- Fix test outcome reported in the replay file (#64).
- Dropped support for Python 3.6 and 3.7 (EOL).
- Added official support for Python 3.10, 3.11 and 3.12.
- Test execution order using
--replay
now follows the recorded order, not the collection order, as was always intended (#52).
- Introduce new
--replay-skip-cleanup
option that skips the cleanup before running the command. This allows to keep previously generated replay files when running new commands.
- Replay files can now contain comments (
#
or//
), to make it easy to comment out tests from them when trying to narrow the tests to find a culprit.
- Add proper support when running with
xdist
in a frozen executable.
- Change the format of the output to be able to add more information. The new output has new information such as
start time, end time, outcome and the node identification, all these data is represented by each line being a
json
format.
- Introduce new
--replay-base-name
option that lets users configure a different name of the replay file. Defaults to.pytest-replay
.
- Drop support for Python 2.
- Normal runs and
xdist
runs no longer clean up each other's files.
- Fix crash
IOError
when tests changed the current working directory in the middle of the testing session.
- Replace the shell scripts by plain text files and add new
--replay
flag which accepts the generated files to run the tests.
- Escape node ids in the generated shell scripts.
- Initial release.