bcftools FTBFS on i386, powerpc, s390x due to failing tests

[Tiago Stürmer Daitx]

bcftools currently FTBFS on yakkety-proposed for i386, powerpc, s390x due to failing tests.

Number of failures per arch:
- i386: 74
- s390x: 10
- powerpc: 6

This numbers are bigger than the tests listed bellow because some tests are repeated with different arguments.

Some of those failures are related to Debian's 819617 bug, which seems to be related to nan handling by both bcftools and hts.

This has also been reported upstream @ https://github.com/samtools/bcftools/issues/389

The discussion upstream includes an issue about chars being unsigned on some archs, but that has already been fixed for 1.3.1 (current version we are using).

Of those, only the test merge.4.out fails on the 3 archs and is also the only failure in common between i386 and the other 2 archs.

s390x and powerpc have 5 common failures:
- idx_count.out
- idx.out
- large_chrom.20.1.2147483647.out
- large_chrom_csi_limit.20.1.2147483647.out
- merge.4.out

In addition to that s390x also fails on convert.gt.noHead.vcf.

The i386 exclusive failures are:
annotate.out
annotate2.out
annotate3.out
annotate4.out
annotate5.out
annotate8.out
annotate9.out
annotate10.out
filter.2.out
merge.abc.out
merge.2.all.out
merge.2.both.out
merge.2.none.out
merge.3.out
norm.merge.out
norm.merge.strict.out
norm.split.and.norm.out
norm.split.out
norm.merge.3.out
query.12.out
query.20.out
query.21.out
view.vectors.A.out
view.vectors.B.out
view.filter.1.out
view.filter.2.out
view.filter.3.out
view.filter.4.out
view.filter.5.out
view.filter.6.out
view.filter.7.out
view.filter.8.out
view.filter.9.out
view.filter.10.out
view.filter.11.out
view.filter.annovar.1.out

The amount of test failures reported is bigger than the amount of tests listed above because some tests are run more than once with different arguments.

Failure example on i386 (slight edited for clarity):
$ wdiff -3 test/merge.4.out test/merge.4.out.new
======================================================================
[-TR=1,2,2;TA=1,1;TG=1,2,3,2,.,3;AN=8;AC=2,2-]
{+TR=1,2,2;TA=1,1;TG=1,2,3,2,nan,3;AN=8;AC=2,2+}
======================================================================
[-TR=1,2,2;TA=1,1;TG=1,2,3,2,.,3;AN=8;AC=2,2-]
{+TR=1,2,2;TA=1,1;TG=1,2,3,2,nan,3;AN=8;AC=2,2+}
======================================================================
[-0/1:245,245,-1.56481e-39,.,.,.:.:.:.:.
  0/1:245,-1.56481e-39,1.09706e-28,.,.,.:.:.:.:.
  0/2:.:.:.:.:245
  0/2:.:.:.:.:245-]
{+0/1:245,245,-1.56481e-39,nan,nan,nan:.:.:.:.
  0/1:245,-1.56481e-39,1.09706e-28,nan,nan,nan:.:.:.:.
  0/2:nan,nan,nan,nan,nan,nan:.:.:.:245
  0/2:nan,nan,nan,nan,nan,nan:.:.:.:245+}
======================================================================
[-TR=1,2,2;TA=1,1;TG=1,2,3,2,.,3;AN=8;AC=2,2-]
{+TR=1,2,2;TA=1,1;TG=1,2,3,2,nan,3;AN=8;AC=2,2+}
======================================================================
[-0/1:245,245,-1.56481e-39,.,.,.:.:.:.:.
  0/1:245,-1.56481e-39,1.09706e-28,.,.,.:.:.:.:.
  0/2:.:.:.:.:245
  0/2:.:1,.,2:.,2:1,.,.,2,.,3:245-]
{+0/1:245,245,-1.56481e-39,nan,nan,nan:.:.:.:.
  0/1:245,-1.56481e-39,1.09706e-28,nan,nan,nan:.:.:.:.
  0/2:nan,nan,nan,nan,nan,nan:.:.:.:245
  0/2:nan,nan,nan,nan,nan,nan:1,.,2:.,2:1,.,.,2,.,3:245+}
======================================================================

[Tiago Stürmer Daitx (tdaitx)]

The failing build of htseq on i386, powerpc, and s390x is holding htseq on yakkety-proposed due to the chain dependency: htseq -> python-pysam -> bcftools.