Resume last program workspace and limit cross-program aggregation to selective scoped views

DEC-0096 - Resume last program workspace and limit cross-program aggregation to selective scoped views

Context

The previous shell model promoted a cross-program Global Today Hub as the default entry for some logged-in learners. The team later clarified that this model optimizes for a minority edge case: most learners focus on one program during a study period, even when they own or occasionally access more than one program.

A default overview page therefore creates disproportionate complexity:

• too many branching cases for active/completed/not-started combinations,
• a pseudo-workspace that competes with real program execution,
• extra cognitive load before learners can resume actual study.

Decision

Logged-in DOL English now uses a program-first resume shell:

• There is no standalone post-login global overview page in the active shell.
• Default entry after login is resolved in this order:
  • valid returnTo -> exact protected target,
  • else last meaningful program workspace,
  • else if restore target is no longer valid -> same program root,
  • else if exactly one valid active program exists -> that program root,
  • else -> onboarding / program choice / discovery.

Last meaningful program workspace means a learner-facing program context that is safe and worth resuming. It must not blindly restore routes such as:

• auth,
• account,
• checkout,
• transient error/success pages,
• expired or invalid deep routes.

Cross-program aggregation policy:

• Cross-program aggregation is no longer a shell mode.
• It is allowed only inside explicitly eligible views where combined data materially helps the learner.
• Confirmed initial use case: schedule / calendar style views.
• Eligible aggregate views must:
  • default to the current program scope first,
  • expose All Programs only as an explicit scope toggle when needed,
  • never pretend that All Programs is a peer program context.

Program context policy:

• Program switcher continues to list real programs only.
• Logged-in deep execution remains one real program at a time.
• Canonical object ownership remains unchanged: every deep object still needs program, object_type, source, and canonical_section.

Decision Value

• Aligns the shell with majority learner behavior instead of optimizing the default path for a rare multi-program case.
• Reduces the number of shell states and edge cases the system must support.
• Shortens time-to-resume because the learner lands directly in a relevant program context.
• Preserves necessary multi-program utility through scoped aggregate views without bringing back a pseudo-workspace.

Rationale

The product should optimize for the behavior that happens most often:

• resume the real program the learner is already focused on,
• keep the shell simple,
• introduce cross-program aggregation only when the view truly benefits from merged data.

This gives DOL English a cleaner long-term navigation model:

• Program Workspace is the default shell reality,
• Deep Object lives under canonical owners,
• Cross-program aggregation survives only as an explicit exception where it adds real value.