Repository navigation
2026-04-02 Engineering Team #86
Description
Agenda
@ActivitySim/engineering
- Remove 3-Zone PR #1056 -- who will finish?
- Requires small changes to example models, and final work to clean up main repo
- Park and Ride Lot Choice #1001 -- in review
- @jpn--: #1027 which does not actually fix PSRC's non-alignment issue #1022
- Delete temporary vars #1062 needs reviewer, looks like maybe needs docs or tests?
- Adding ignore list to skim load for unused skims while running with sharrow #1036 @jpn-- to review
- PopulationSim geographies #217 -- will get to this eventually, is it a priority?
- Other topics...
Activity
jpn-- commentedon Apr 2, 2026
Meeting Overview
Agenda
The meeting covered a series of open pull requests and development priorities for ActivitySim, including the three-zone removal PR, park and ride lot choice, PRC non-alignment bug, a summarize script fix, skim loading ignore list, PopulationSim integration, desired accessibility speedup, and a broader discussion on testing strategy.
Participants
Duration
Approximately 18 minutes
Executive Summary
The team reviewed a backlog of open pull requests for ActivitySim, establishing sequencing priorities and assigning reviewers. A key decision was made to complete the three-zone removal PR before merging the park and ride lot choice PR, since the latter is failing three-zone-related tests. The group also had a productive discussion on incrementally building out the test suite as new features are contributed, with agreement that engineering team members should help external contributors add appropriate tests.
Meeting Notes
Three-Zone Removal PR
David Hensle has an open draft PR to remove all three-zone code. Jeff Newman identified a failing test caused by a data file that was removed as part of the three-zone cleanup but may need to be present for the two-zone tests as well. There is also a network settings entry in the SANDAG example that references a no-op setting which should be removed. David confirmed he can address both issues and complete the PR.
Park and Ride Lot Choice PR
Jeff Newman has begun reviewing the park and ride lot choice PR. Because this PR is failing some three-zone tests, the team agreed to complete the three-zone removal first and then proceed with the park and ride review. David noted the tests are extensive and suggested Jeff could do a high-level structural review of the park and ride PR even before the three-zone work is merged.
PRC Non-Alignment Issue
Jeff Newman opened a PR intending to fix the PSRC non-alignment issue but the fix did not work. He identified this as his current highest-priority item.
Summarize Script Fix (Delete Temporary Variables)
Sijia Wang clarified that this PR was submitted by Andrea from WSP Australia as part of work on the Sydney ActivitySim model. The PR makes a small change to the summarize step. Jeff suggested that the review is not large but that the appropriate testing approach and documentation should be worked out between Sijia and Andrea. The appropriate test would verify that Python references to the variables are deleted, allowing garbage collection. Jeff marked Sijia as a reviewer.
Skim Loading Ignore List PR
Jeff noted this PR is straightforward and that he just needs to complete his review.
PopulationSim Integration
Jeff noted there is an open request to work with Joe Flood (@JoeJimFlood) on PopulationSim integration. Nick Fournier has already reviewed the PR, made some revisions in a PR against Joe's branch, and notified him, but has not heard back. The team agreed this does not need to jump the queue unless Joe follows up. Jeff asked Nick to ping Joe to see if they can move it forward.
Explicit Error Testing and Documentation PR
Jan Zill mentioned he would be opening a PR related to explicit error testing and documentation immediately after the meeting.
Desired Accessibility Speedup (PR #1031)
David flagged PR #1031 for Jeff to review, noting it is small and straightforward. Jeff confirmed the tests he re-ran that morning appeared to be passing, with the only prior failures having been in the external examples.
Memory Leak / Recompilation Issue in Sharrow
David sent Jeff an email with details on why models were recompiling in Sharrow, following up on a discussion from a previous meeting. Jeff had not yet read it and agreed to follow up.
Testing Strategy Discussion
David observed that the team is implicitly moving toward writing tests for all new contributions, and raised the question of whether this should be formalized. He cited the skim loading work and Andrea's summarize PR as examples of this emerging pattern. Jeff agreed the direction makes sense and that it should be formalized, but that incrementally adding tests alongside development is preferable to a large-scale retroactive effort. Jeff also noted that support funds can be used to help write tests for externally contributed features, and committed to reviewing the prior testing plan discussion and laying out a more concrete plan in the coming weeks.
Action Items
JoeJimFlood commentedon Apr 2, 2026
@nick-fournier I had commented on your pull request letting you know that it didn't seem to fix the issue I was having.