On Mon, 17 Mar 2025, Jacek Caban wrote:
On 17.03.2025 16:14, Martin Storsjö wrote:
On Sun, 16 Mar 2025, Jacek Caban wrote:
Allows using additional macros like __arm64ec__ or __SIZEOF_LONG_DOUBLE__.
---
Commit message typo, "standart"
mingw-w64-crt/Makefile.am | 16 +++++------
mingw-w64-crt/def-include/crt-aliases.def.in | 2 +-
mingw-w64-crt/def-include/func.def.in | 10 +++----
.../api-ms-win-crt-math-l1-1-0.def.in | 6 ++--
mingw-w64-crt/lib-common/msvcr120_app.def.in | 6 ++--
mingw-w64-crt/lib-common/msvcrt.def.in | 24 ++++++++--------
mingw-w64-crt/lib-common/ntdll.def.in | 28 +++++++++----------
mingw-w64-crt/lib-common/ntdllcrt.def.in | 8 +++---
.../lib-common/ucrtbase-common.def.in | 4 +--
9 files changed, 52 insertions(+), 52 deletions(-)
diff --git a/mingw-w64-crt/Makefile.am b/mingw-w64-crt/Makefile.am
index 15642b3ff..217c0325f 100644
--- a/mingw-w64-crt/Makefile.am
+++ b/mingw-w64-crt/Makefile.am
@@ -4167,28 +4167,28 @@ endif
$(RANLIB) $@
lib32/%.def: lib32/%.def.in
- $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -undef -P
-I$(top_srcdir)/def-include -DDEF_I386 > $@
+ $(MKDIR_P) $(@D) && $(CPP) -x c $(CPPFLAGS32) $< -Wp,-w -P
-I$(top_srcdir)/def-include > $@
lib64/%.def: lib64/%.def.in
- $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -undef -P
-I$(top_srcdir)/def-include -DDEF_X64 > $@
+ $(MKDIR_P) $(@D) && $(CPP) -x c $(CPPFLAGS64) $< -Wp,-w -P
-I$(top_srcdir)/def-include > $@
libarm32/%.def: libarm32/%.def.in
- $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -undef -P
-I$(top_srcdir)/def-include -DDEF_ARM32 > $@
+ $(MKDIR_P) $(@D) && $(CPP) -x c $(CPPFLAGSARM32) $< -Wp,-w -P
-I$(top_srcdir)/def-include > $@
libarm64/%.def: libarm64/%.def.in
- $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -undef -P
-I$(top_srcdir)/def-include -DDEF_ARM64 > $@
+ $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -P
-I$(top_srcdir)/def-include -DDEF_ARM64 > $@
Why doesn't this use CPPFLAGSARM64, for consistency? We do reference that
variable in a number of places, even though we never really set it.
Good point, I will add it.
lib32/%.def: lib-common/%.def.in
- $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -undef -P
-I$(top_srcdir)/def-include -DDEF_I386 > $@
+ $(MKDIR_P) $(@D) && $(CPP) -x c $(CPPFLAGS32) $< -Wp,-w -P
-I$(top_srcdir)/def-include > $@
lib64/%.def: lib-common/%.def.in
- $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -undef -P
-I$(top_srcdir)/def-include -DDEF_X64 | sed -E 's/^([^ ]+)@[0-9]+(
|$$)/\1\2/' > $@
+ $(MKDIR_P) $(@D) && $(CPP) -x c $(CPPFLAGS64) $< -Wp,-w -P
-I$(top_srcdir)/def-include | sed -E 's/^([^ ]+)@[0-9]+( |$$)/\1\2/' > $@
libarm32/%.def: lib-common/%.def.in
- $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -undef -P
-I$(top_srcdir)/def-include -DDEF_ARM32 | sed -E 's/^([^ ]+)@[0-9]+(
|$$)/\1\2/' > $@
+ $(MKDIR_P) $(@D) && $(CPP) -x c $(CPPFLAGSARM32) $< -Wp,-w -P
-I$(top_srcdir)/def-include | sed -E 's/^([^ ]+)@[0-9]+( |$$)/\1\2/' > $@
libarm64/%.def: lib-common/%.def.in
- $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -undef -P
-I$(top_srcdir)/def-include -DDEF_ARM64 | sed -E 's/^([^ ]+)@[0-9]+(
|$$)/\1\2/' > $@
+ $(MKDIR_P) $(@D) && $(CPP) -x c $< -Wp,-w -P
-I$(top_srcdir)/def-include -DDEF_ARM64 | sed -E 's/^([^ ]+)@[0-9]+(
|$$)/\1\2/' > $@
This probably is ok. I think I initially did this with the setup where
preprocessing could be done with any plain "cpp" executable (as we also use
-undef), but this change does make sense, with what you want to do in the
following patch.
It's worth pointing out to other readers here, that the CPPFLAGS* variables
themselves don't really add much of value here, it's mostly the fact that
you remove -undef, that the default compiler defines become available.
It's important if we build both 32-bit and 64-bit libraries, to make sure
that -m32 is passed when appropriate.
Yes, in that particular case it's important.
// Martin
_______________________________________________
Mingw-w64-public mailing list
Mingw-w64-public@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
