Re: [PATCH 3/3] Enable bash-ptest

[Richard Purdie <richard.purdie@linuxfoundation.org>]

On Thu, 2012-08-23 at 16:20 +0200, Björn Stenberg wrote:
> Patch Makefile.in to allow test programs be built on host and ran on
> target.
> Patch tests/run-all to output PASS/FAIL for each testcase.
> Patch recipe to build and install test programs and test suite.
> 
> ---
>  .../bash/bash-4.2/build-tests.patch                |   29 ++++++++++++++++++++
>  meta/recipes-extended/bash/bash-4.2/run-ptest      |    2 +
>  .../bash/bash-4.2/test-output.patch                |   19 +++++++++++++
>  meta/recipes-extended/bash/bash.inc                |   18 ++++++++++++
>  meta/recipes-extended/bash/bash_4.2.bb             |    5 +++-
>  5 files changed, 72 insertions(+), 1 deletions(-)
>  create mode 100644 meta/recipes-extended/bash/bash-4.2/build-tests.patch
>  create mode 100644 meta/recipes-extended/bash/bash-4.2/run-ptest
>  create mode 100644 meta/recipes-extended/bash/bash-4.2/test-output.patch
> 
> diff --git a/meta/recipes-extended/bash/bash-4.2/build-tests.patch b/meta/recipes-extended/bash/bash-4.2/build-tests.patch
> new file mode 100644
> index 0000000..e92fab3
> --- /dev/null
> +++ b/meta/recipes-extended/bash/bash-4.2/build-tests.patch
> @@ -0,0 +1,29 @@
> +diff -uNr a/Makefile.in b/Makefile.in
> +--- a/Makefile.in	2012-06-14 10:15:15.063465304 +0200
> ++++ b/Makefile.in	2012-06-14 10:47:02.985115849 +0200
> +@@ -827,18 +827,21 @@
> + 	fi
> + 
> + recho$(EXEEXT):		$(SUPPORT_SRC)recho.c
> +-	@$(CC_FOR_BUILD) $(CCFLAGS_FOR_BUILD) -o $@ $(SUPPORT_SRC)recho.c ${LIBS_FOR_BUILD}
> ++	@$(CC) $(CCFLAGS) $(LDFLAGS) -o $@ $<
> + 
> + zecho$(EXEEXT):		$(SUPPORT_SRC)zecho.c
> +-	@$(CC_FOR_BUILD) $(CCFLAGS_FOR_BUILD) -o $@ $(SUPPORT_SRC)zecho.c ${LIBS_FOR_BUILD}
> ++	@$(CC) $(CCFLAGS) $(LDFLAGS) -o $@ $<
> + 
> + printenv$(EXEEXT):	$(SUPPORT_SRC)printenv.c
> +-	@$(CC_FOR_BUILD) $(CCFLAGS_FOR_BUILD) -o $@ $(SUPPORT_SRC)printenv.c ${LIBS_FOR_BUILD}
> ++	@$(CC) $(CCFLAGS) $(LDFLAGS) -o $@ $<
> + 
> + xcase$(EXEEXT):	$(SUPPORT_SRC)xcase.c
> +-	@$(CC_FOR_BUILD) $(CCFLAGS_FOR_BUILD) -o $@ $(SUPPORT_SRC)xcase.c ${LIBS_FOR_BUILD}
> ++	@$(CC) $(CCFLAGS) $(LDFLAGS) -o $@ $<
> + 
> + test tests check:	force $(Program) $(TESTS_SUPPORT)
> ++	$(MAKE) runtest
> ++
> ++runtest:
> + 	@-test -d tests || mkdir tests
> + 	@cp $(TESTS_SUPPORT) tests
> + 	@( cd $(srcdir)/tests && \


As it stands, I can bet on upstream saying "no" to this for good reason.
A better concept might be to introduce $(CC_FOR_TEST) which could
default to CC_FOR_BUILD. This would let us do what you're thinking here
yet keep upstream happy too.

Just thinking out loud really...

Cheers,

Richard