mirror of
https://github.com/HDFGroup/hdf5.git
synced 2024-11-27 02:10:55 +08:00
081ac87fed
Bug fix Description: This checkin addresses two bugs. Firstly, I used the wrong semantics when detecting if $MAKE is defined to the empty string or undefined. Secondly, pmake on IRIX does not work well with automake's dependency tracking if the build is not an in-place build. Solution: Fixed semantics in configure.in. Disabled automatic dependency tracking automatically on IRIX. This is not the optimal fix, but it should work fine as long as no one is doing HDF5 development on IRIX, and it is much easier than trying to detect the exact situations in which dependency tracking fails (or than trying to fix dependency tracking!). Platforms tested: heping, modi4 (pmake and gmake)
81 lines
2.8 KiB
Bash
81 lines
2.8 KiB
Bash
# -*- shell-script -*-
|
|
#
|
|
# This file is part of the HDF5 build script. It is processed shortly
|
|
# after configure starts and defines, among other things, flags for
|
|
# the various compile modes.
|
|
#
|
|
# See BlankForm in this directory for details
|
|
|
|
# Disable dependency tracking on IRIX unless the user specifically asks for
|
|
# it.
|
|
# IRIX's pmake confuses automake (as of version 1.9) if dependency tracking
|
|
# is enabled and it is not an in-place build. Simply disabling dependency
|
|
# tracking on IRIX is simpler to implement than detecting pmake, detecting
|
|
# when a build is not in-place, and then disabling dependency tracking.
|
|
if test -z "${enable_dependency_tracking}"; then
|
|
enable_dependency_tracking="no"
|
|
fi
|
|
|
|
# The default compiler is `cc' and there is no ranlib.
|
|
if test "X-" = "X-$CC"; then
|
|
CC=cc
|
|
CC_BASENAME=cc
|
|
fi
|
|
RANLIB=:
|
|
|
|
case "X-$CC_BASENAME" in
|
|
X-gcc)
|
|
CFLAGS="$CFLAGS -Wsign-compare" #Only works for some versions
|
|
DEBUG_CFLAGS="-g -fverbose-asm"
|
|
DEBUG_CPPFLAGS=
|
|
PROD_CFLAGS="-O3"
|
|
PROD_CPPFLAGS=
|
|
PROFILE_CFLAGS="-pg"
|
|
PROFILE_CPPFLAGS=
|
|
;;
|
|
|
|
*)
|
|
# Do *not* use -ansi because it prevents hdf5 from being able
|
|
# to read modification dates from the file. On some systems it
|
|
# can also result in compile errors in system header files
|
|
# since hdf5 includes a couple non-ANSI header files.
|
|
#CFLAGS="$CFLAGS -ansi"
|
|
|
|
# Always turn off these compiler warnings:
|
|
CFLAGS="$CFLAGS -woff 799"
|
|
|
|
# Extra debugging flags
|
|
DEBUG_CFLAGS=-g
|
|
DEBUG_CPPFLAGS=
|
|
|
|
# Extra production flags
|
|
# Note: higher optimizations relax alignment requirements needed.
|
|
PROD_CFLAGS="-O -s"
|
|
PROD_CPPFLAGS=
|
|
|
|
# Extra profiling flags
|
|
PROFILE_CFLAGS=-pg
|
|
PROFILE_CPPFLAGS=
|
|
;;
|
|
esac
|
|
|
|
# Hard set flag to indicate that the 'unsigned long long' to floating-point
|
|
# value conversion are broken by the compilers (as of 4/27/04 - QAK)
|
|
hdf5_cv_sw_ulong_to_fp_bottom_bit_works=${hdf5_cv_sw_ulong_to_fp_bottom_bit_works='no'}
|
|
|
|
# Set flags to avoid conversion between 'long double' and integers because of
|
|
# SGI's compiler problems. For both IRIX64 6.5 and IRIX 6.5, the compilers
|
|
# have the following problems,
|
|
# long double -> signed char : incorrect rounding
|
|
# long double -> unsigned char : incorrect rounding
|
|
# long double -> short : incorrect rounding
|
|
# long double -> unsigned short : incorrect rounding
|
|
# long double -> long or long long: incorrect value
|
|
# long double -> unsigned long or long long : incorrect value
|
|
#
|
|
# long or long long -> long double : correct value but incorrect bit pattern
|
|
# unsigned long or long long -> long double : correct value but incorrect bit pattern
|
|
# (1/5/05 - SLU)
|
|
hdf5_cv_sw_ldouble_to_int_works=${hdf5_cv_sw_ldouble_to_int_works='no'}
|
|
hdf5_cv_sw_integer_to_ldouble_works=${hdf5_cv_sw_integer_to_ldouble_works='no'}
|