Repository navigation
Confirm SUBSIDE can live without the un-ported Lit screens #69
Description
Question
Replacing the Lit UI at mint.tacc.utexas.edu removes these screens for SUBSIDE users, because they do not exist in ui-react and were never in the SOW's scope:
- Analysis, Emulators, Messages
models-compare,models-calibrate,models-cromo- Thread Visualize and Summary steps
model-view(the 3,256-LOC model detail screen)
Max knows the answers here — this is a short conversation, not an outreach to TACC. Resolve it by walking the list and recording, per screen, whether SUBSIDE uses it and what happens when it disappears.
Several are dead or stubbed in the Lit app anyway: Messages lost its backend when Firebase was removed, and several Analysis screens are placeholders. Say so explicitly per screen rather than dismissing the list wholesale — the resolution is the record that T9 flips against.
Outcome is one of: (a) none are needed, flip freely; (b) a specific screen is needed, which either graduates a scoping conversation or forces T9 to keep Lit reachable at a second path.
Activity
mosoriob commentedon Aug 8, 2026
Resolution
Outcome (a): none of the un-ported screens are needed. The flip is free.
Walked the list against both codebases rather than from memory. Two of the ticket's own
premises turned out to be wrong, so the per-screen record below supersedes the question body.
Per screen
ui/src/screens/messages/actions.tsis entirely commented-out Firebase (/* TODO: Add this to graphql), andmessages-post.tsis 0 bytes. No loss.<p>This page is in progress</p>. No loss.ui/src/app/mint-app.ts:350, so it is URL-only today. Max confirms commenting the link out was the retirement. No loss.models-calibrateCYCLESdefault. Leftover from a different project, not SUBSIDE. No loss.models-compare,models-cromoui-react/src/pages/modeling/thread/MintSummary.tsx, wired atMintThread.tsx:309, a 1:1 port. The ticket listed this as missing; it is not.mint-visualize.ts, 320 LOC). But it only iframes URLs fromgetVisualizationURLs, which needs a MINT visualization service SUBSIDE does not have — so it renders empty in Lit too. Parity, not a gap. Not flagged on handover.model-viewConfigurationDetailat/modelconfigurations/:slugid. SUBSIDE users arrive through the configure workflow, not the explorer browse surface, so this covers the SOW. No loss.Consequences for the rest of the map
The conversation also settled the deployment strategy, because outcome (a) removed the only
user-facing reason to stage the rollout:
beside. This invalidates the premise of Smoke-test ui-react on a Vercel preview against TACC's real Hasura and CKAN #76, which is rescoped to a Vercel smoke test.
TACC's Hasura and CKAN; production is then upgraded directly.
mint.tacc.utexas.eduatui-reactand leaves theuiHelm component deployed but off theroute — the only undo available if the client hits something during sign-off. The component
is removed after sign-off, which graduates out of the map's fog as its own ticket.
#77 no longer has to decide whether to keep Lit reachable at a second path: it does not, and
the fallback is an unrouted deployment rather than a second URL.