On Sun, Sep 30, 2018 at 02:11:30PM +0200, Andreas Metzler wrote:
> On 2018-09-29 Adrian Bunk <b...@debian.org> wrote:
> > Source: plotutils
> > Version: 2.6-9
> > Severity: serious
> > Tags: ftbfs
>
> > https://tests.reproducible-builds.org/debian/rb-pkg/unstable/i386/plotutils.html
>
> > ...
> > FAIL: plot2hpgl
> > ===============
>
> > FAIL plot2hpgl.test (exit status: 1)
>
> That is a strange one. Building with -O0 succeeds. Starting from there I
> tried to find the offending source.
>...
> I do not get this.
The root of the problem is that the x87 FPU has more precision internally,
which can result in better but different results.
Does normal operation of plotutils need exact results,
or are these tests that are too strict?
In the latter case the tests should be fixed.
In the former case the following would fix the tests
(with a performance penalty for plotutils on i386):
--- debian/rules.old 2018-09-30 12:13:23.905785679 +0000
+++ debian/rules 2018-09-30 12:16:11.629784080 +0000
@@ -1,7 +1,14 @@
#!/usr/bin/make -f
# -*- makefile -*-
-export DEB_CFLAGS_MAINT_APPEND := -Wall
+DEB_CFLAGS_MAINT_APPEND := -Wall
+ifneq (,$(filter $(DEB_HOST_ARCH), i386))
+ DEB_CFLAGS_MAINT_APPEND += -ffloat-store
+ DEB_CXXFLAGS_MAINT_APPEND += -ffloat-store
+endif
+export DEB_CFLAGS_MAINT_APPEND
+export DEB_CXXFLAGS_MAINT_APPEND
+
export DEB_BUILD_MAINT_OPTIONS := hardening=+all
override_dh_auto_configure:
> cu Andreas
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
