Hi Eric,
> obstack-printf-posix-tests had a dependency on obstack-printf-tests.
> Somehow in the conversion to caching, we no longer pick up implicit
> dependencies of tests on their main module (that is, obstack-printf-tests no
> longer implies obstack-printf).
Wheee... You are right. This implicit dependency is documented:
"A test modules @code{foo-tests} implicity depends on the corresponding
non-test module @code{foo}."
but it got lost through my commits yesterday:
$ ./gnulib-tool --extract-dependencies obstack-printf-posix-tests
obstack-printf-posix
obstack-printf-tests
became
$ ./gnulib-tool --extract-dependencies obstack-printf-posix-tests
obstack-printf-tests
Fixed as follows:
2010-02-08 Bruno Haible <br...@clisp.org>
gnulib-tool: Fix typo in 2010-02-07 commit.
* gnulib-tool (func_get_dependencies): Fix typo in last commit.
Reported by Eric Blake.
--- gnulib-tool.orig Mon Feb 8 21:28:01 2010
+++ gnulib-tool Mon Feb 8 21:27:39 2010
@@ -1870,7 +1870,7 @@
*-tests)
fgd1="$1"
func_remove_suffix fgd1 '-tests'
- echo "$fgdl"
+ echo "$fgd1"
;;
esac
# Then the explicit dependencies listed in the module description.
> * modules/obstack-printf-posix (Depends-on): Add obstack-printf.
This is not needed. The modules 'obstack-printf-posix' and 'obstack-printf'
are two modules that use the same source code but different .m4 macros.
There is no need for running gl_FUNC_OBSTACK_PRINTF when we are also running
gl_FUNC_OBSTACK_PRINTF_POSIX.
> + * modules/obstack-printf-posix-tests: Delete unused test module.
I disagree. This dependency is right: In order to test 'obstack-printf-posix',
we use the same source code as for testing 'obstack-printf', and - so far -
nothing more.
I would leave both module descriptions as they are. Sorry for the gnulib-tool
bug that introduced the confusion.
Bruno
