On Thu, 22 Sep 2022 12:28:56 +1000
Jonathan Gray <j...@jsg.id.au> wrote:
> On Wed, Sep 21, 2022 at 10:13:44PM -0400, George Koehler wrote:
> This uses the compiler builtin __BYTE_ORDER__ instead of endian.h
> BYTE_ORDER.
>
> Both could be avoided by using htole32() instead. Then the two
> ifdefs could be removed as well.
My 1st patch used swap32; this 2nd patch uses le32toh and works
equally well. ok for either swap32 or le32toh?
Index: Makefile
===================================================================
RCS file: /cvs/ports/graphics/webp-pixbuf-loader/Makefile,v
retrieving revision 1.1.1.1
diff -u -p -r1.1.1.1 Makefile
--- Makefile 12 Aug 2022 11:42:55 -0000 1.1.1.1
+++ Makefile 22 Sep 2022 03:12:56 -0000
@@ -3,6 +3,7 @@ COMMENT= WebP GDK Pixbuf Loader library
GH_ACCOUNT= aruiz
GH_PROJECT= webp-pixbuf-loader
GH_TAGNAME= 0.0.6
+REVISION= 0
CATEGORIES= graphics
Index: patches/patch-io-webp_c
===================================================================
RCS file: patches/patch-io-webp_c
diff -N patches/patch-io-webp_c
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-io-webp_c 22 Sep 2022 03:12:56 -0000
@@ -0,0 +1,27 @@
+Use <endian.h> on BE_ARCHS; OpenBSD doesn't have <byteswap.h>.
+
+Index: io-webp.c
+--- io-webp.c.orig
++++ io-webp.c
+@@ -12,9 +12,7 @@
+
+ #include "io-webp.h"
+
+-#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+-#include <byteswap.h>
+-#endif
++#include <endian.h>
+
+ #define IMAGE_READ_BUFFER_SIZE 65535
+
+@@ -278,9 +276,7 @@ gdk_pixbuf__webp_anim_load_increment (gpointer co
+ /* The next 4 bytes give the size of the webp container less
the 8 byte header. */
+ uint32_t anim_size = *(uint32_t *) (buf + 4); /* gives file
size not counting the first 8 bytes. */
+
+- #if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+- anim_size = bswap_32(anim_size);
+- #endif
++ anim_size = le32toh(anim_size);
+
+ uint32_t file_size = anim_size + 8;
+ if (file_size < size) {
