/
2026-07-30 Frontend Working Group Meeting Notes

2026-07-30 Frontend Working Group Meeting Notes

https://drive.google.com/file/d/1pbMTIfWeuskwRfBQxWPYcsr_x7CLVsGH/view?usp=drive_link All public Working Group meetings follow the Recording Policy for Open edX Meetings

 Time, location

 Agenda

  • @Mary Gwozdz Competency Management: new MFE or not?

  • @Max Frank and Varun Kumar

      1. Datadog related PR raised in openedx frontend-base repo

      2. UTM Params support in frontend-platform

📝Notes

 Decisions

  1. CBE features will be built in frontend-app-authoring, as opposed to a new MFE

 Action items

Previous Action Items

🎥Recording

 Participants

@Jesper Hodge @Adolfo Brandes @Brian Smith @Max Frank @Mary Gwozdz @Sarina Canelake @Marco Morales @Edward Byun @Sam Daitzman @Phillip Ball

🤖 Summary

Frontend Working Group — July 30, 2026

Main topic: where to build competency management (CBE)

Unicon is picking up an extensive competency-management authoring feature, and the question was whether to build it as a new frontend-base app or inside frontend-app-authoring. Adolfo framed it as a rehash of the in-person discussion, where the leaning was to keep it in authoring since it's a core (if optional) feature. Braden had flagged in the issue that a separate app would let the team move faster by avoiding refactoring work, but he couldn't attend and registered "concerns, okay in principle."

Arguments that came up:

  • Brian was initially drawn to the new-app route: authoring is already gigantic and hard to maintain, and this would establish a pattern for breaking pieces out and start the migration at 1-of-21 pages rather than 0%. But he conceded that reusing components across a non-frontend-base app and a frontend-base app would be painful, so a separate app wouldn't actually save refactor time.

  • Jesper pushed back on the premise that a new app is less effort — everything is already set up in authoring, it just needs a page and components. He argued more lines of code isn't the same as more complexity: like taxonomy editing, this would be a self-contained page that could later be lifted out with essentially a copy-paste. More repos is its own maintenance burden.

  • Marco wanted clarity on how interconnected taxonomies and competencies actually are, since without more context he'd have guessed separate app.

  • Mary clarified the MVP: pages fully separate, reusing the entire tagging data model with a competency model alongside it. Course-outline integration isn't in MVP (likely just warnings and delete prevention); Jenna's longer-term 2027 vision is competency criteria in the course-outline sidebar panel.

  • Sam raised the design-side risk: the Figma files contain many near-duplicates of undeclared standards (tagging, filtering, resizable sidebar, hierarchy objects), and shipping without pulling out the underlying components means later having to unify three slightly different interfaces instead of two. His worry: it gets built two and a half times instead of once.

Outcome: no hard objections to building inside frontend-app-authoring, with the middle ground agreed as: build new pieces to be reusable (in src/generic or similar), actively look for refactor opportunities, and file well-documented follow-up issues under a dedicated epic identifying exactly where each new component should replace existing occurrences — but don't do the sweeping refactor now, given Willow's two-to-three-month window and the breakage risk from Redux coupling. Adolfo stressed the issues are a project deliverable, not an optional afterthought.

Related threads: Jesper and Phillip agreed the designs should be triaged into "this is Paragon," "this is a Paragon composition," "this twists Paragon in an unintended way," and "this should become a new Paragon component" — with tickets to port later if not feasible now. Adolfo noted Paragon 24 (breaking changes, landing in Xylon) makes this timing convenient. Brian wants Paragon 24 to draw a hard line at the molecule level, moving organisms like DataTable out into their own installable packages so they can move faster than Paragon's major-version cadence.

Open item: Jesper will circle back with Mary and Jenna on whether Willow is a hard requirement, and bring it back to the group if the release date turns out to be negotiable.

2U items (rushed, five minutes): Max introduced Varun and noted they're working heavily in frontend-base with private apps built on the template. Varun's fix adds the missing type-interface entry so a custom Datadog logging service can be set in site config without a TypeScript error — Adolfo sees no objection, will review formally; as a fix it gets a point release and can be backported to Verawood by bumping the frontend-base version. Second item, via Jason: persist UTM parameters into cookies/local storage on the frontend, mirroring existing edx-platform middleware, with no proprietary code. Adolfo has no objection in principle, on the condition it lands in frontend-base as well as frontend-platform. An issue and discuss-forum post already exist and will be shared in Slack.