Project Meeting 2022.09.08

Agenda

Jeff to make a first pass at proposed changes to the governance documentation to better reflect what we do now. This will be brought to the consortium to decide whether or not what we do now is the best path forward and that the document should be updated.
Jeff to schedule a consultant bench meeting to discuss who is available to code review Task 8: Flexible Number of Tours and Stops pull request.

PR #592
A variety of non-core changes have been made to modernize the package's supporting infrastructure:
- setuptools: setup.py ▶️ pyproject.toml
- version numbering: bumpversion (manual script) ▶️ setuptools_scm (automatic based on git tags with intermediate values)
- continuous integration testing: travis ▶️ github actions (parallel testing, each test can run longer than 1 hour, free)
- docs: sphinx_rtd_theme ▶️ pydata_sphinx_theme (adds versioning to docs)
- add prototype_mwcog example
A question was raised about how we define a new release/version of ActivitySim.
- There are technical steps required to release an official version, which is documented here:
- On the administrative side, project management needs to approve a release and put it to a vote for the consortium. The process is documented in the governance documentation. However, this document isn't always consistent with how we actually work. At some point, it would be worth reviewing this document and updating as needed, to reflect real and practical process that we have or would like to implement.

PR #599
Code runs. 2 zone systems work great. 3 zone systems run but need additional debugging to validate results.
Currently does not pass the prototype_sandag_xborder tests. Code runs without crash but results do not match the non-sharrow outputs. However, in diagnosing this problem, it was discovered that this model is unstable in prior versions. The older non-sharrow code gives different outcomes when the model is run with different numbers of processes. A new issue has been opened for this pre-existing bug.

PR by Sept 30th
RSG to make preliminary review of runtime performance
RSG hopes to have more results to share by next Thursday, but will confirm by Wednesday if they have generated enough results to share.

PR #581
Extended to 4 outbound and 3 inbound stops
Modified trip scheduling probabilities for additional stops/trips after extension
Nobody assigned to review. ACTION ITEM: Jeff to put together a group meeting to discuss who has available resources.

PR #600 Windows installer, contributed by PTV
PR #577 Fix to control reproducible random seeding correctly
PR #584 Dependabot warning of a security issue with numpy < 1.22
Various issues RSG has uncovered with using estimation mode. #591 #595 #596 #597