- update to 0.9.2:
* Added GitHub Actions integration. Tests and
coverage are run in the cloud after every checkin.
Thanks to [Dan Pope](https://github.com/lordmauve)
for gently walking me through this!
* Fixed metadata in the `pyproject.toml` file.
* Dropped support for Python 3.5. (I assumed I already
had, but it was still listed as being supported
in the project metadata.)
* Added badges for testing, coverage,
and supported Python versions.
* API change: the `Parser` attribute `breadcrumbs` has been
renamed to `stack`. It was previously undocumented anyway,
though as of 0.9.1 it's now documented. The previous name
`breadcrumbs` has been kept as an alias for now, but will
be removed before 1.0.
* Added the `line_number` and `source` attributes to the
`Parser` object, for the convenience of pragma handlers.
* Refactored `parser_include` slightly. No change to
functionality or behavior, just a small code cleanup pass.
* Added a "lines per second" output metric to the
benchmark program.
* From this point forward, Perky only supports reading and
writing files in
[UTF-8](https://en.wikipedia.org/wiki/UTF-8).
If you need to work with a different encoding, you'll have
to handle loading it form and saving it to disk yourself.
You'll have to use `loads` and `dumps` to handle converting
between Perky string format and native Python objects.
* Optimized Perky some more. It's roughly 11% faster than 0.8.1.