On 13/12/23 09:06, Martin Storsjö wrote:
On Wed, 13 Dec 2023, Mark Harmstone wrote:
Signed-off-by: Mark Harmstone <m...@harmstone.com>
---
mingw-w64-crt/lib-common/wsmsvc.def | 3675 +++++++++++++++++++++++++++
mingw-w64-crt/lib32/Makefile.am | 1 +
mingw-w64-crt/lib32/wsmsvc.def | 3673 ++++++++++++++++++++++++++
mingw-w64-crt/lib64/Makefile.am | 1 +
mingw-w64-crt/libarm64/Makefile.am | 1 +
5 files changed, 7351 insertions(+)
create mode 100644 mingw-w64-crt/lib-common/wsmsvc.def
create mode 100644 mingw-w64-crt/lib32/wsmsvc.def
Note that we do have a version of this def file already, for arm32 only.
As we're adding one version in lib-common and reusing it for arm64, I wonder if
we should reuse/share it for arm32 as well? Are there many differences between
these symbols between x86_64 and i386, or between x86_64 and arm32? Depending
on that, we should either keep all 4 separate (I can generate ones for all 4
architectures by dumping DLLs, but if we generate it from import libraries you
should be able to do it yourself as well), or share them more than this.
Thanks Martin. Yes, I should have pointed out that this was deliberate. It
looks like the MSVC name mangling is slightly different between arm32 and i386
(but the same for x86_64 and arm64).
On the other hand, I wonder how much utility there is to this at all - as no
mingw tools really can use the MSVC C++ ABI anyway, so is it worth adding this
at all? (Or should we add it as a stub, empty def file?)
Well, there is that. It's worth adding wsmsvc though, as there's some C symbols
right at the end.
I would rather like to add support for MSVC name mangling to GCC one day, but I
suspect it'd be quite a big job...
Mark
_______________________________________________
Mingw-w64-public mailing list
Mingw-w64-public@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
