python-pygobject fails to build from scratch

python-pygobject fails to build from scratch

[Mike (mwester)]

The patch below causes build failures for me :(

Specifically, with an empty tmp directory (and parallel make specified,
but I don't think that makes a difference in this case),
gobject-generate-constants is not yet built and staged.  In fact, it is
both created and used by this makefile, so there's no way that this
recipe should use the one in staging (in fact, if it's working for folks
today, it's because their builds are finding (an older) one left around
in staging).

Since I don't know what problem this patch and the others in the
associated commit were solving, I can't provide a fix, other than to
note that backing out just this patch makes it Work For Me(tm).

Thanks,
Mike (mwester)




--- packages/python/python-pygobject/generate-constants.patch
1726f18a8e65fb7fe7a860bd1bd79e09c0dd2de1
+++ packages/python/python-pygobject/generate-constants.patch
1726f18a8e65fb7fe7a860bd1bd79e09c0dd2de1
@@ -0,0 +1,18 @@
+Index: pygobject-2.14.2/gobject/Makefile.am
+===================================================================
+--- pygobject-2.14.2.orig/gobject/Makefile.am
++++ pygobject-2.14.2/gobject/Makefile.am
+@@ -21,11 +21,11 @@ if PLATFORM_WIN32
+ common_ldflags += -no-undefined
+ endif
+
+-constants.py: generate-constants$(EXEEXT) constants.py.in
++constants.py: constants.py.in
+ 	rm -f constants.py
+ 	cp $(srcdir)/constants.py.in constants.py
+ 	chmod 644 constants.py
+-	$(top_builddir)/gobject/generate-constants$(EXEEXT) >> constants.py
++	gobject-generate-constants$(EXEEXT) >> constants.py
+ 	chmod 444 constants.py
+
+ generate_constants_CFLAGS = $(GLIB_CFLAGS)

Re: python-pygobject fails to build from scratch

[Michael 'Mickey' Lauer]

Am Samstag 28 Juni 2008 17:28:56 schrieb Mike (mwester):
> The patch below causes build failures for me :(
>
> Specifically, with an empty tmp directory (and parallel make specified,
> but I don't think that makes a difference in this case),
> gobject-generate-constants is not yet built and staged.  In fact, it is
> both created and used by this makefile, so there's no way that this
> recipe should use the one in staging (in fact, if it's working for folks
> today, it's because their builds are finding (an older) one left around
> in staging).

That's strange. The situation is that pygobject 2.14.x compiles a helper which 
it calls during the build process which fails. Since we already had a 
pygobject-native for various other packages laying around, i made that stage 
the necessary helper and made pygobject DEPEND on pygobject-native.

Can you check whether pygobject-native is getting build and staged for you at 
the time pygobject is going to be compiled?

Update: Sorry, forget what I said, I had a patch to pygobject-native to 
commited :/

It should work now.
-- 
:M: