[PATCH 1/5] Always build wayland-scanner

[Thierry Reding]

On Tue, May 27, 2014 at 09:22:30AM +0300, Pekka Paalanen wrote:
> On Mon, 19 May 2014 17:12:39 +0200
> Thierry Reding <thierry.reding at gmail.com> wrote:
> 
> > From: Thierry Reding <treding at nvidia.com>
> > 
> > When cross-compiling, wayland-scanner cannot be used to generate files
> > because it is built for the host architecture. To overcome this, users
> > currently need to instruct the build not to create wayland-scanner and
> > provide a native variant in the PATH. This has the disadvantage of not
> > installing the scanner either and in turn users running a distribution
> > that was cross-built cannot use it.
> > 
> > This commit changes the build process to always create and install the
> > wayland-scanner binary. The configure script determines whether to use
> > the version being built or a native version found in the PATH.
> > 
> > Signed-off-by: Thierry Reding <treding at nvidia.com>
> > ---
> >  Makefile.am  | 11 +++--------
> >  configure.ac | 31 ++++++++++++++++---------------
> >  2 files changed, 19 insertions(+), 23 deletions(-)
> > 
> > diff --git a/Makefile.am b/Makefile.am
> > index c15d8b8b853d..028d0d4837b5 100644
> > --- a/Makefile.am
> > +++ b/Makefile.am
> > @@ -61,25 +61,20 @@ nodist_libwayland_client_la_SOURCES =		\
> >  
> >  pkgconfig_DATA += src/wayland-client.pc src/wayland-server.pc
> >  
> > -if ENABLE_SCANNER
> > -wayland_scanner = $(top_builddir)/wayland-scanner
> >  bin_PROGRAMS = wayland-scanner
> >  wayland_scanner_SOURCES = src/scanner.c
> >  wayland_scanner_LDADD = $(EXPAT_LIBS) libwayland-util.la
> >  $(BUILT_SOURCES) : wayland-scanner
> >  pkgconfig_DATA += src/wayland-scanner.pc
> > -else
> > -wayland_scanner = wayland-scanner
> > -endif
> >  
> >  protocol/%-protocol.c : $(top_srcdir)/protocol/%.xml
> > -	$(AM_V_GEN)$(MKDIR_P) $(dir $@) && $(wayland_scanner) code < $< > $@
> > +	$(AM_V_GEN)$(MKDIR_P) $(dir $@) && $(WAYLAND_SCANNER) code < $< > $@
> >  
> >  protocol/%-server-protocol.h : $(top_srcdir)/protocol/%.xml
> > -	$(AM_V_GEN)$(MKDIR_P) $(dir $@) && $(wayland_scanner) server-header < $< > $@
> > +	$(AM_V_GEN)$(MKDIR_P) $(dir $@) && $(WAYLAND_SCANNER) server-header < $< > $@
> >  
> >  protocol/%-client-protocol.h : $(top_srcdir)/protocol/%.xml
> > -	$(AM_V_GEN)$(MKDIR_P) $(dir $@) && $(wayland_scanner) client-header < $< > $@
> > +	$(AM_V_GEN)$(MKDIR_P) $(dir $@) && $(WAYLAND_SCANNER) client-header < $< > $@
> >  
> >  BUILT_SOURCES =					\
> >  	$(nodist_libwayland_server_la_SOURCES)	\
> > diff --git a/configure.ac b/configure.ac
> > index 393310cd6699..d14142bbc8c3 100644
> > --- a/configure.ac
> > +++ b/configure.ac
> > @@ -31,6 +31,17 @@ AC_PROG_CC
> >  LT_PREREQ([2.2])
> >  LT_INIT
> >  
> > +if test "x$cross_compiling" = "xyes"; then
> > +	AC_PATH_PROG(WAYLAND_SCANNER, wayland-scanner)
> > +
> > +	if test "x$WAYLAND_SCANNER" = "x"; then
> > +		AC_MSG_ERROR([Cross-compiling but wayland-scanner not found.])
> > +	fi
> > +else
> > +	WAYLAND_SCANNER='$(top_builddir)/wayland-scanner'
> > +	AC_SUBST([WAYLAND_SCANNER])
> > +fi
> 
> Is it ok to have AC_SUBST inside a condition?
> I found one message where it seems to imply, that it is not a good
> practice:
> http://lists.gnu.org/archive/html/autoconf/2010-01/msg00091.html
> 
> Otherwise looks fine to me.

I think it doesn't matter at all where AC_SUBST is placed since it will
be evaluated outside of the shell context anyway. If you prefer I can
move it outside of the conditional block, though.

Thierry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/wayland-devel/attachments/20140527/4932ec99/attachment.sig>