summaryrefslogtreecommitdiff
path: root/igor.py
Commit message (Collapse)AuthorAgeFilesLines
* build: no need for check_eol, pylint does most of itNed Batchelder2023-01-301-72/+0
|
* test: more flexibility in setting a static context locallyNed Batchelder2023-01-241-0/+2
|
* build: igor tweaks for scriv start markerNed Batchelder2023-01-221-3/+5
|
* build: more-correct version bumpingNed Batchelder2022-12-031-5/+8
|
* style: fix spellingNed Batchelder2022-11-281-4/+4
| | | | un-executed, white space, time stamp.
* build: cheats for commit-specific installsNed Batchelder2022-11-201-1/+7
|
* build: allow for .devN version numbersNed Batchelder2022-11-171-2/+11
|
* test: simplify the metacov flow a bitNed Batchelder2022-11-081-4/+1
| | | | | | | - default COVERAGE_METAFILE=.metacov for convenience - don't make json and xml reports, if we need them we'll do it explicitly.
* test: no need to combine during each test run, do it at the endNed Batchelder2022-11-081-2/+0
|
* build: automate release-related editsNed Batchelder2022-09-301-43/+83
|
* build: more low-tech release automationNed Batchelder2022-06-021-6/+18
|
* build: always use three-component version numbersNed Batchelder2022-05-301-5/+2
| | | | | Someone thought we didn't use semantic versioning because we said "6.4" instead of "6.4.0". Don't trim .0.
* build: more make targets for release stepsNed Batchelder2022-05-221-3/+0
|
* build: Build and test tweaks to help with cross-compilationSteve Dower2022-04-301-2/+16
| | | | | | | | | | | | | | | For context, I've been testing whether a range of popular libraries are going to work on Windows ARM64. (This requires compiling on a regular x64 machine and then copying the wheel to an ARM64 one for testing.) The good news is, coverage seems to be just fine without any changes. However, because of a few assumptions in the test suite about always testing an in-place build, I had to make some tweaks to be able to run tests. My proposed tweaks should be fine for current uses, but they also allow the following: * support `SETUPTOOLS_EXT_SUFFIX` when building to override the pyd tag on Windows (used with `setup.py build_ext -L <path>` to point at [pythonarm64](https://www.nuget.org/packages/pythonarm64/) import libraries to do the cross-compile) * allow `COVERAGE_INSTALL_ARGS` to override how the tests install coverage into a venv (allows to point at a wheel rather than rebuilding from the sources) * allow `python igor.py remove_extension --from-install` to delete the extension module from `site-packages` rather than only the source tree * other changes to allow removing the `coverage` directory from the source tree before tests so that the installed copy will be used instead. I've tested these on my own Windows ARM64 machine, though unfortunately there aren't any available on CI services yet. If you wanted to start releasing (preview) wheels for win-arm64 you can, but there's no support (yet) in `cibuildwheel` or `build` to do it (because those tools don't really have a concept of cross-compilation for Windows at all... yet ;-) ). For the linter
* build: version bump and more release detailsNed Batchelder2022-02-011-2/+2
|
* build: prep for 6.3.16.3.1Ned Batchelder2022-02-011-1/+8
|
* build: 'python igor.py cheats' shows a few things needed during releasesNed Batchelder2022-01-261-0/+40
|
* debug: why can't we clean up the C extension on windows?Ned Batchelder2021-11-211-4/+10
|
* style: convert more string formatting to f-stringsNed Batchelder2021-11-111-3/+3
|
* test: when we measure our coverage, get a json report alsoNed Batchelder2021-10-311-0/+1
|
* build: force 'build' to shut upNed Batchelder2021-10-261-0/+7
|
* style: use the official designation for utf-8Ned Batchelder2021-10-101-1/+1
| | | | | Yes, this is completely unimportant. Don't ask me why I bothered, I'm not really sure.
* test: make metacov work with xdistNed Batchelder2021-10-081-1/+2
| | | | | | | | | I was using pytest.__file__ as a directory, but in 5.3.2, pytest changed to be a package, so that wasn't the site-packages directory anymore, and our .pth file was written someplace useless. Now we don't rely on the structure of pytest (why did we ever?), and it works again.
* test: show contexts in metacov reportNed Batchelder2021-08-091-1/+4
|
* style: fix typos discovered by codespell (#1197)Christian Clauss2021-08-011-1/+1
| | | | | | | | | | | python3 -m pip install codespell codespell --ignore-words-list="ba,cant,datas,hart,linke,ned,nin,overthere,upto" --skip="*.js" * Fix typos discovered by codespell * datas * intgers ==> integers
* build: update pylint and remove some unneeded warning suppressionNed Batchelder2021-07-181-8/+2
|
* test: change how we keep mismatched actual outputNed Batchelder2021-07-151-3/+0
| | | | | | | | | | | | Now when a goldtest has a failure, the actual mismatched output will be written to the tests/actual directory. Along the way, I removed some obsolete settings which were only used by unittest and unittest_mixins, which we no longer use: - COVERAGE_KEEP_TMP - COVERAGE_ENV_ID - $TMPDIR/coverage_test
* refactor: pyupgrade --py36-plus *.pyNed Batchelder2021-05-021-16/+15
|
* test: improve zipfile testNed Batchelder2021-04-141-35/+35
| | | | | | | Before this commit, the GetZipBytesTest.test_get_encoded_zip_files test was flaky on Python 3.10.0a7. Since I had just added new files to the common zip file, I tried splitting the newly added stuff into its own file, and that seemed to fix the problem.
* test: remove the .egg testNed Batchelder2021-04-141-16/+3
| | | | | | People don't use .egg much anymore, distutils is showing deprecation warnings, and coverage.py only deals with them the same way it deals with .zip files, so let's just rely on a .zip test to cover that.
* build: update pylintNed Batchelder2021-04-111-1/+1
|
* build: update to latest pylintNed Batchelder2021-02-261-1/+1
|
* build: nicer publishing of HTML reportNed Batchelder2021-02-241-1/+5
| | | | Also, this correctly combines results from different runners.
* style: singleton comparisons should use isNed Batchelder2021-01-311-1/+1
| | | | | | I guess the original line was wrong, but it would have been nice for unittest2pytest to fix it for me: https://github.com/pytest-dev/unittest2pytest/issues/52
* COVERAGE_ONE_TRACER runs just one tracerNed Batchelder2021-01-101-3/+21
| | | | It chooses the appropriate tracer based on the implementation.
* Indicate the git revision if running a local buildNed Batchelder2021-01-101-0/+4
|
* Remove unneeded CI and kitting supportNed Batchelder2020-11-281-1/+2
| | | | | | | | | We no longer: - Use Travis - Use AppVeyor - Build manylinux wheels locally - Build other wheels locally
* Don't complain about trivial problems in doc/helpNed Batchelder2020-08-181-0/+1
|
* Fix some checking configurationsNed Batchelder2020-07-041-0/+1
|
* A test of running coverage when it has been zipped. #862Ned Batchelder2020-01-111-0/+4
|
* I'm not sure why I was deleting these meta env varsNed Batchelder2019-12-011-1/+1
| | | | Don't we want to measure the coverage of subprocesses we start in tests?
* Better auto-spelling of contributors' namesNed Batchelder2019-11-111-1/+2
|
* Details of context tracingNed Batchelder2019-11-101-1/+1
|
* Quiet pylintNed Batchelder2019-10-061-1/+0
|
* No need for format indexes (mostly)Ned Batchelder2019-09-011-1/+1
|
* Make it easier for me to measure test_function contextsNed Batchelder2019-06-141-2/+3
|
* Silence a DeprecationWarningNed Batchelder2019-05-091-1/+3
| | | | | | "DeprecationWarning: Use inspect.signature() instead of inspect.getfullargspec()" This started appearing with Python 3.8a4.
* Keep test directories shorterNed Batchelder2018-11-121-1/+3
|
* Keep test tempdirs distinct, for saving and comparing test outputNed Batchelder2018-11-111-7/+12
|
* Debug-time environment variables can be set with set_env.pyNed Batchelder2018-11-031-0/+2
|
View Lorry Controller Status