Download raw body.
py-zstd, don't force -O3, avoid -fno-lto
On Mon, 20 Oct 2025 19:54:31 +0100
Stuart Henderson <stu@spacehopper.org> wrote:
> -fno-lto breaks build on sparc64, which is bad for a "nojunk" port
>
> ok?
Thanks for finding this, I committed it.
>
>
> Index: Makefile
> ===================================================================
> RCS file: /cvs/ports/archivers/py-zstd/Makefile,v
> diff -u -p -r1.1.1.1 Makefile
> --- Makefile 27 Aug 2025 21:29:55 -0000 1.1.1.1
> +++ Makefile 20 Oct 2025 18:53:30 -0000
> @@ -1,6 +1,7 @@
> COMMENT = zstd bindings for python
>
> MODPY_DISTV = 1.5.7.2
> +REVISION = 0
> DISTNAME = zstd-${MODPY_DISTV}
> PKGNAME = py-${DISTNAME}
>
> Index: patches/patch-setup_py
> ===================================================================
> RCS file: patches/patch-setup_py
> diff -N patches/patch-setup_py
> --- /dev/null 1 Jan 1970 00:00:00 -0000
> +++ patches/patch-setup_py 20 Oct 2025 18:53:30 -0000
> @@ -0,0 +1,26 @@
> +- disable -fno-lto; base-gcc doesn't understand it
> +- don't build with -O3 by default
> +
> +Index: setup.py
> +--- setup.py.orig
> ++++ setup.py
> +@@ -355,8 +355,8 @@ if BUILD_SPEED3:
> + 'msvc': ['/O3', ],
> + 'mingw32': ['-O3',],
> + 'unix': ['-O3',],
> +- 'clang': ['-O3',],
> +- 'gcc': ['-O3',],
> ++ 'clang': ['',],
> ++ 'gcc': ['',],
> + }
> + ###
> + # DVERSION - pass module version string
> +@@ -399,7 +399,7 @@ if SUP_DEBUG:
> + COPT[comp].extend([ '-DZSTD_DEBUG=1','-g',
> + ])
> +
> +-if BUILD_NO_LTO:
> ++if 0 and BUILD_NO_LTO:
> + for comp in COPT:
> + if comp == 'msvc':
> + pass
py-zstd, don't force -O3, avoid -fno-lto