Bump to rc2, fix more big-endian issues, re-enable tests
This commit is contained in:
+29
-11
@@ -5,12 +5,15 @@
|
||||
%endif
|
||||
%global __provides_exclude_from .*/site-packages/.*\\.so$
|
||||
%global with_html 0
|
||||
%ifarch %{power64} s390x
|
||||
# disable tests on alt arches until resolved by upstream
|
||||
%global run_tests 0
|
||||
%else
|
||||
|
||||
# It seems like there's some kind of weird occasional error where a
|
||||
# build (often aarch64 or ppc64) will fail in one of the Stix font
|
||||
# tests with a huge RMS difference, but if you run the same build again,
|
||||
# you won't get the same error. Unless someone can figure out what's
|
||||
# going on, we just have to keep re-running the build until it doesn't
|
||||
# happen.
|
||||
%global run_tests 1
|
||||
%endif
|
||||
|
||||
|
||||
# On RHEL 7 onwards, don't build with wx:
|
||||
%if 0%{?rhel} >= 7
|
||||
@@ -48,11 +51,11 @@
|
||||
# Use the same directory of the main package for subpackage licence and docs
|
||||
%global _docdir_fmt %{name}
|
||||
|
||||
%global rctag b4
|
||||
%global rctag rc2
|
||||
|
||||
Name: python-matplotlib
|
||||
Version: 2.0.0
|
||||
Release: 0.6%{?rctag:.%{rctag}}%{?dist}
|
||||
Release: 0.7%{?rctag:.%{rctag}}%{?dist}
|
||||
Summary: Python 2D plotting library
|
||||
Group: Development/Libraries
|
||||
# qt4_editor backend is MIT
|
||||
@@ -69,15 +72,20 @@ Patch9: python-matplotlib-qhull.patch
|
||||
# https://github.com/matplotlib/matplotlib/issues/7134
|
||||
# https://github.com/matplotlib/matplotlib/issues/7158
|
||||
# https://github.com/matplotlib/matplotlib/issues/7159
|
||||
# https://github.com/matplotlib/matplotlib/issues/7797
|
||||
Patch10: python-matplotlib-increase-tests-tolerance.patch
|
||||
Patch11: python-matplotlib-increase-tests-tolerance-aarch64.patch
|
||||
Patch11: python-matplotlib-increase-tests-tolerance-aarch64ppc64.patch
|
||||
Patch13: python-matplotlib-increase-tests-tolerance-i686.patch
|
||||
# These two patches fix some integer type issues which broke matplotlib
|
||||
# These four patches all fix integer type issues which broke matplotlib
|
||||
# badly on ppc64 (big-endian)
|
||||
# https://github.com/matplotlib/matplotlib/pull/7768
|
||||
Patch14: https://github.com/matplotlib/matplotlib/commit/b0e4b6708d71df80999764eb4b65cc1d388a521f.patch
|
||||
# https://github.com/matplotlib/matplotlib/pull/7781
|
||||
Patch15: 0001-Fix-integer-types-for-font-metrics-in-PyGlyph-class.patch
|
||||
# https://github.com/matplotlib/matplotlib/pull/7791
|
||||
Patch16: 0001-Use-reliable-int-type-for-mesh-size-in-draw_quad_mes.patch
|
||||
# https://github.com/matplotlib/matplotlib/pull/7796
|
||||
Patch17: 0001-Only-byte-swap-16-bit-PNGs-on-little-endian-7792.patch
|
||||
|
||||
BuildRequires: freetype-devel
|
||||
BuildRequires: libpng-devel
|
||||
@@ -418,14 +426,16 @@ sed -i 's/\(USE_FONTCONFIG = \)False/\1True/' lib/matplotlib/font_manager.py
|
||||
%patch9 -p1 -b .qh
|
||||
%endif
|
||||
%patch10 -p1 -b .tests
|
||||
%ifarch aarch64
|
||||
%patch11 -p1 -b .tests-aarch64
|
||||
%ifarch aarch64 %{power64}
|
||||
%patch11 -p1 -b .tests-aarch64ppc64
|
||||
%endif
|
||||
%ifarch i686
|
||||
%patch13 -p1 -b .tests-i686
|
||||
%endif
|
||||
%patch14 -p1 -b .inttype
|
||||
%patch15 -p1 -b .moreints
|
||||
%patch16 -p1 -b .yetmoreints
|
||||
%patch17 -p1 -b .pngswap
|
||||
|
||||
chmod -x lib/matplotlib/mpl-data/images/*.svg
|
||||
chmod -x lib/matplotlib/{dates,sankey}.py
|
||||
@@ -624,6 +634,14 @@ PYTHONPATH=%{buildroot}%{python3_sitearch} \
|
||||
%endif
|
||||
|
||||
%changelog
|
||||
* Tue Jan 10 2017 Adam Williamson <awilliam@redhat.com> - 2.0.0-0.7.rc2
|
||||
- Update to 2.0.0rc2
|
||||
- Fix more big-endian integer issues
|
||||
- Apply the 'aarch64' test tolerance patch on ppc64 also (it's affected by same issues)
|
||||
- Tweak the 'i686' test tolerance patch a bit (some errors are gone, some new ones)
|
||||
- Re-enable test suite for all arches
|
||||
- Note a remaining quasi-random test issue that causes build to fail sometimes
|
||||
|
||||
* Mon Jan 09 2017 Adam Williamson <awilliam@redhat.com> - 2.0.0-0.6.b4
|
||||
- Fix another integer type issue which caused more issues on ppc64
|
||||
|
||||
|
||||
Reference in New Issue
Block a user