Context
While reworking #86 (BinderHub launch option) we decided the launch interface direction is Colab-first: Colab is the launch target QuantEcon actually uses, primarily because it provides GPU access for the lectures that need it. BinderHub was dropped before merging (flaky in practice — see #26, kept open as a demand-driven future request).
That raises the same question for the Private JupyterHub option that ships in the launcher today (LaunchButton.tsx — radio + service-URL input, using the nbgitpuller git-pull URL with the .notebooks repo convention).
Question
Is the Private JupyterHub option used by anyone? If demand is zero, removing it would collapse the launcher from a popover-with-radio-group to a single direct launch-to-Colab button — a real interface simplification, and it would also shrink PLAN.md Phase 2 (the repo/branch generalisation and nested-path items then only need to cover Colab).
Proposed resolution path
- Leave the option in place for now (it works and predates this discussion).
- Gather demand/usage signal from lecture maintainers.
- If unused: remove and simplify the launcher; if used: keep and fold into the Phase 2 generalisation work.
🤖 Generated with Claude Code
Context
While reworking #86 (BinderHub launch option) we decided the launch interface direction is Colab-first: Colab is the launch target QuantEcon actually uses, primarily because it provides GPU access for the lectures that need it. BinderHub was dropped before merging (flaky in practice — see #26, kept open as a demand-driven future request).
That raises the same question for the Private JupyterHub option that ships in the launcher today (
LaunchButton.tsx— radio + service-URL input, using the nbgitpullergit-pullURL with the.notebooksrepo convention).Question
Is the Private JupyterHub option used by anyone? If demand is zero, removing it would collapse the launcher from a popover-with-radio-group to a single direct launch-to-Colab button — a real interface simplification, and it would also shrink PLAN.md Phase 2 (the repo/branch generalisation and nested-path items then only need to cover Colab).
Proposed resolution path
🤖 Generated with Claude Code