Re: [PATCH 1/6] ldbl-128ibm-compat: Add regular character printing functions

["Gabriel F. T. Gomes" <gabriel@inconstante.net.br>]

On Wed, 14 Aug 2019, Florian Weimer wrote:

>* Gabriel F. T. Gomes:
>
>> diff --git a/sysdeps/ieee754/ldbl-128ibm-compat/Makefile b/sysdeps/ieee754/ldbl-128ibm-compat/Makefile
>> index 412beb5b5c..102973add9 100644
>> --- a/sysdeps/ieee754/ldbl-128ibm-compat/Makefile
>> +++ b/sysdeps/ieee754/ldbl-128ibm-compat/Makefile  
>
>> +# The following rule ensures that the string conversion routines will
>> +# always be built in IBM long double mode, with additional support for
>> +# IEEE binary128, through the use of -mabi=ibmlongdouble and -mfloat128.
>> +ldbl-ibm128-files := $(objpfx)test-%-ibm128^ \
>> +		     $(foreach r,$(ldbl-extra-routines),$(objpfx)$(r)^) \
>> +		     $(foreach r,$(ldbl-extra-routines),$(objpfx)$(r)-internal^)
>> +obj-suf-foreach = $(foreach suf,$(all-object-suffixes),$(subst ^,$(suf),$(1)))  
>
>Why isn't the assignment to obj-suf-foreach dead?

Oh!  There's a hunk missing in this patch.  It should have the following
extra statement:

+$(call obj-suf-foreach,$(ldbl-ibm128-files)): \
+  sysdep-CFLAGS += -mabi=ibmlongdouble

I accidentally removed it when preparing the patch for submission.  Sorry
and thanks for catching it.

It does not affect the test results because there's nothing currently
adding -mabi=ieeelongdouble to the build of powerpc64le files.  Anyway,
I'll reinstate this hunk locally, because it makes it clear *and explicit*
that these files must be built with -mabi=ibmlongdouble, regardless of
what becomes the default (if the default changes).

>Shouldn't this patch have abilist updates?

I kept the abilist change in the last commit on the branch [1] I mentioned
in the cover letter.  In that commit, this sysdeps directory
(ldbl-128ibm-compat) gets actually added to the Implies list.

The idea behind this decision is that the header redirections in our
installed headers (e.g.: stdio-ldbl.h) is implemented in a way that it
works for all functions or for none, so, I'm adding the functions to the
repository now (and testing with the development branch), but actually
enabling them only happens in the last commit.

[1] https://sourceware.org/git/?p=glibc.git;a=shortlog;h=refs/heads/gabriel/powerpc-ieee128-printscan