[PATCH] 18969 - multiple string test failures due to missing locale dependencies

[Martin Sebor]

I'm not sure what the protocol is -- do I need explicit approval
for this or can I commit the patch without one?

The follow-on patch is here:
   http://sourceware.org/ml/libc-alpha/2015-09/msg00723.html

On 09/28/2015 05:01 PM, Martin Sebor wrote:
> After committing the patch I noticed a number of other similar
> problems in other Makefiles. I tried to find a complete set by
> grepping tests (files that match the tst*.c and test*.c shell
> pattern) for calls to setlocale with a named locale, testing
> a subset of those, and fixing the errors I got by making the
> same tweaks to the Makefile as in the first patch. It's
> possible I still missed some.
>
> Attached is a follow-on patch with these tweaks.
>
> Martin
>
> PS In the process of this testing I ran into another, unrelated
> problem. Some of the Makefiles add test output files to a variable
> named tests-special. When make is invoked to build and run just
> one or a few tests, the invocation fails with an error like the
> one below. The error can be avoided by adding the .out files to
> the variable only when the tests variable is defined in the
> Makefile rather than on the command line. I haven't included
> these fixes in this patch since it's an unrelated problem.
> Instead, unless there are objections, I will post another patch
> later to correct this problem.
>
> make[2]: *** No rule to make target
> `/build/glibc-trunk/power8/stdio-common/tst-printf-bz18872.out', needed
> by `/build/glibc-trunk/power8/stdio-common/tst-printf-bz18872-mem.out'.
>   Stop.
>