head	1.1;
branch	1.1.1;
access;
symbols
	netbsd-10-2-RELEASE:1.1.1.1
	netbsd-11-0-RELEASE:1.1.1.2
	netbsd-11-0-RC7:1.1.1.2
	netbsd-11-0-RC6:1.1.1.2
	netbsd-11-0-RC5:1.1.1.2
	netbsd-11-0-RC4:1.1.1.2
	netbsd-11-0-RC3:1.1.1.2
	netbsd-11-0-RC2:1.1.1.2
	netbsd-11-0-RC1:1.1.1.2
	netbsd-11:1.1.1.2.0.2
	netbsd-11-base:1.1.1.2
	netbsd-10-1-RELEASE:1.1.1.1
	netbsd-10-0-RELEASE:1.1.1.1
	netbsd-10-0-RC6:1.1.1.1
	netbsd-10-0-RC5:1.1.1.1
	netbsd-10-0-RC4:1.1.1.1
	netbsd-10-0-RC3:1.1.1.1
	netbsd-10-0-RC2:1.1.1.1
	netbsd-10-0-RC1:1.1.1.1
	ctwm-4-1-0:1.1.1.2
	netbsd-10:1.1.1.1.0.2
	netbsd-10-base:1.1.1.1
	ctwm-4-0-3:1.1.1.1
	CTWM:1.1.1;
locks; strict;
comment	@# @;


1.1
date	2021.04.11.08.36.54;	author nia;	state Exp;
branches
	1.1.1.1;
next	;
commitid	PjWN65car4ihSQOC;

1.1.1.1
date	2021.04.11.08.36.54;	author nia;	state Exp;
branches;
next	1.1.1.2;
commitid	PjWN65car4ihSQOC;

1.1.1.2
date	2023.07.05.07.36.07;	author nia;	state Exp;
branches;
next	;
commitid	W6xbaV6G6L8elAvE;


desc
@@


1.1
log
@Initial revision
@
text
@#
# Compiler/stdlib feature checks for ctwm
#


# Expect and try to enforce a C99 capable compiler.  There doesn't seem
# an obvious way to be sure in a fully portable way, but we probably
# don't work well in places that compile with something other than a
# program called like 'cc', and a cc that supports C99 should accept -std
# calls, so that's good enough.  Lacking it is not (yet) a fatal error,
# but is a sign that it's a compiler or platform we're moving further
# away from.
#
# Known alternate spellings:
#   -xc99  (Sun C 5.10 SunOS_i386, sunstudio12.1, OpenIndiana)
include(CheckCCompilerFlag)
set(c99_flag_options -std=c99 -xc99)
foreach(_C99_FLAG ${c99_flag_options})
	# CheckCCompilerFlag calls into CheckCSourceCompiles, which won't do
	# anything if the result var is already set in the cache, so we have
	# to unset it.  Otherwise, the second and later invocations don't
	# actually do anything, and it'll never check any flag after the
	# first.
	unset(COMPILER_C99_FLAG CACHE)
	check_c_compiler_flag(${_C99_FLAG} COMPILER_C99_FLAG)
	if(COMPILER_C99_FLAG)
		set(C99_FLAG ${_C99_FLAG})
		break()
	endif(COMPILER_C99_FLAG)
endforeach(_C99_FLAG)
if(C99_FLAG)
	message(STATUS "Enabling C99 flag: ${C99_FLAG}")
	add_definitions(${C99_FLAG})
else()
	message(WARNING "Compiler doesn't support known C99 flag, "
			"building without it.")
endif(C99_FLAG)


# With -std=c99, GNU libc's includes get strict about what they export.
# Particularly, a lot of POSIX stuff doesn't get defined unless we
# explicitly ask for it.  Do our best at checking for what's there...
check_include_files(features.h HAS_FEATURES_H)
if(HAS_FEATURES_H)
	# Check if including it with our args sets __USE_ISOC99; that's a
	# sign it's what we're looking for here.
	check_symbol_exists(__USE_ISOC99 features.h SETS_USE_ISOC99)
	if(SETS_USE_ISOC99)
		# OK, it does.  Assume that's a good enough test that things are
		# acting as we expect.
		set(GLIBC_FEATURE_FLAGS
			"-D_POSIX_C_SOURCE=200809L"
			"-D_XOPEN_SOURCE=700"
			)
		# asprintf() seems to need _GNU_SOURCE; no other way to expose it
		# I can find.
		check_symbol_exists(__GNU_LIBRARY__ features.h SETS_GNU_LIBRARY)
		if(SETS_GNU_LIBRARY)
			list(APPEND GLIBC_FEATURE_FLAGS "-D_GNU_SOURCE")
		endif(SETS_GNU_LIBRARY)

		message(STATUS "Enabling glibc feature macros: ${GLIBC_FEATURE_FLAGS}")
		add_definitions(${GLIBC_FEATURE_FLAGS})
	endif()
endif(HAS_FEATURES_H)


# Some compilers (like Sun's) don't take -W flags for warnings.  Do a
# quick check with -Wall.  They're mostly for devs, so we don't care THAT
# much, I guess...  maybe we should be more thorough about checking the
# flags we use too, but worry about what when it becomes an issue.
check_c_compiler_flag("-Wall" COMPILER_TAKES_WALL)
if(NOT COMPILER_TAKES_WALL)
	message(STATUS "Compiler doesn't like -Wall, disabling warnings.")
	set(NO_WARNS 1)
endif(NOT COMPILER_TAKES_WALL)

if(NOT NO_WARNS)
	add_definitions(${STD_WARNS})
	message(STATUS "Enabling standard warnings.")
endif(NOT NO_WARNS)
@


1.1.1.1
log
@ctwm-4.0.3
@
text
@@


1.1.1.2
log
@ctwm-4.1.0
@
text
@a13 4
# cmake 3.1+ has C_STANDARD and related vars that seem like they'd help
# with this, but it's unclear that they actually solve the whole
# problem...
#
d17 47
a63 81
set(MANUAL_C_STD_FLAG true)
if(NOT MANUAL_C_STD_FLAG)
	# This is the Better Way(tm), but is disabled by default because, as
	# with the manual one below, the added arg doesn't apply in
	# check_symbol_exists(), so it screws up the tests below.  I'm unable
	# to find a way to get info from cmake about what arg it would add
	# for the specified standard, so we can't pull it out manually to add
	# like we do our found C99_FLAG below, so...
	if(NOT "c_std_99" IN_LIST CMAKE_C_COMPILE_FEATURES)
		message(WARNING "cmake doesn't know about c99 support for this "
			"compiler, trying manual search...")
		set(MANUAL_C_STD_FLAG true)
	else()
		message(STATUS "Enabling C99 mode")
		set(CMAKE_C_EXTENSIONS false)
		set(CMAKE_C_STANDARD 99)
	endif()
endif()
if(MANUAL_C_STD_FLAG)
	set(c99_flag_options -std=c99 -xc99)
	foreach(_C99_FLAG ${c99_flag_options})
		# CheckCCompilerFlag calls into CheckCSourceCompiles, which won't do
		# anything if the result var is already set in the cache, so we have
		# to unset it.  Otherwise, the second and later invocations don't
		# actually do anything, and it'll never check any flag after the
		# first.
		unset(COMPILER_C99_FLAG CACHE)
		check_c_compiler_flag(${_C99_FLAG} COMPILER_C99_FLAG)
		if(COMPILER_C99_FLAG)
			set(C99_FLAG ${_C99_FLAG})
			break()
		endif(COMPILER_C99_FLAG)
	endforeach(_C99_FLAG)
	if(C99_FLAG)
		message(STATUS "Enabling C99 flag: ${C99_FLAG}")
		add_definitions(${C99_FLAG})
	else()
		message(WARNING "Compiler doesn't support known C99 flag, "
				"building without it.")
	endif(C99_FLAG)
endif()



# With -std=c99, some systems/compilers/etc enable ANSI strictness, and
# so don't export symbols from headers for e.g. a lot of POSIX etc stuff.
# So we may need some extra -D's.
# Some refs:
# https://www.gnu.org/software/libc/manual/html_node/Feature-Test-Macros.html
# https://illumos.org/man/5/standards

# Somewhat irritatingly, check_symbol_exists() doesn't use the extra
# flags we set above for the C standard.  So we have to add it manually,
# and stash up the old C_R_D.  x-ref above.
set(OLD_CMAKE_REQUIRED_DEFINITIONS ${CMAKE_REQUIRED_DEFINITIONS})
list(APPEND CMAKE_REQUIRED_DEFINITIONS ${C99_FLAG})

# What might and will we add?
set(AVAIL_SYM_FLAGS -D_POSIX_C_SOURCE=200809L -D_XOPEN_SOURCE=700
	-D_GNU_SOURCE -D__EXTENSIONS__)
set(EXTRA_SYM_FLAGS "")

# Abstract
macro(_check_func_flag FUNC HEADER)
	unset(_HAS_FUNC CACHE)
	check_symbol_exists(${FUNC} ${HEADER} _HAS_FUNC)
	if(NOT _HAS_FUNC)
		foreach(_SFLAG ${AVAIL_SYM_FLAGS})
			unset(_HAS_FUNC CACHE)
			list(APPEND CMAKE_REQUIRED_DEFINITIONS ${_SFLAG})
			check_symbol_exists(${FUNC} ${HEADER} _HAS_FUNC)
			list(REMOVE_ITEM CMAKE_REQUIRED_DEFINITIONS ${_SFLAG})
			if(_HAS_FUNC)
				message(STATUS "${FUNC}() needs ${_SFLAG}")
				list(APPEND EXTRA_SYM_FLAGS ${_SFLAG})
				break()
			endif()
		endforeach()
		if(NOT _HAS_FUNC)
			message(WARNING "Couldn't find def for ${FUNC}, not good...")
		endif()
d65 1
a65 24
	unset(_HAS_FUNC CACHE)
endmacro(_check_func_flag)

# strdup is POSIX, so see if we have to ask for that.  Probably
# _POSIX_C_SOURCE.
_check_func_flag(strdup string.h)

# isascii falls into XOPEN on glibc, POSIX on Illumos.
_check_func_flag(isascii ctype.h)

# asprintf() is even weirder.  glibc apparently usually uses _GNU_SOURCE,
# Illumos has a pure __EXTENSIONS__
_check_func_flag(asprintf stdio.h)

if(EXTRA_SYM_FLAGS)
	list(REMOVE_DUPLICATES EXTRA_SYM_FLAGS)
	message(STATUS "Adding extra visibility flags: ${EXTRA_SYM_FLAGS}")
	add_definitions(${EXTRA_SYM_FLAGS})
endif()

# And restore
set(CMAKE_REQUIRED_DEFINITIONS ${OLD_CMAKE_REQUIRED_DEFINITIONS})


@