On 2023/05/13 09:18, Peter Hessler wrote: > darktable fails to build on arm64, and it seems it is because of badly > balanced parenthesis. With the patch in-tree, it always sets > DT_SUPPORTED_X86 1, and then arm also sets DT_SUPPORTED_ARMv8A 1 which > blows up later in the check. > > Thankfully we don't even need this patch, so simply remove it. > > builds fine on amd64 and arm64. > > OK? > > > Index: graphics/darktable/Makefile > =================================================================== > RCS file: /cvs/openbsd/ports/graphics/darktable/Makefile,v > retrieving revision 1.117 > diff -u -p -u -p -r1.117 Makefile > --- graphics/darktable/Makefile 8 May 2023 20:55:21 -0000 1.117 > +++ graphics/darktable/Makefile 13 May 2023 07:14:36 -0000 > @@ -6,6 +6,7 @@ V= 4.2.1 > DISTNAME= darktable-$V > PORTROACH= limitw:1,even skipv:release-$V > EXTRACT_SUFX= .tar.xz > +REVISION= 0
Bump doesn't hurt but not really needed, it's either a noop or a build fix depending on arch. > CATEGORIES= graphics x11 > > Index: graphics/darktable/patches/patch-src_is_supported_platform_h > =================================================================== > RCS file: graphics/darktable/patches/patch-src_is_supported_platform_h > diff -N graphics/darktable/patches/patch-src_is_supported_platform_h > --- graphics/darktable/patches/patch-src_is_supported_platform_h 8 May > 2023 20:55:21 -0000 1.6 > +++ /dev/null 1 Jan 1970 00:00:00 -0000 > @@ -1,19 +0,0 @@ > -Index: src/is_supported_platform.h > ---- src/is_supported_platform.h.orig > -+++ src/is_supported_platform.h > -@@ -24,13 +24,13 @@ > - #error "Unfortunately we only work on litte-endian systems." > - #endif > - > --#if (defined(__amd64__) || defined(__amd64) || defined(__x86_64__) || > defined(__x86_64)) > -+#if (defined(__amd64__) || defined(__amd64) || defined(__x86_64__) || > defined(__x86_64) || defined(__OpenBSD__)) definitely ok, that was clearly wrong > - #define DT_SUPPORTED_X86 1 > - #else > - #define DT_SUPPORTED_X86 0 > - #endif > - > --#if defined(__aarch64__) && (defined(__ARM_64BIT_STATE) && > defined(__ARM_ARCH) && (defined(__ARM_ARCH_8A) || __ARM_ARCH_PROFILE == 'A') > || defined(__APPLE__) || defined(__MINGW64__)) > -+#if defined(__aarch64__) && (defined(__ARM_64BIT_STATE) && > defined(__ARM_ARCH) && (defined(__ARM_ARCH_8A) || __ARM_ARCH_PROFILE == 'A') > || defined(__APPLE__) || defined(__OpenBSD__) || defined(__MINGW64__)) > - #define DT_SUPPORTED_ARMv8A 1 > - #else > - #define DT_SUPPORTED_ARMv8A 0 I haven't gone to boot the arm64 to check but I guess we define __ARM_ARCH_PROFILE (which was newly added to the ifdef upstream somewhere between the older and recently committed port update). Anyway if it builds it's right.
