Package: pandas / 1.1.5+dfsg-2

Metadata

Package Version Patches format
pandas 1.1.5+dfsg-2 3.0 (quilt)

Patch series

view the series file
Patch File delta Description
deb_nonversioneer_version.patch | (download)

pandas/__init__.py | 8 1 + 7 - 0 !
pandas/tests/api/test_api.py | 2 1 + 1 - 0 !
pandas/tests/test_common.py | 2 1 + 1 - 0 !
3 files changed, 3 insertions(+), 9 deletions(-)

 don't try to use git to find the version number

Needed as Debian buildds use tarballs

deb_doc_donotoverride_PYTHONPATH.patch | (download)

doc/make.py | 5 3 + 2 - 0 !
1 file changed, 3 insertions(+), 2 deletions(-)

 don't try to import from the source directory

Needed as we build the extension modules elsewhere

xfail_tests_nonintel_io.patch | (download)

pandas/_testing.py | 4 4 + 0 - 0 !
pandas/io/pytables.py | 11 10 + 1 - 0 !
pandas/io/stata.py | 5 5 + 0 - 0 !
pandas/tests/io/pytables/test_store.py | 11 11 + 0 - 0 !
pandas/tests/io/pytables/test_timezones.py | 3 3 + 0 - 0 !
pandas/tests/io/test_common.py | 6 3 + 3 - 0 !
pandas/tests/io/test_stata.py | 11 7 + 4 - 0 !
7 files changed, 43 insertions(+), 8 deletions(-)

 hdf5 and stata i/o are broken on some architectures

Fix some issues, warn on use and xfail tests for the remainder

armhf TestHDF5Store::test*encoding only sometimes crashes
(1.1.3+dfsg-1 passed on build but failed autopkgtest)

HDF5 and Stata are known to fail on big-endian architectures
Stata also fails on qemu-ppc64el, but not real ppc64el

deb_disable_googleanalytics.patch | (download)

pandas/tests/io/data/html/spam.html | 44 4 + 40 - 0 !
web/pandas/_templates/layout.html | 8 0 + 8 - 0 !
2 files changed, 4 insertions(+), 48 deletions(-)

 avoid privacy breach by google analytics
deb_no_strict_data.patch | (download)

setup.cfg | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 don't fail tests because we don't ship test data

This applies to users running them; our build/autopkgtest re-enable
this and use the data from the source tree

mathjax path.patch | (download)

doc/source/conf.py | 2 2 + 0 - 0 !
1 file changed, 2 insertions(+)

 use debian packaged mathjax

xfail_tests_nonintel_nannat.patch | (download)

pandas/core/dtypes/cast.py | 8 8 + 0 - 0 !
pandas/tests/dtypes/cast/test_downcast.py | 4 4 + 0 - 0 !
pandas/tests/frame/indexing/test_where.py | 4 4 + 0 - 0 !
pandas/tests/frame/test_analytics.py | 4 4 + 0 - 0 !
pandas/tests/indexes/datetimes/test_datetime.py | 4 4 + 0 - 0 !
pandas/tests/reductions/test_reductions.py | 4 4 + 0 - 0 !
pandas/tests/series/test_constructors.py | 6 5 + 1 - 0 !
pandas/tests/test_algos.py | 4 4 + 0 - 0 !
8 files changed, 37 insertions(+), 1 deletion(-)

 xfail nan <-> nat tests on non-x86 and warn on cast

pd.Series([np.nan]).astype('datetime64[ns]')[0] = pd.NaT on x86
but 1970-01-01 on arm* because float NaN -> int is undefined:
https://github.com/numpy/numpy/issues/8325
https://github.com/pandas-dev/pandas/issues/17792
https://github.com/pandas-dev/pandas/issues/26964

On s390x it's the maximum _positive_ value (2**63-1 ns = year 2262)

skip_noencoding_locales.patch | (download)

pandas/_config/localization.py | 2 2 + 0 - 0 !
1 file changed, 2 insertions(+)

 don't test datetime in locales with no encoding

Some datetime tests run the test in every available locale.
If this set includes locales without an encoding (currently dsb_DE
and sah_RU), it fails due to Python bug
https://bugs.python.org/issue20088

Failure log
https://tests.reproducible-builds.org/debian/rbuild/buster/amd64/pandas_0.23.3+dfsg-3.rbuild.log.gz

use_system_intersphinx.patch | (download)

doc/source/conf.py | 14 7 + 7 - 0 !
1 file changed, 7 insertions(+), 7 deletions(-)

 use packaged intersphinx indexes

contributor_list_not_in_tarball.patch | (download)

doc/sphinxext/contributors.py | 3 1 + 2 - 0 !
1 file changed, 1 insertion(+), 2 deletions(-)

 don't try to read a contributor list from the git log

Debian packages are built from tarballs, so there isn't a git log.

tests_dont_assume_endian.patch | (download)

pandas/tests/arrays/boolean/test_astype.py | 3 2 + 1 - 0 !
pandas/tests/arrays/boolean/test_construction.py | 3 2 + 1 - 0 !
pandas/tests/arrays/integer/test_dtypes.py | 3 2 + 1 - 0 !
pandas/tests/frame/methods/test_to_records.py | 4 4 + 0 - 0 !
pandas/tests/io/parser/test_c_parser_only.py | 2 1 + 1 - 0 !
pandas/tests/scalar/timedelta/test_arithmetic.py | 2 1 + 1 - 0 !
6 files changed, 12 insertions(+), 5 deletions(-)

 fix or skip tests that assume the wrong endianness

This is a bug in the tests not pandas itself -
the expected values explicitly specify little-endian

fix_random_seeds.patch | (download)

doc/source/getting_started/comparison/comparison_with_r.rst | 1 1 + 0 - 0 !
doc/source/user_guide/advanced.rst | 1 1 + 0 - 0 !
doc/source/user_guide/visualization.rst | 1 1 + 0 - 0 !
3 files changed, 3 insertions(+)

 use fixed seeds for reproducible pseudorandomness

xfail_c_locale.patch | (download)

pandas/tests/config/test_localization.py | 1 1 + 0 - 0 !
1 file changed, 1 insertion(+)

 xfail a test that doesn't work in the c locale

sphinx_no_pandas_theme.patch | (download)

doc/source/conf.py | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 don't require the pandas theme

(TODO - package it - here as a multi-upstream-tarball package?)

privacy.patch | (download)

doc/source/development/meeting.rst | 6 1 + 5 - 0 !
1 file changed, 1 insertion(+), 5 deletions(-)

 link to rather than embed google calendar

find_test_data.patch | (download)

pandas/conftest.py | 8 6 + 2 - 0 !
pandas/tests/util/test_util.py | 1 1 + 0 - 0 !
2 files changed, 7 insertions(+), 2 deletions(-)

 allow tests to use the data files in the source tree

We don't ship these in the package,
but do want to run the tests that use them

stable_test_urls.patch | (download)

pandas/tests/io/excel/test_readers.py | 2 1 + 1 - 0 !
pandas/tests/io/parser/test_common.py | 2 1 + 1 - 0 !
pandas/tests/io/parser/test_network.py | 2 1 + 1 - 0 !
pandas/tests/io/test_html.py | 2 1 + 1 - 0 !
4 files changed, 4 insertions(+), 4 deletions(-)

 use test urls that are less likely to disappear

Avoid 404 errors in stable when upstream reorganize the test data
(happened to two of these in 0.25 -> 1.0).

It is _not_ necessary to update the tag version on every package release,
only if these tests fail because they expect moved/changed data.

remove_ccbysa_snippets.patch | (download)

doc/source/user_guide/cookbook.rst | 22 22 + 0 - 0 !
doc/source/user_guide/index.rst | 1 0 + 1 - 0 !
pandas/io/sql.py | 10 5 + 5 - 0 !
pandas/tests/groupby/aggregate/test_other.py | 24 0 + 24 - 0 !
pandas/tests/groupby/test_categorical.py | 23 0 + 23 - 0 !
pandas/tests/indexing/multiindex/test_chaining_and_caching.py | 22 0 + 22 - 0 !
pandas/tests/indexing/multiindex/test_setitem.py | 28 0 + 28 - 0 !
pandas/tests/indexing/test_chaining_and_caching.py | 8 0 + 8 - 0 !
pandas/tests/io/parser/test_common.py | 18 0 + 18 - 0 !
9 files changed, 27 insertions(+), 129 deletions(-)

 remove code from stack overflow

Stack Overflow content is CC-BY-SA licensed,
which this package is not supposed to be.  These snippets may be
too small to be copyrightable, but removing them to be safe.

https://lists.debian.org/debian-legal/2020/04/threads.html#00018

numba_fail_32bit.patch | (download)

pandas/core/window/numba_.py | 6 6 + 0 - 0 !
pandas/tests/groupby/transform/test_numba.py | 4 4 + 0 - 0 !
pandas/tests/window/conftest.py | 3 2 + 1 - 0 !
pandas/tests/window/test_numba.py | 11 11 + 0 - 0 !
4 files changed, 23 insertions(+), 1 deletion(-)

 allow some numba errors on non-amd64, warn on non-x86

Specifying the exception type allows only explicit errors,
not silently wrong answers

Numba has been observed to give wrong answers on mipsel,
and crash on armel (LLVM ERROR) and s390x (segfault).

hurd_compat.patch | (download)

pandas/tests/io/parser/test_common.py | 2 1 + 1 - 0 !
pandas/tests/io/parser/test_multi_thread.py | 7 6 + 1 - 0 !
pandas/tests/io/test_common.py | 4 2 + 2 - 0 !
pandas/tests/test_downstream.py | 5 5 + 0 - 0 !
4 files changed, 14 insertions(+), 4 deletions(-)

 avoid test failures on hurd

Allow multiprocessing to be unavailable
Accept any errno not just 2 for (intentionally) nonexistent files
(Hurd appears to use 2**30+2)

matplotlib33_compat.patch | (download)

pandas/tests/plotting/test_datetimelike.py | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 matplotlib 3.3 compatibility fixup

968208_rounding.patch | (download)

pandas/tests/window/moments/test_moments_rolling.py | 7 6 + 1 - 0 !
1 file changed, 6 insertions(+), 1 deletion(-)

 don't fail test for rounding difference on i386

fsspec_version.patch | (download)

pandas/tests/io/test_fsspec.py | 2 1 + 1 - 0 !
pandas/tests/io/test_pickle.py | 2 1 + 1 - 0 !
2 files changed, 2 insertions(+), 2 deletions(-)

 skip rather than fail tests if fsspec is too old

(which it currently is in Debian)

35080_plot_tests.patch | (download)

pandas/tests/plotting/test_converter.py | 5 4 + 1 - 0 !
1 file changed, 4 insertions(+), 1 deletion(-)

 clean up so test_ts_plot_with_tz doesn't fail

skip_test_missing_required_dependency.patch | (download)

pandas/tests/test_downstream.py | 1 1 + 0 - 0 !
1 file changed, 1 insertion(+)

 skip test that fails from outside the source tree

python39_compat.patch | (download)

pandas/core/computation/pytables.py | 4 4 + 0 - 0 !
1 file changed, 4 insertions(+)

 fix pytables expressions with python 3.9

ast.Index has been replaced by a bare value, so we need to do the
conversion from Constant to int

xlrd_976620.patch | (download)

doc/source/whatsnew/v1.1.5.rst | 10 10 + 0 - 0 !
pandas/io/excel/_base.py | 95 91 + 4 - 0 !
pandas/tests/io/excel/test_readers.py | 7 6 + 1 - 0 !
pandas/tests/io/excel/test_writers.py | 9 7 + 2 - 0 !
pandas/tests/io/excel/test_xlrd.py | 46 45 + 1 - 0 !
5 files changed, 159 insertions(+), 8 deletions(-)

 default to openpyxl not xlrd for read_excel

xlrd 1.2 fails if defusedxml (needed for odf) is installed

Bug: https://github.com/pandas-dev/pandas/pull/35029
Bug-Debian: https://bugs.debian.org/976620
i386_time_overflow.patch | (download)

pandas/tests/indexes/datetimes/test_ops.py | 3 2 + 1 - 0 !
1 file changed, 2 insertions(+), 1 deletion(-)

 avoid year 2038 overflow in test

https://tests.reproducible-builds.org/debian/rbuild/unstable/i386/pandas_1.1.4+dfsg-1.rbuild.log.gz

979621_dead_url.patch | (download)

doc/source/user_guide/io.rst | 6 1 + 5 - 0 !
pandas/tests/io/test_html.py | 4 2 + 2 - 0 !
2 files changed, 3 insertions(+), 7 deletions(-)

 stop using now-dead url in tests/examples

jedi018_compat.patch | (download)

pandas/tests/arrays/categorical/test_warnings.py | 2 1 + 1 - 0 !
pandas/tests/frame/test_api.py | 2 1 + 1 - 0 !
pandas/tests/indexes/test_base.py | 2 1 + 1 - 0 !
pandas/tests/resample/test_resampler_grouper.py | 2 1 + 1 - 0 !
pandas/tests/series/test_api.py | 2 1 + 1 - 0 !
5 files changed, 5 insertions(+), 5 deletions(-)

 don't require external warnings that no longer exist

(The cause is not actually known to be jedi 0.18, but it moved to a
new upstream version since the last pandas and ipython didn't.)