On 2015/05/15 23:26, Mark Kettenis wrote:
> Use $CC to link shared library to make sure crtbeginS.o gets linked in.
>
> Guess the configure.in patch isn't really needed, but good for
> completeness. What's the policy here?
Generally it's preferred to patch only the input file/s (leave the
generated configure script unpatched), and setup the port to run
autoconf at build time.
We make an exception for "quick and dirty" fixes which have no hope of
going upstream, and are difficult to work out how to patch the input
file. But in that case, we're more likely to hit conflicts when
updating the port later (and these are often nasty with autoconf
generated files).
So I would prefer to drop patch-configure, and add this on top of
your diff (2.69 because that's what upstream used originally).
Then it's OK sthen@.
--- Makefile 14 Mar 2015 22:26:21 -0000 1.39
+++ Makefile 15 May 2015 22:13:48 -0000
@@ -39,7 +39,8 @@ MODPY_RUNDEP= No
LIB_DEPENDS= textproc/libxml
SEPARATE_BUILD= Yes
-CONFIGURE_STYLE= gnu
+CONFIGURE_STYLE= autoconf
+AUTOCONF_VERSION= 2.69
CONFIGURE_ARGS= ${CONFIGURE_SHARED} \
--enable-threads \
--with-libtool \
> Anybody within the ports team regularly pushing stuff upstream for bind?
I can pass it on. (I should feed them the socket.c "stdint.h" patch too).
