RLS: 1.3.4

[simonjayhawkins]

Tracking issue for the 1.3.4 release.

Currently scheduled for October 17, 2021. (could release earlier if there are issues with the pandas 1.3.3 py3.10 wheels once Python 3.10.0 final is released xref #41940 (comment))

https://github.com/pandas-dev/pandas/milestone/90

List of open regressions: https://github.com/pandas-dev/pandas/issues?q=is%3Aopen+is%3Aissue+label%3ARegression

performance regressions between 1.2.5 and 1.3.x https://simonjayhawkins.github.io/fantastic-dollop/#regressions?sort=3&dir=desc&branch=1.3.x (significant ones have now been addressed)

cc @pandas-dev/pandas-core @pandas-dev/pandas-triage

[jbrockmendel]

performance regressions between 1.2.5 and 1.3.x

Some of these are pretty clearly false-positives. if I hit the "ignore" button in my browser, does that do something on the back-end to mark it for everybody?

[simonjayhawkins]

if you hit the ignore it gives a code snippet at the bottom in the show ignore regressions section to include in the config. If you paste it here, i'll add it to the config.

[jbrockmendel]

"regressions_first_commits": {
    "^tslibs\\.normalize\\.Normalize\\.time_normalize_i8_timestamps\\(10000, None\\)@1\\.3\\.x$": "1c38dfaa",
    "^tslibs\\.normalize\\.Normalize\\.time_normalize_i8_timestamps\\(10000, datetime\\.timezone\\(datetime\\.timedelta\\(seconds=3600\\)\\)\\)@1\\.3\\.x$": "1c38dfaa",
    "^tslibs\\.normalize\\.Normalize\\.time_normalize_i8_timestamps\\(10000, datetime\\.timezone\\.utc\\)@1\\.3\\.x$": "1c38dfaa",
    "^tslibs\\.normalize\\.Normalize\\.time_normalize_i8_timestamps\\(1000000, datetime\\.timezone\\(datetime\\.timedelta\\(seconds=3600\\)\\)\\)@1\\.3\\.x$": "1c38dfaa",
    "^tslibs\\.normalize\\.Normalize\\.time_normalize_i8_timestamps\\(1000000, datetime\\.timezone\\.utc\\)@1\\.3\\.x$": "1c38dfaa",
    "^tslibs\\.period\\.TimePeriodArrToDT64Arr\\.time_periodarray_to_dt64arr\\(1, 10000\\)@1\\.3\\.x$": "3dfbf496",
    "^tslibs\\.period\\.TimePeriodArrToDT64Arr\\.time_periodarray_to_dt64arr\\(100, 6000\\)@1\\.3\\.x$": "3dfbf496",
    "^tslibs\\.tz_convert\\.TimeTZConvert\\.time_tz_convert_from_utc\\(1, datetime\\.timezone\\.utc\\)@1\\.3\\.x$": "e0c57672",
    "^tslibs\\.tz_convert\\.TimeTZConvert\\.time_tz_convert_from_utc\\(100, datetime\\.timezone\\.utc\\)@1\\.3\\.x$": "e0c57672",
}

[lithomas1]

MacPython/pandas-wheels#157 needs to be merged before release.

[simonjayhawkins]

MacPython/pandas-wheels#157 needs to be merged before release.

merged. Thanks @lithomas1

[simonjayhawkins]

As discussed in the dev meeting this week, we will release after #43291 is merged which is moreless ready so planning to release tomorrow, assuming #43291 is merge later today.

[simonjayhawkins]

closing off milestone to new additions now. will start pre-release checks after #44066 is merged.

[jreback]

merged

[simonjayhawkins]

pre-release checklist

• wheels building test wheel build status of 1.3.x branch MacPython/pandas-wheels#135
• release readiness passing https://github.com/simonjayhawkins/pandas-release/actions?query=workflow%3A%22Tag+Release%22
• release notes checked https://pandas.pydata.org/pandas-docs/dev/whatsnew/v1.3.4.html
• ci checks https://github.com/pandas-dev/pandas/commits/1.3.x
• no manual backports outstanding https://github.com/pandas-dev/pandas/pulls?q=is%3Apr+label%3A%22Still+Needs+Manual+Backport%22+is%3Aclosed
• no open issues/PRs (except this one) https://github.com/pandas-dev/pandas/milestone/90

[simonjayhawkins]

• no open issues/PRs (except this one) https://github.com/pandas-dev/pandas/milestone/90

will close #43715 after release.

[simonjayhawkins]

• wheels building test wheel build status of 1.3.x branch MacPython/pandas-wheels#135

known flakey test

=========================== short test summary info ===========================
FAILED test_venv/lib/site-packages/pandas/tests/frame/methods/test_rename.py::TestRename::test_rename_inplace
= 1 failed, 157447 passed, 19595 skipped, 862 xfailed, 10 xpassed, 259 warnings in 1004.09s (0:16:44) =

[simonjayhawkins]

starting release now

• tag locally
• build sdist
• pip test
• conda test
• build docs
• push tag https://github.com/pandas-dev/pandas/tags
• open PR to build wheels RLS v1.3.4 MacPython/pandas-wheels#158
• upload docs https://pandas.pydata.org/pandas-docs/version/1.3.4/
• wheels in staging area https://anaconda.org/multibuild-wheels-staging/pandas/files
• github release - https://github.com/pandas-dev/pandas/releases
• upload wheels to PyPI https://pypi.org/project/pandas/
• change link for stable docs https://pandas.pydata.org/docs/
• start v1.3.5 release notes DOC: Start v1.3.5 release notes #44058
• conda-forge bot picked up on release pandas v1.3.4 conda-forge/pandas-feedstock#121
• available on conda-forge https://anaconda.org/conda-forge/pandas/files
• announce

[simonjayhawkins]

• available on conda-forge https://anaconda.org/conda-forge/pandas/files

except linux_aarch64_numpy1.19python3.7.____73_pypy

last successful build was pandas 1.3.1 (which IIRC required a couple for restarts)

from https://cloud.drone.io/conda-forge/pandas-feedstock the successful builds are much quicker

I recall seeing somewhere that the hardware allocated varies, I think it was sometimes 2 cores and sometimes 4, which made sense as could get the wheels built with some restarts.

Maybe this is no longer the case since did restart both 1.3.2 and 1.3.3 over a couple of days after those releases without success.

(Note that the number of downloads for linux_aarch64_numpy1.19python3.7.____73_pypy is negligible, so maybe not even worth peoples time reading this post)

[jreback]

thank you @simonjayhawkins !

[simonjayhawkins]

except linux_aarch64_numpy1.19python3.7.____73_pypy

this is now built using Azure. xref conda-forge/pandas-feedstock#123