, OBJDUMP, CXX, CXXFLAGS, and CPPLAGS.
    • Versions.def
    • Makefile.gdb: Use configure exported 'objdump' via $(OBJDUMP). Use configure exported 'ldd' via $(LDD). Use configure exported 'gdb' via $(GDB).
    • dfp/stdlib.h: Comment update.
    • dfp/float.h: New file adds _Decimal[32|64|128] types #ifdef __cplusplus and if they aren't added by the compiler. At the time of this writing the compiler hasn't added them to float.h yet.
    • dfp/decimal: New file adds operator<< for decimal[32|64|128] C++ types. This use C++ template metaprogramming to create functions that call printf.
    • printf_dfp.c: Comment change for a/A spec char.
    • TODO: New item.

  • Miscellaneous changes

    • Makefile.in (libdfp_tests): Add test-expd, testing expd[64|128](-1.0) which was previously failing. (libdfp_tests.so): New dummy rule to add $(top_srcdir)/tests/scaffold.c as a dependency so that scaffold.c changes result in rebuilding the libdfp_test .so files on make check.
    • tests/test-expd.c: New file testing expd[32|64|128].
    • tests/scaffold.c (_AVC _AVC_P): New macros which do a value compare but allow a +/- variation.
    • README.user: Added instructions for C++ types compatibility with ISO-C DFP types.
    • sysdeps/powerpc/dfpu/expd64.c: Existing code only accounts for positive 'n' in e^n. This fix allows for e^-n, which is just 1/(e^n).

  • Miscellaneous changes

    • README.user: Updated with instructions for using #include <dfp/ >.
    • sysdeps/soft-dfp/dpd/Versions: Moved __dpd_extend* and __dpd_trunc* symbol definitions from here to...
    • sysdeps/dpd/Versions: here, since these are needed for all versions of libdfp, not just the soft-dfp version.

  • Miscellaneous changes

    • Makefile.in: Replace erroneous usage of $? with $^. Fix libdfp_tests based rules to correct prerequisite usage so that a complete rebuild isn't required when only a few of the test case dependencies change.
    • Makefile.gdb: Correct prerequisite usage to not require a complete rebuild when only a few of the test case dependencies change.

  • Miscellaneous changes

    • dfp/stdlib.h: Added #include <features.h> to correct include order problem.
    • dfp/wchar.h: Likewise.
    • tests/test-stdlib.c: New test to verify <features.h> inclusion.
    • tests/test-wchar.c: New test to verify <features.h> inclusion.
    • Makefile.in (libdfp_tests): Added test-stdlib and test-wchar.

  • Miscellaneous changes

    • configure: Regenerated.
    • configure.ac: Incremented the version number to 1.0.8.
    • Versions.def: Incremented the version number to 1.0.8.
    • tests/test-bfp-conversions.c: Added tests for extendsddf extendsdtf extenddfdd extendddtf extenddftd and extendtftd.

  • Miscellaneous changes

    • base-math/Makefile: Moved 'classify' from sysdeps/soft-dfp/Makefile.
    • base-math/classify.c: Moved from sysdeps/soft-dfp/ because it's required for the power6 build as well.
    • base-math/extendsftd.c: Likewise.
    • base-math/extendsdtf.c: Likewise.
    • base-math/extenddfdd.c: Likewise.
    • base-math/trunctdsf.c: Likewise.
    • base-math/trunctfsd.c: Likewise.
    • base-math/extendddtf.c: Likewise.
    • base-math/extenddftd.c: Likewise.
    • base-math/extendsfsd.c: Likewise.
    • base-math/extendtftd.c: Likewise.
    • base-math/truncdddf.c: Likewise.
    • base-math/trunctfdd.c: Likewise.
    • base-math/trunctddf.c: Likewise.
    • base-math/truncsdsf.c: Likewise.
    • base-math/trunctdtf.c: Likewise.
    • base-math/extendsddf.c: Likewise.
    • base-math/extendsfdd.c: Likewise.
    • base-math/truncddsf.c: Likewise.
    • base-math/truncdfsd.c: Likewise.
    • tests/scaffold.c (_VC_P): Updated spacing in "Expected" and "Result" output.
    • tests/test-bfp-conversions.c: Added float -> _Decimal128 tests.
    • sysdeps/soft-dfp/Makefile: Moved 'classify' to base-math/Makefile.
    • sysdeps/soft-dfp/classify.c: Moved to base-math/ because it's required for the power6 build as well.
    • sysdeps/soft-dfp/extendsftd.c: Likewise.
    • sysdeps/soft-dfp/extendsdtf.c: Likewise.
    • sysdeps/soft-dfp/extenddfdd.c: Likewise.
    • sysdeps/soft-dfp/trunctfsd.c: Likewise.
    • sysdeps/soft-dfp/trunctdsf.c: Likewise.
    • sysdeps/soft-dfp/extenddftd.c: Likewise.
    • sysdeps/soft-dfp/extendddtf.c: Likewise.
    • sysdeps/soft-dfp/extendsfsd.c: Likewise.
    • sysdeps/soft-dfp/extendtftd.c: Likewise.
    • sysdeps/soft-dfp/truncdddf.c: Likewise.
    • sysdeps/soft-dfp/trunctfdd.c: Likewise.
    • sysdeps/soft-dfp/trunctddf.c: Likewise.
    • sysdeps/soft-dfp/truncsdsf.c: Likewise.
    • sysdeps/soft-dfp/trunctdtf.c: Likewise.
    • sysdeps/soft-dfp/extendsfdd.c: Likewise.
    • sysdeps/soft-dfp/extendsddf.c: Likewise.
    • sysdeps/soft-dfp/truncddsf.c: Likewise.
    • sysdeps/soft-dfp/truncdfsd.c: Likewise.

  • Miscellaneous changes

    • Makefile.in (libdfp_tests): Added test-bfp-conversions.
    • base-math/Makefile (libdfp_files): Moved extendsfsd extendsddf extendsdtf extendsfdd extenddfdd extendddtf extendsftd extenddftd extendtftd truncsdsf truncdfsd trunctfsd truncddsf truncdddf trunctfdd trunctdsf trunctddf trunctdtf from sysdeps/soft-dfp/Makefile since they're required in all builds.
    • tests/test-bfp-conversions.c: New file to test bfp -> dfp and dfp -> bfp conversions.
    • include/convert.h: Moved from sysdeps/soft-dfp/ in preparation for moving extend and trunc bfp->dfp and dfp->bfp conversion functions.
    • sysdeps/soft-dfp/Makefile (libdfp_files): Removed extendsfsd extendsddf extendsdtf extendsfdd extenddfdd extendddtf extendsftd extenddftd extendtftd truncsdsf truncdfsd trunctfsd truncddsf truncdddf trunctfdd trunctdsf trunctddf trunctdtf and moved to base-math/Makefile since they're required in all builds.
    • sysdeps/soft-dfp/extendsfsd.c: Replaced magic number '24' with __FLT_MANT_DIG__.
    • sysdeps/soft-dfp/convert.h: Moved to include/convert.h.
    • sysdeps/soft-dfp/truncdfsd.c: Replaced magic number '53' with __DBL_MANT_DIG__.

  • Miscellaneous changes

    • tests/gdb_start_address.py: New file which provides the start_address() python convenience function to GDB.
    • Makefile.gdb (*.gdb): Break on _dl_start_user for powerpc64 and _dl_main_dispatch on powerpc32. Call new python convenience function start_address() to get the start address of libc and libdfp for the scripts.

  • Miscellaneous changes

    • sysdeps/powerpc/powerpc32/power6/fpu/Makefile (ASFLAGS): Remove HAVE_ASM_PPC_REL16 as it is no longer needed since we no longer support building the power[6|7] optimized libdfp with -mbss-plt.
    • sysdeps/powerpc/powerpc32/power6/fpu/floatunsdisd.S: Remove HAVE_ASM_PPC_REL16 guard and remove the bss-plt form of GOT access. We no longer support building the power[6|7] optimized libdfp with -mbss-plt.
    • sysdeps/powerpc/powerpc32/power6/fpu/fixunssddi.S: Likewise.
    • sysdeps/powerpc/powerpc32/power6/fpu/fixunsdddi.S: Likewise.
    • sysdeps/powerpc/powerpc32/power6/fpu/fixunstddi.S: Likewise.
    • sysdeps/powerpc/powerpc32/power6/fpu/floatunsdidd.S: Likewise.
    • sysdeps/powerpc/powerpc32/power6/fpu/floatunsditd.S: Likewise.

  • Miscellaneous changes

    • configure: Regenerated.

  • Miscellaneous changes

    • Makefile.in: Added $(CFLAGS) to .S.o: rules.
    • configure.ac: Added -DSHARED to picflag in order to enforce PIC when compiling hand rolled asm functions.
    • sysdeps/powerpc/powerpc32/power6/fpu/Makefile: Added -DHAVE_ASM_PPC_REL16 to ASFLAGS in order to enforce secure-plt style GOT access.

  • Miscellaneous changes

    • sysdeps/s390/dfpu/fe_decround.c (__init_printf_dfp_getround): Added function definition for callback and added callback pointer. The printf_dfp () function was graciously skipping over the rounding mode modifiers due to this missing definition.
    • README.developer: Updated compatibility matrix to put x86 and x86_64 under TODO. Changed all instances of at05 to at4.0.

  • Miscellaneous changes

    • fe_decround.c (__init_printf_dfp_getround): Remove unnecessary prototype and declare constructor 'static. Remove unnecessary hidden_def(__init_printf_dfp_getround).
    • sysdeps/powerpc/dfpu/fe_decround.c (__init_printf_dfp_getround): Remove unnecessary prototype and declare constructor 'static. Remove unnecessary hidden_def(__init_printf_dfp_getround).
    • sysdeps/s390/dfpu/fe_decround.c: Added <dfpfenv_private.h>
    • printf_dfp.c: Add hidden_def (__register_printf_dfp).
    • include/libdfp-symbols.h (hidden_def): Fix macro definition.

  • Miscellaneous changes

    • printf_dfp.c (__printf_dfp): Finished cleaning up 'decimal'.
    • sysdeps/powerpc/sysdep.h: Updated copyright year.

  • Miscellaneous changes

    • README.maintainer: Updated.
    • sysdeps/powerpc/powerpc32/power6/fpu/fixunstdsi.S (END): Fix erroneous __dpd_fixunstssi to __dpd_fixunstdsi.
    • sysdeps/powerpc/powerpc32/sysdep.h (PROF): Removed PROF macros. (CALL_MCOUNT): Removed CALL_MCOUNT. (ENTRY): Removed CALL_MCOUNT and C_SYMBOL_NAME from macro body. Replaced C_LABEL with LABEL. (C_TEXT): Removed macro as unnecessary.
    • sysdeps/powerpc/sysdep.h (C_LABEL): Removed macro.
    • include/libdfp-symbols.h (LABEL): New macro which replaces C_LABEL by removing preceding underscore.
    • printf_dfp.c (__dfp_ais): Return -1 if info->user doesn't match any of the DFP mod_[H|D|DD] identifiers. (__printf_dfp): Started cleaning up 'decimal' per Jakub Jelinek's suggestion. Fixed bug found by Jakub Jelinek where random garbage is printed if the modifier is not H/D/DD. The default was dropping into the _Decimal32 case when it should have been returning -2.

  • Miscellaneous changes

    • README.maintainer: Added todo for "Commit Access"

  • Miscellaneous changes

    • README.maintainer: New file.
    • README.developer: Added several TODO items.
    • TODO: Added an item.
    • CONTRIBUTORS: Moved DCO from here to...
    • CONTRIBUTING: here. 2011-1-21 Ryan S. Arnold [email protected]
    • configure: Regenerated.
    • configure.ac: Incremented the version number to 1.0.7.
    • Versions.def: Incremented the version number to 1.0.7. 2011-1-21 Andreas Krebbel [email protected]
    • sysdeps/s390/dfpu/numdigits.h (setexp): Bias the exponent and return '1' for z zero mantissa, which is what is done for soft-dfp.
    • sysdeps/s390/dfpu/fe_decround.c (__fe_dec_setround): Enable extended rounding modes (5,6 and 7) which are not covered by the standard. 2011-1-18 Ryan S. Arnold [email protected]
    • configure: Regenerated.
    • tests/test-isfinite.c: Changed expectation on isfinite() for negative numbers on Power5 to '1'. Power6 (hardware isfinite) returns '-1' on isfinite. The spec simply says that non-zero means finite.
    • tests/test-logd.c: Removed half completed 'quantize compare' functionality which was throwing a false error.
    • tests/test-decode.c: Removed dangling printf that wasn't useful.
    • tests/test-printf.c: Added a _Decimal128 test for printing (1.0DL / 1.000000e-123DL) which has a sufficiently negative exponent that it forced the 'index' (rounding digit) position to be negative and caused a segv when __printf_dfp() tried to erroneously round the result on s390. We got lucky up until now on PowerPC due to the kernel reserving space on the top of the stack.
    • configure.ac: Incremented the version number to 1.0.6.
    • Versions.def: Incremented the version number to 1.0.6. 2011-1-18 Andreas Krebbel [email protected]
    • printf_dfp.c (__printf_dfp): Fixed segv where, if the exponent is sufficiently negative it pushes 'index' (the digit used to round values to the left) into a negative number. In that case we simply don't need to round at all. 2011-1-18 Ryan S. Arnold [email protected]
    • tests/test-strtod.c: Adjusted assumptions of tests from truncation on too long of an input to rounding to nearest. 2011-1-18 Andreas Krebbel [email protected]
    • strtod32.c (FUNCTION_L_INTERNAL): Check an extra digit on the input to see if we need to round rather than truncate. 2011-1-17 Ryan S. Arnold [email protected]
    • configure: Regenerated.
    • tests/test-strtod.c: Removed unnecessary test cases.
    • configure.ac: Incremented the version number to 1.0.5.
    • Versions.def: Incremented the version number to 1.0.5.

  • Miscellaneous changes

    • Makefile.in (libdfp_tests): Added test-log10d.
    • tests/test-decode.c: Added decode of log10d128(0.0000000011DL) to verify that it prints correctly in declets.
    • tests/test-logd.c (main): Added pbuf[CHAR_MAX] so that tests work properly.
    • tests/test-log10d.c: New test-case which verifies that printf now works properly for the result of log10d128(0.0000000011DL).
    • sysdeps/dpd/dpd-private.c: Fixed dpd_to_char[0x3f4] from "974" to "774" since the previous was a typo which was making values print incorrectly.

  • Miscellaneous changes

    • configure: Regenerated.
    • configure.ac: Removed addition of -fPIC to CFLAGS and instead export it to Makefile.in who is now responsible for adding it to CFLAGS. Set (picflags): Default to -fpic since the GOT for libdfp should be small enough that -fPIC isn't needed.
    • Makefile.in: (BACKEND_CFLAGS): New variable which passes -mcpu=foo and -fpic to the backend sub make invocation in -DEFS to get around the fact that libdecnumber/Makefile.in won't allow overridden CFLAGS.
    • sysdeps/dpd/dpd-private.c: Fixed 'used when uninitialized' warning by switching "else if" case to "else" for Decimal128.

  • Miscellaneous changes

    • strtod32.c: Fix precedence of arithmetic operation.
    • tests/test-strtod.c: Add new testcase and fix wrong tests.

  • Miscellaneous changes

    • configure: Regenerated.
    • configure.ac: Incremented the version number to 1.0.4
    • README.user: Marked empty sections with (TODO).
    • Versions.def: Incremented the version number to 1.0.4

  • Miscellaneous changes

    • README.user: Added clarification of how to include dfp/stdlib.h, dfp/fenv.h and dfp/wchar.h.

  • Miscellaneous changes

    • README.user: Added clarification of how to include dfp/math.h.

  • Miscellaneous changes

    • dfp/math.h: Move #include_next <math.h> so that if the compiler invocation has -I//include/dfp/ but doesn't define __STDC_WANT_DEC_FP__ that they still get the system math.h.

  • Miscellaneous changes

    • CONTRIBUTORS: Added Luis Machado. Added DCO requirements.

  • Miscellaneous changes

    • strtod32.c (FUNCTION_L_INTERNAL): Fix comparison operator.

  • Miscellaneous changes

    • sysdeps/s390/dfpu/numdigits.h: Remove include guards.
    • sysdeps/powerpc/dfpu/numdigits.h: Likewise.
    • sysdeps/soft-dfp/dpd/numdigits.h: Likewise.

  • Miscellaneous changes

    • tests/test-printf.c (printf_d128s): Added 0.9999999999DL which reported 0.100000 rather than 1.000000 and was fixed by Andreas' patch from 2010-10-04.

  • Miscellaneous changes

    • printf_dfp.c (__printf_dfp): Fix up nd when rounding overflows beyond first digit.

  • Miscellaneous changes

    • sysdeps/soft-dfp/floatsisd.c: Don't generate additional decimal digit.

  • Miscellaneous changes

    • README.developer: Documentation update.
    • TODO: Mark some items as [DONE].

  • Miscellaneous changes

    • init_dfp.c: Changed #include <dfp/dfp.h> to #include <dfp.h>.

  • Miscellaneous changes

    • tests/test-amort.c: Added license and copyright.
    • tests/test-printf.c: Updated copyright year.
    • tests/test-strtod.c: Added license and copyright. Corrected incorrect expected values.
    • strtod32.c: Fixed several conversion bugs.

  • Miscellaneous changes

    • Makefile.in: Removed test-left_justify from execution since it is a known failure case for non-power6 systems. No worries since left_justify isn't used right now.
    • tests/test-amort.c: Removed spurious #include <dfp.h>.
    • tests/test-get_digits.c: Likewise.
    • tests/test-decode.c: Likewise.
    • tests/test-quantize.c: Likewise.
    • tests/test-printf.c: Likewise.
    • tests/test-fpclassify.c: Likewise.
    • tests/test-numdigits.c: Likewise.
    • tests/test-left_justify.c: Likewise.
    • tests/test-isinf.c: Likewise.
    • tests/test-isnan.c: Likewise.
    • tests/test-isfinite.c: Likewise.
    • tests/test-strtod.c: Likewise. Also changed one value compare test to a decode test since the visual representation was 0.0.
    • README.user: Added further user notes.
    • dfp/dfp.h: Moved this to ...
    • include/dfp.h: here since the user no longer has to call the printf_dfp registration hooks, and thus doens't need the prototype. configure: Regenerated. configure.ac: Added AC_SUBST(mzarch) back. We shouldn't be editing CFLAGS. Makefile.in: Pass $(mzarch) along with CFLAGS if it is defined.

  • Miscellaneous changes

    • configure: Regenerated.
    • Makefile.in: Added special sed script which changes libdecnumber/Makefile:CFLAGS = foo into CFLAGS ?= foo. This allows libdfp/Makefile's sub make invocation of libdecnumber/Makefile to accept the updated CFLAGS. This was necessary to allow configure to update CFLAGS with -mzarch for s390 z9-ec and z10. GCC should fix this in their version of libdecnumber.
    • configure.ac: -mzarch added to CFLAGS when s390 system is z9-ec or z10.
    • README.developer: Added indication that configure.ac sets -mzarch in CFLAGS when appropriate.

  • Miscellaneous changes

    • Makefile.gdb (libdfp_tests.gdb): Add LD_LIBRARY_PATH=./:$LD_LIBRARY_PATH to ldd invocations to make sure it reports ./libdfp.so.1 as a dependency and not the system or toolchain libdfp.so.1.

  • Miscellaneous changes

    • ieee754r/Versions: Added __fabsd* and __fpclassifyd* under the correct libdfp version number 1.0.3.

  • Miscellaneous changes

    • dfp/math.h: Restructure to cleanup presentation.

  • Miscellaneous changes

    • ieee754r/Versions: Moved some of the __is* functions under LIBDFP_1.0.2 and LIBDFP_1.0.3. Removed unnecessary __is* functions from export that don't have polymorphic versions in math.h
    • dfp/math.h: Removed accidental undef of isunordered. This was never implemented as a polymorphic function because it doesn't need to be (per the spec). Removed redundant polymoprhic isnan().
    • Versions.def: Added Versions LIBDFP_1.0.2 and LIBDFP_1.0.3 for Libdfp.

  • Miscellaneous changes

    • configure: Regenerated.
    • configure.ac: Apparently AC_CANONICAL_SYSTEM is only for compilers. I removed it, and support for --target. Now you must use --build only.
    • README.developer: Removed directions for using --target.

  • Miscellaneous changes

    • sysdeps/powerpc/dfpu/fe_decround.c (stdio.h): Removed unnecessary include.

  • Miscellaneous changes

    • sysdeps/powerpc/dfpu/fe_decround.c (__fe_dec_getround): The structure which defined the DRN was to small and the compiler started generating code that wasn't properly extracting the DRN field from the structure holding the FPSCR value. This is fixed.
    • tests/test-round.c: Testcase to prove fe_dec_getround() and fe_dec_setround().
    • Makefile.in (libdfp_tests): Added test-round.

  • Miscellaneous changes

    • configure: Regenerated.
    • config.log: Removed as unnecessary.
    • configure.ac: --target now works in lieu of --build. --host is no longer necessary.
    • README.developer: Removed directive to use --host and replaced with a diective to use --target.

  • Miscellaneous changes

    • configure: Regenerated file.
    • Makefile.in: Removed last comments dealing with -mzarch.
    • configure.ac: Removed last fragments dealing with -mzarch.

  • Miscellaneous changes

    • ieee754r/isinfd32.c: Removed unnecessary static char buffer used for testing.

  • Miscellaneous changes

    • ieee754r/isinfd32.c: Removed test-code that shouldn't have been checked in.

  • Miscellaneous changes

    • ieee754r/isinfd32.c (isinf): Changed Infinity check such that DEC_NAN doesn't pass as inf. There was an incorrect mask check which was true for DEC_NAN as well as Inf.
    • tests/test-isfinite.c: Added Conditional expectations for s390 for isfinite return values from negative number inputs.

  • Miscellaneous changes

    • tests/test-logd.c: Removed unnecessary comments.
    • tests/test-fpclassify.c: Removed unnecessary comments.
    • configure.ac: Removed unnecessary comments.
    • printf_dfp.c: Removed unnecessary comments.

  • Miscellaneous changes

    • Makefile.in: Removed commented out code.
    • tests/test-logd.c: Corrected expected values for a _Decimal64 NAN declet series.
    • tests/test-fpclassify.c: Added cases for + and - __DEC[32|64|128]_SUBNORMAL_MIN__.
    • tests/scaffold.c: Modified spacing in macro output for 'Success'.
    • tests/test-isfinite.c: Corrected expected values for negative numbers to return -1. Corrected expected value for DEC_NAN to '0'.
    • tests/test-strtod.c: Cleaned up. Still failing.
    • sysdeps/powerpc/dfpu/fpclassifyd32.c (fpclassifyd32): Since _Decimal32 is promoted to _Decimal64 and __DEC32_SUBNORMAL_MIN__ falls within the bounds of _Decimal64 a special case to range check for _Decimal32 as required.
    • sysdeps/powerpc/dfpu/powd64.c: Whitespace correction.
    • strtod32.c: Started to cleanup the code.

  • Miscellaneous changes

    • Makefile.in: Added test-numdigits, test-get_digits, and test-left_justify.
    • base-math/Makefile: Fixed missing truncd* exports which was exposed by GCC 4.4.
    • base-math/lttd2.c (lttd3): Renamed lttd3 to lttd2. Typo.
    • tests/test-get_digits.c: Added test-cases to show previously failing numdigits for soft-dfp when coefficient was zero. It was returning '0' intead of '1', which is what hardware DFP returns.
    • tests/test-printf.c: Added test-cases to show failing tests.
    • tests/test-fpclassify.c: (printf_d32s): Changed DENORMAL to SUBNORMAL in constant usage. GCC changed usage.
    • tests/test-numdigits.c: New test.
    • tests/scaffold.c: Updated output formatting of macros to make output clearer (columns and test numbering).
    • tests/test-left_justify.c: New test.
    • tests/test-strtod.c: Added test-cases to show failing tests.
    • sysdeps/powerpc/dfpu/numdigits.h (_NUMDIGITS_H): Removed guard to allow multiple #includes with different preconditions. (numdigits): New implementation which avoids clobbers and avoids modifying a parameter. (left_justify): New implementation which avoids clobbers and avoids modifyinig a parameter. (getexp): New implementation which avoids modifying a parameter.
    • sysdeps/powerpc/dfpu/fe_decround.c (__fe_dec_setround): Fixed errorneous usage of mtfsfi instruction. Previous usage resulted in the set rounding mode not being permanent.
    • sysdeps/soft-dfp/dpd/numdigits.h (numdigits): Return '1' if coefficient is zero, which is what hardware DFP does. (__dfp_declet_to_dpd): Moved to ...
    • sysdeps/dpd/dpd-private.h: ... this file.
    • sysdeps/dpd/dpd-private.c: Formatting.
    • printf_dfp.c (printf_dfp): Major fixes to correct a variety of printing errors. It should now fully support %f, %e, %g, and %a in the myriad of bizarre combinations.

  • Miscellaneous changes

    • Makefile.in: Added test-strtod testcase.
    • tests/test-logd.c: Cleaned up some comments.
    • tests/scaffold.c: Cleaned up _QC_P macro. It is incomplete at this point.
    • tests/test-strtod.c: New testcase.
    • sysdeps/powerpc/dfpu/isfinited128.c: Removed spurious <dfpmacro.h>.
    • sysdeps/powerpc/dfpu/isfinited32.c: Likewise.
    • sysdeps/powerpc/dfpu/isfinited64.c: Likewise.
    • sysdeps/powerpc/dfpu/is_template.h: Added <dfpmacro.h> and removed redundant macro definitions.
    • strtod32.c: Fixed bug when calling newlocale(LC_ALL_MASK,NULL,NULL); Changed it to newlocale(LC_ALL_MASK, setlocale (LC_ALL, NULL),NULL);

  • Miscellaneous changes

    • tests/test-logd.c: Added more expressive output on failure.
    • sysdeps/powerpc/dfpu/isfinited128.c: Added #include <dfpmacro.h>.
    • sysdeps/powerpc/dfpu/isfinited32.c: Likewise.
    • sysdeps/powerpc/dfpu/isfinited64.c: Likewise.

  • Miscellaneous changes

    • configure: Regenerated.
    • ieee754r/Versions: Re-added __isfinited* for backward compatability.
    • configure.ac: Increment library version to 1.0.3 due to interface changes.
    • README.user: Whitespace.
    • ieee754r/isfinited32.c: Use 'isfinite' instead of 'finite'.
    • sysdeps/powerpc/dfpu/isfinited128.c: Use 'isfinite' instead of
    • 'finite'.
    • sysdeps/powerpc/dfpu/isfinited32.c: Use 'isfinite' instead of
    • 'finite'.
    • sysdeps/powerpc/dfpu/isfinited64.c: Use 'isfinite' instead of
    • 'finite'.
    • dfp/math.h: Whitespace.

  • Miscellaneous changes

    • ieee754r/isfinited128.c: Change finite to isfinite.
    • ieee754r/isfinited32.c: Likewise.
    • ieee754r/isfinited64.c: Likewise.

  • Miscellaneous changes

    • Makefile.in: Added new test cases. Changed finited to isfinited.
    • ieee754r/scalbnd32.c: Changed usage of finited to isfinited.
    • ieee754r/scalblnd32.c: Likewise.
    • ieee754r/cbrtd32.c: Likewise.
    • ieee754r/expd32.c: Likewise.
    • ieee754r/nextafterd32.c: Likewise.
    • ieee754r/exp2d32.c: Likewise.
    • ieee754r/Makefile: Likewise.
    • ieee754r/expm1d32.c: Likewise.
    • ieee754r/hypotd32.c: Likewise.
    • ieee754r/powd32.c: Likewise.
    • ieee754r/ldexpd32.c: Likewise.
    • ieee754r/fdimd32.c: Likewise.
    • ieee754r/nearbyintd32.c: Likewise.
    • ieee754r/coshd32.c: Likewise.
    • ieee754r/lgammad32.c: Likewise.
    • ieee754r/sinhd32.c: Likewise.
    • ieee754r/tgammad32.c: Likewise.
    • ieee754r/nexttowardd32.c: Likewise.
    • ieee754r/Versions: Likewise.
    • ieee754r/tand32.c: Likewise.
    • ieee754r/frexpd32.c: Likewise.
    • ieee754r/isfinited64.c: Renamed to this from...
    • ieee754r/finited64.c: ...this.
    • ieee754r/isfinited32.c: Renamed to this from...
    • ieee754r/finited32.c: ...this.
    • ieee754r/isfinited128.c: Renamed to this from...
    • ieee754r/finited128.c: ...this.
    • tests/test-logd.c: New test.
    • tests/test-quantize.c: New test.
    • tests/test-fpclassify.c: New test.
    • tests/scaffold.c: Started adding _QC and QC_P.
    • tests/test-isfinite.c: Renamed this...
    • tests/test-finite.c: ...from this.
    • sysdeps/powerpc/dfpu/logd64.c: Added hard-dfp optimizations.
    • sysdeps/powerpc/dfpu/isinfd128.c: Likewise.
    • sysdeps/powerpc/dfpu/isfinited128.c: Likewise.
    • sysdeps/powerpc/dfpu/expd64.c: Likewise.
    • sysdeps/powerpc/dfpu/rintd128.c: Likewise.
    • sysdeps/powerpc/dfpu/isfinited32.c: Likewise.
    • sysdeps/powerpc/dfpu/powd64.c: Likewise.
    • sysdeps/powerpc/dfpu/rintd32.c: Likewise.
    • sysdeps/powerpc/dfpu/finited32.c: Likewise.
    • sysdeps/powerpc/dfpu/logd32.c: Likewise.
    • sysdeps/powerpc/dfpu/isnormald64.c: Likewise.
    • sysdeps/powerpc/dfpu/fpclassifyd128.c: Likewise.
    • sysdeps/powerpc/dfpu/expd32.c: Likewise.
    • sysdeps/powerpc/dfpu/isnand128.c: Likewise.
    • sysdeps/powerpc/dfpu/is_template.h: Likewise.
    • sysdeps/powerpc/dfpu/powd32.c: Likewise.
    • sysdeps/powerpc/dfpu/ddlogtbls.h: Likewise.
    • sysdeps/powerpc/dfpu/isnormald128.c: Likewise.
    • sysdeps/powerpc/dfpu/isfinited64.c: Likewise.
    • sysdeps/powerpc/dfpu/rintd64.c: Likewise.
    • sysdeps/powerpc/dfpu/finited64.c: Likewise.
    • sysdeps/powerpc/dfpu/isnormald32.c: Likewise.
    • dfp/math.h: added islessgreaterd prototypes, etc. Added isnormald prototypes.

  • Miscellaneous changes

    • configure: Updated version to 1.0.2.
    • Makefile.in: Added test cases.
    • ieee754r/Versions: Exported __isnand*, __isinfd*, etc.
    • tests/test-printf.c: Define _WANT_PC.
    • tests/test-decode.c: Define _WANT_DC.
    • tests/scaffold.c: Added _VC_P and _VC as well as _WANT_PC, _WANT_DC, and _WANT_VC.
    • tests/test-fpclassify.c: New classification function test.
    • tests/test-isnan.c: Likewise.
    • tests/test-isinf.c: Likewise.
    • tests/test-finite.c: Likewise.
    • tests/test-param.c: Likewise.
    • tests/test-quantize.c: Likewise.
    • sysdeps/powerpc/dfpu/quantized64.c: Hard-dfp implementation.
    • sysdeps/powerpc/dfpu/quantized128.c: Likewise.
    • sysdeps/powerpc/dfpu/fpclassifyd64.c: Likewise.
    • sysdeps/powerpc/dfpu/isnand32.c: Likewise.
    • sysdeps/powerpc/dfpu/isinfd32.c: Likewise.
    • sysdeps/powerpc/dfpu/isnand64.c: Likewise.
    • sysdeps/powerpc/dfpu/isinfd64.c: Likewise.
    • sysdeps/powerpc/dfpu/finited32.c: Likewise.
    • sysdeps/powerpc/dfpu/quantized32.c: Likewise.
    • sysdeps/powerpc/dfpu/finited64.c: Likewise.
    • sysdeps/powerpc/dfpu/fpclassifyd32.c: Likewise.

  • Miscellaneous changes

    • Makefile.in: Allow make check to work without --with-glibc-build.
    • tests/test-printf.c: Removed register_printf_dfp() invocation.
    • tests/test-amort.c: New file which is an amortiation benchmark.
    • tests/scaffold.c: Changed <decode.h> to "decode.h".
    • tests/decode.h (_DFP_DECODE_H): Added _DFP onto guard.
    • tests/debug-test.sh: Allow make check to work without --with-glibc-build by either debugging the app or the loader.
    • configure.ac: Incrememented the library version from 1.0.1 to 1.0.2 due to the addition of the libdfp constructor which no longer requires register_printf_dfp().
    • README.user: More todos.
    • include/init_dfp.h (__libdfp_init): New file to add constructor prototype.
    • Makefile.gdb: Allow make check to work without --with-glibc-build.
    • init_dfp.c (__libdfp_init): New file. Add a constructor to call register_printf_dfp().

  • Miscellaneous changes

    • configure: Regenerated.
    • Makefile.in: Remove appending of cc_msize to CFLAGS. Remove -e from submake which builds libdecnumber.
    • configure.ac: Remove cc_msize and cc_mzarch and accompanying AC_SUBST. The preferred method is to require the user to pass -m[31|32|64] in CFLAGS as an env variable when configure is invoked.
    • README.user: New file.
    • README.developer: Updated to direct user to pass -m[31|32|64] in CFLAGS and to always use --build and --host values that align with the -m values.

  • Miscellaneous changes

    • Makefile.in: Removed unneccesary rule -- backend: $(dfp_backend)/$(dfp_backend_lib).

  • Miscellaneous changes

    • Makefile.in: Fixed rule $(top_builddir)/$(dfp_backend_lib): to $(dfp_backend)/$(dfp_backend_lib) to correct problem with make -j*

  • Miscellaneous changes

    • Makefile.in: Replaced libdfp.so.1 generation using ln -s' to using ldconfig -l' in both $(top_builddir)/ and $(install_root)$(libdir)/. Added creation of linkname symlink libdfp.so. Strip trailing / from any install_root= paths.

  • Miscellaneous changes

    • Makefile.in: Fixed `ln -s' to not have the goofy '(cd..)', e.g. (cd $(libdir) ln -s ) when making the .so.1 -> -1.0.1.so symlinks.

  • Miscellaneous changes

    • Makefile.in: Fixed libdir=@prefix@/lib to libdir=@libdir@ and includedir=@prefix@/include to includedir=@includedir@. This will allow configure override with --libdir= and --includedir= respectively. This was preventing installation into //lib64. Fix header install so that dfp/README isn't installed with the header files.

  • Miscellaneous changes

    • Makefile.in: `install' doesn't handle symlinks so we've done another 'ln -s' in the install stage to set $(SHARED_LIBRARY) -&gt; ./$(SHARED_VERSIONED_LIBRARY) in the $(install_root)/$(libdir)/ directory.

  • Miscellaneous changes

    • Makefile.in: Fixed erroneous attempt to install libdfp.a even if --enable-static=no was configured.

  • Miscellaneous changes

    • Makefile.in: Fixed missing libdecnumber.a rule dependency by changing the backend rule to be libdecnumber.a' rather than libdecnumber'. Fixed ln -s' error when rebuilding without a make clean by adding -f' to always force overwrite the previous versioned library symlink.
    • README: Renamed to README.developer
    • README.developer: New file.

  • Miscellaneous changes

    • configure: Regenerated.
    • configure.ac: Added additional help text when warning about the glibc headers lacking the printf-hooks support.
    • README: Clarified version number.
    • TODO: Updated.

  • Miscellaneous changes

    • configure: Regenerated.
    • configure.ac: Add check for glibc headers that support the printf-hooks. Add runtime check for printf-hooks in libc specified by --with-glibc-build.
    • TODO: Updated.

  • Miscellaneous changes

    • configure: Regenerated.
    • configure.ac: Updated libdfp to version 1.0.1 to indicate that power7 support has been added.

  • Miscellaneous changes

    • configure.ac: Added POWER 7 support.
    • README: Added notes on POWER 7 support.
    • sysdeps/powerpc/powerpc32/power7: New directory.
    • sysdeps/powerpc/powerpc32/power7/Implies: Imply Power6.
    • sysdeps/powerpc/powerpc32/power7/fpu: New directory.
    • sysdeps/powerpc/powerpc32/power7/fpu/Implies: Imply Power6.
    • sysdeps/powerpc/powerpc64/power7: New directory.
    • sysdeps/powerpc/powerpc64/power7/Implies: Imply Power6.
    • sysdeps/powerpc/powerpc64/power7/fpu: New directory.
    • sysdeps/powerpc/powerpc64/power7/fpu/Implies: Imply Power6.

  • Miscellaneous changes

    • configure: Regenerated.
    • configure.ac (AC_CACHE_CHECK): Re-added check for decimal-float-support in the compiler by checking whether the compiler can handle the _Decimal64 data type rather than checking whether the compiler -v output had a particular flag set.

  • Miscellaneous changes

    • configure: Regenerated.
    • configure.ac (AC_PROG_CC): Re-added AC_PROG_CC and removed custom $(CC). Removed -m flags from CC and pushed this responsibility onto Makefile.in to put into CFLAGS. These changes allow a user to prepend some custom flags onto CFLAGS at configure time. (dfp_opts): Changed to $(cc_mzarch). No need to be obscure until we have to.
    • Makefile.in (cc_mzarch): New variable used on s390 only, otherwise empty. (default_cflags): Inherit defaults from configure. Add -m$(cc_msize) and $(cc_mzarch). (default_asflags): Inherit defaults from configure. Add -m$(cc_msize) and $(cc_mzarch). (+cflags): Simplified CFLAGS override mechanism. (+asflags): Simplified ASFLAGS override mechanism. ($(top_builddir)/$(dfp_backend)): Manually pass overridden CFLAGS to backend 'make' since the -m flags aren't passed in the top level configure's CFLAGS. ($(top_builddir)/$(SHARED_VERSION_LIBRARY)): Added $(CFLAGS) since it's no longer part of $(CC). ($(libdfp_tests)): Added $(CFLAGS) since it's no longer part of $(CC).

  • Miscellaneous changes

    • configure: Regenerated.

  • Miscellaneous changes

    • configure.ac: Add -mzarch when building hw dfp on S/390. This is needed for 31 bit which would otherwise default to -mesa.

  • Miscellaneous changes

    • configure: Regenerated.
    • Makefile.in (.c.o, .c.os): Added -include $(top_builddir)/config.h to pick up _BACKEND_DPD or _BACKEND_BID. ($(top_builddir)/$(STATIC_LIBRARY)): Add $(top_builddir) to dfp_backend path.
    • README: Added WARNING about making sure to link to a glibc that's the same bitness (32|64) and the same optimization as the libdfp that's being configure. Added --with-glibc-build= examples.
    • Versions (libdfp): LIBDFP_1.0.0 tagged.
    • Versions.def (libdfp): LIBDFP_1.0.0 tagged.
    • ieee754r/Versions.def (libdfp): LIBDFP_1.0.0 tagged.
    • config.h.in: Regenerated.
    • configure.ac: (AC_INIT): Tagged version 1.0.0. (_DPD_BACKEND): AC_DEFINE_UNQUOTED; Set based on backend. (_BID_BACKEND): AC_DEFINE_UNQUOTED; Set based on backend. (--with-cpu): Removed submachine_opt="-mcpu=$withval". ($machine): On s390 cc_msize=31 NOT 32. (AC_SUBST): CFLAGS and ASFLAGS. (AC_CACHE_CHECK): Removed check for compiler feature decimal-float-support.
    • ieee754r/frexpd32.c (INTERNAL_FUNCTION_NAME): Initialize 'result' to DEC_NAN because compiler can't see that it's set in a macro and generates warnings.
    • sysdeps/bid/Versions (libdfp): LIBDFP_1.0.0 tagged
    • sysdeps/dpd/Versions (libdfp): LIBDFP_1.0.0 tagged
    • sysdeps/dpd/dpd-private.c (__get_digits_d32, __get_digits_d64, __get_digits_d128): initialize exp = 0;
    • sysdeps/powerpc/dfpu/numdigits.h (numdigits): Conditionally define 'f' based on _Decimal size to prevent the compiler from messing up the register order requirements.
    • sysdeps/soft-dfp/Versions (libdfp): LIBDFP_1.0.0 tagged
    • sysdeps/soft-dfp/bid/Versions (libdfp): LIBDFP_1.0.0 tagged
    • sysdeps/soft-dfp/dpd/Versions (libdfp): LIBDFP_1.0.0 tagged
    • tests/decode.h: New file providing non-exported decoded[32|64|128] prototypes used in tests and debugging.
    • tests/scaffold.c (_DC, _DC_P): Added macros for testing against decoded[32|64|128] output.
    • tests/test-param.c (main): Updated to use scaffold.c and test the values against the decoded[32|64|128] values that're expected.

  • Miscellaneous changes

    • ieee754r/powd32.c (INTERNAL_FUNCTION_NAME): Set __isnan -> FUNC_D(__isnan). Not strictly necessary since __isnan is a polymorphic classification function, but this was we avoid the extra steps of detecting the type/size of the input parameter and call the appropriate __isnand[32|64|128] directly.

  • Miscellaneous changes

    • ieee754r/powd32.c: Add dfp prefix to __isnan invocation.

  • Miscellaneous changes

    • sysdeps/s390/dfpu/fenv_libdfp.h: Move file to ...
    • sysdeps/s390/fpu/fenv_libdfp.h: ... here.

  • Miscellaneous changes

    • configure.ac: Add a march vs mcpu gcc option check.

  • Miscellaneous changes

    • Makefile.in (prefix): Set based on @prefix@ to prevent missing --exec-prefix from setting @exec_prefix@ to the string '$(prefix)', i.e. not the context of $prefix. ($(top_builddir)/$(SHARED_VERSION_LIBRARY)): Added -W,no-whole-archive to close -W,whole-archive to SHARED_VERSION_LIBRARY rule. Removed -nodefaultlibs -nostdlib which prevented linking against libc and libm which libdfp relies upon.
    • tests/test-printf.c (printf_d128s): Added correct "expected" strings.
    • sysdeps/soft-dfp/dpd/numdigits.h (left_justify): Changed erroneous FUNC_D (getexp(x)) to FUNC_D (getexp) (x) to correct macro invocation.

  • Miscellaneous changes

    • README: Updated the S/390 availability section.

  • Miscellaneous changes

    • Makefile.in: Updated comment to reflect update from 0.0.1 to 0.0.9.
    • configure.ac (AC_INIT): Updated version from 0.0.1 to 0.0.9.
    • README: Added two TODO chapters and clarified the intro paragraphs.
    • Versions.def: Update package versions for symbols.
    • Versions: Update package versions for symbols.

  • Miscellaneous changes

    • Makefile.in: Added test-param to libdfp_tests.
    • tests/TODO: New file.
    • tests/README: N/A.
    • tests/test-param.c: New test to verify integrity of parameters that are spilled to the stack on being passed to a new function.

  • Miscellaneous changes

    • tests/test-printf.c: Added mechanism to easily add new printf test cases based upon GLIBC's stdio-common/tfformat.c testcase.

  • Miscellaneous changes

    • tests/scaffold.c: New file which holds _C, _C_P, _PC, and _PC_P macros used for printing debugging messages in the event of test case failures.

  • Miscellaneous changes

    • Makefile.in: For .out rules send stdout to /dev/null and stderr to the .out file.
    • tests/test-printf.c: Added framework for generically adding printf tests.
    • Makefile.gdb: Squelch echos of .conf and .gdb file creation.

  • Miscellaneous changes

    • Makefile.in: Change test-debug to > debug-test.
    • tests/debug-test.sh: Change test-debug.conf -> debug-test.conf.
    • Makefile.gdb: Change test-debug to debug-test.

  • Miscellaneous changes

    • tests/test-debug.sh: Removed and renamed to ...
    • tests/debug-test.sh: New.

  • Miscellaneous changes

    • CONTRIBUTORS: Added Carlos Eduardo Seo for contribution of Makefile.in make install and make install-headers feature.
    • README: Added section on Make Rules.
    • TODO: Added additional todo items.

  • Miscellaneous changes

    • Makefile.in: Added 'make install0headers' feature.

  • Miscellaneous changes

    • printf_dfp.c: (padn, wpadn): Changed usage of size_t to 'int' to avoid compiler warnings about unsigned to signed comparisons.

  • Miscellaneous changes

    • tests/test-printf.c: Added print statements for _Decimal32 and _Decimal128.
    • sysdeps/powerpc/dfpu/numdigits.h (numdigits getexp): Use FUNC_D wrapper. (FUNC_D PASTE): Conditionally define these macros.

  • Miscellaneous changes

    • include/printf_dfp.h (__d32_ais, __d64_ais, __d128_ais): Removed
    • printf_dfp.c (__printf_dfp): Changed width from unsigned into to int in order to prevent negative default. (__register_printf_dfp): Replaced individual decimal sized specifier registrations with a unified registrations. (__d32_ais, __d64_ais, __d128_ais): Removed and replaced with unified __dfp_ais().

  • Miscellaneous changes

    • ieee754r/ldexpd32.c (ldexpd32): Use FUNC_D wrapper around getexp.
    • ieee754r/frexpd32.c (frexpd32): Use FUNC_D wrapper around numdigits and getexp.
    • sysdeps/soft-dfp/dpd/numdigits.h (numdigits getexp): Use FUNC_D wrapper.

  • Miscellaneous changes

    • Makefile.in: Added 'make install' feature.

  • Miscellaneous changes

    • include/printf_dfp.h: Add #include <stdardg.h>.
    • sysdeps/s390/dfpu/numdigits.h: Wrap function definitions in FUNC_D macro. Add PASTE and FUNC_D macro definitions.
    • strtod32.c (setexp): Refer to 'getexp' using FUNC_D() macro. Add PASTE and FUNC_D macro definitions. Reposition dfp*_private.h header inclusions.

  • General non-functiopnal maintenance.

    • Makefile.in: Cleanup comments.
    • README: Fix typo, spacings, and text.

  • This commit adds the port of libdfp from a GLIBC add-on to a standalone library. This library was written by Ryan S. Arnold [email protected], Joseph Kerian [email protected], Pete Eberlein [email protected], Steve Munroe [email protected], Andreas Krebbel [email protected], et al.

  • This commit adds the top-level sub-directories for the stand-alone libdfp project. It adds the top-level directory libdfp/ per Mark Mitchell's direction on [email protected] on Mon, 29 Jun 2009.