Based on the paper: DBSP: Automatic Incremental View Maintenance for Rich Query Languages.
pip install steppinggit clone git@github.com:leontrolski/stepping.git
python -m venv .env
source .env/bin/activate
pip install -e '.[dev]'
pytest
mypy src tests- Implement
ZSetPythonas rust btree. - In
interleave_changes, use indexedZSetPythonunder the hood. - Skip out the middle man with
Groupedand just use an indexedZSet? Does this lead to performance benefits inst.group? - Instead of
create_tables=Truecan we emit SQL and write it. - Revisit
st.compile(...). - Can
run.iterationmake nicer error messages? - Python 3.12. Use built in
batched. Can we use the newUnpacksyntax for nicer action types? - Look at 11.8 "Window aggregates"
- Replace
integrate_delaywith a nice transform. Similarly, transform shared delays. - Decide whether to make
...SQLclasses protocols. - Test arbitrary depth grouped nesting and joining in a grouped setting (Does this even make sense to do?).
- Fork
ormsgpack- work out how versioning gubbins works. - Make dump extra quick - see https://github.com/gukoff/dtparse.
- Make rust btree wrapper that implements:
OrderedMap[tuple[bytes | int | float, ...], bytes]
ascending: tuple[bool, ...]
# bump version
python -m pip install build twine
python -m build
twine check dist/*
twine upload dist/*flyctl launch
flyctl deploy
flyctl ips list -a stepping-docs
# set A record to @, IPv4
# set AAAA record to @, IPv6
flyctl certs create -a stepping-docs stepping.site
cd docs/doks
npm install
npm run start
npm run build
cd ..; flyctl deploy; cd -
cd docs
python ../scripts/md.py ../ $(find -L ../docs-md -name '**.md')