You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Discussion of Task 11B progress, including trip scheduling choice implementation, test infrastructure improvements, code formatting updates, and UV installation documentation.
The engineering team discussed progress on Task 11B deliverables, with significant focus on testing infrastructure and maintenance policies. RSG completed work on trip scheduling choice with reproducible random numbers, including a new unit test. The team decided to recommend retiring an removing unmaintained example models from the repository while ensuring proper unit tests exist for all unique components. Additional topics covered included transitioning from Black to Ruff for code formatting and updating UV installation documentation based on agency feedback.
Meeting Notes
Trip Scheduling Choice Implementation (RSG)
RSG completed the trip scheduling choice work with reproducible random numbers. The ARC test that was previously disabled has been turned back on and is now passing. The implementation includes a new unit test specifically for the trip scheduling choice component. The team confirmed that ARC's production model does use this trip scheduling component, making this work relevant to consortium members.
Testing Infrastructure and Example Models
The team had an extensive discussion about testing philosophy and maintenance of example models in the repository. Several legacy prototype models (ARC, MWCOG, SEMCOG) remain in the repository but have diverged, sometimes significantly, from agencies' production implementations. These models are contributing to maintenance overhead without providing value.
The team decided to adopt a new testing approach that includes maintaining only two canonical example models (MTC and SANDAG) for integration testing, creating unit tests for individual components rather than maintaining multiple full example models, retiring legacy prototype models that are no longer actively maintained by their respective agencies.
The settings checker identified issues with the prototype ARC example using legacy configuration structures. Rather than supporting multiple configuration approaches indefinitely, the team agreed to deprecate legacy methods and provide migration support to consortium members still using deprecated features.
Stefan Coe suggested creating a linked list of agency implementations on the ActivitySim documentation homepage, allowing agencies to reference each other's work without maintaining outdated examples in the main repository.
Code Formatting Transition
Jeff Newman proposed transitioning from Black (2022 version) to Ruff for code formatting. Ruff is faster, more up-to-date, and can also provide linting capabilities. The team agreed to this transition with configuration to auto-fix formatting issues similar to Black's behavior. Limited linting rules will be enabled initially to catch obvious issues without overwhelming the codebase with warnings.
UV Installation Documentation
Sijia Wang provided an update on UV installation outreach efforts. Three agencies (SANDAG, PSRC, SEMCOG) have successfully adopted UV. Documentation improvements were identified including guidance for global installation across multiple users on the same machine, integration with OpenM++, and clarification of use cases for "uv add activitysim" versus "uv sync."
The team agreed to update documentation first before conducting a second round of agency outreach, allowing more agencies to benefit from improved guidance.
SANDAG Example Model Update
WSP's work to update the SANDAG example model to match SANDAG's production model is substantially complete for the non-Sharrow version and awaits SANDAG review. The Sharrow version compatibility work remains pending. A few minor bugs were identified in the production model during this process.
Action Items
Jeff Newman (@jpn--): Review and merge RSG's trip scheduling choice pull request; create pull request to transition from Black to Ruff for code formatting with appropriate configuration; complete Sharrow version compatibility for SANDAG example model
David Hensle (@dhensle): Turn off ARC test in the trip scheduling choice pull request and flag for review; prepare park and ride discussion topic for next week's meeting
Jeff Newman (@jpn--): Create separate pull request to remove ARC example model from consortium repository
Sijia Wang (@i-am-sijia): Draft documentation updates for UV installation including global installation guidance and use case clarifications for team review
** Jeff Newman (@jpn--)*: Draft document testing philosophy and policies in ActivitySim documentation; create inventory of components requiring unit tests; develop plan to retire remaining legacy prototype models
Product/Community Team: Review and approve decision to retire non-canonical example models before implementation
SANDAG (@bhargavasana or designee): Review WSP's pull request updating SANDAG example model to production model consistency, and decide whether SANDAG should adopt fixes implemented for the consortium back into the SANDAG production model
Jeff Newman (@jpn--): Review CS pull requests from Joe Flood for error definition consolidation and logging cleanup
Activity
jpn-- commentedon Oct 23, 2025
ActivitySim Engineering Meeting Notes
Meeting Overview
Agenda
Discussion of Task 11B progress, including trip scheduling choice implementation, test infrastructure improvements, code formatting updates, and UV installation documentation.
Participants
Duration
63 minutes
Executive Summary
The engineering team discussed progress on Task 11B deliverables, with significant focus on testing infrastructure and maintenance policies. RSG completed work on trip scheduling choice with reproducible random numbers, including a new unit test. The team decided to recommend retiring an removing unmaintained example models from the repository while ensuring proper unit tests exist for all unique components. Additional topics covered included transitioning from Black to Ruff for code formatting and updating UV installation documentation based on agency feedback.
Meeting Notes
Trip Scheduling Choice Implementation (RSG)
RSG completed the trip scheduling choice work with reproducible random numbers. The ARC test that was previously disabled has been turned back on and is now passing. The implementation includes a new unit test specifically for the trip scheduling choice component. The team confirmed that ARC's production model does use this trip scheduling component, making this work relevant to consortium members.
Testing Infrastructure and Example Models
The team had an extensive discussion about testing philosophy and maintenance of example models in the repository. Several legacy prototype models (ARC, MWCOG, SEMCOG) remain in the repository but have diverged, sometimes significantly, from agencies' production implementations. These models are contributing to maintenance overhead without providing value.
The team decided to adopt a new testing approach that includes maintaining only two canonical example models (MTC and SANDAG) for integration testing, creating unit tests for individual components rather than maintaining multiple full example models, retiring legacy prototype models that are no longer actively maintained by their respective agencies.
The settings checker identified issues with the prototype ARC example using legacy configuration structures. Rather than supporting multiple configuration approaches indefinitely, the team agreed to deprecate legacy methods and provide migration support to consortium members still using deprecated features.
Stefan Coe suggested creating a linked list of agency implementations on the ActivitySim documentation homepage, allowing agencies to reference each other's work without maintaining outdated examples in the main repository.
Code Formatting Transition
Jeff Newman proposed transitioning from Black (2022 version) to Ruff for code formatting. Ruff is faster, more up-to-date, and can also provide linting capabilities. The team agreed to this transition with configuration to auto-fix formatting issues similar to Black's behavior. Limited linting rules will be enabled initially to catch obvious issues without overwhelming the codebase with warnings.
UV Installation Documentation
Sijia Wang provided an update on UV installation outreach efforts. Three agencies (SANDAG, PSRC, SEMCOG) have successfully adopted UV. Documentation improvements were identified including guidance for global installation across multiple users on the same machine, integration with OpenM++, and clarification of use cases for "uv add activitysim" versus "uv sync."
The team agreed to update documentation first before conducting a second round of agency outreach, allowing more agencies to benefit from improved guidance.
SANDAG Example Model Update
WSP's work to update the SANDAG example model to match SANDAG's production model is substantially complete for the non-Sharrow version and awaits SANDAG review. The Sharrow version compatibility work remains pending. A few minor bugs were identified in the production model during this process.
Action Items