Repository navigation
Epic: Website compute surfaces (#493) - #511
neuromechanist wants to merge 17 commits into
Conversation
* Capture compute access fixtures from the backend * Parse compute access onto the session user * Transcribe the compute access copy * Add the compute gap block and affiliation row * Derive compute state and add the request client * Add refusal renderer, compute flag, docs paths, personas * Add the Settings compute access card * Add local backend recipe and ADR 0030 * Bind the local backend to loopback and guard it * Open the terms link in a new tab * Fix the terms fallback and empty refusal lists * Record the D2 hand-off in ADR 0030
* Mirror the compute read vocabulary * Capture compute read fixtures * Add SU and time formatters * Add usage chart geometry * Add compute job state badges * Add the compute read client * Add compute page copy keys and Settings link * Add compute view models and labels * Add chart color tokens * Route /compute to the app host * Add compute badge, panel and notice components * Add access, balances and queue panels * Add the usage chart panel * Add the jobs table with Load more * Add the compute page * Link compute from the user menu * Guard the compute page wiring * Offer one way back after a bad cursor * Seed compute data in the local backend * Record the compute dashboard phase * Fix the notice prop type * Rename SU formats, share contract extractors * Separate the chart series by lightness too * Keep a chart stack inside the plot * Use American spelling in compute text * Leave modified clicks on Load more alone * Tighten usage states, caption unit and labels * Mirror the reworded compute copy * Soften the compute access lede
* Capture admin compute fixtures from the backend * Mirror and drift-test the compute admin contract * Add formatSu and job state badges * Add the admin compute API client * Add admin compute view models * Transcribe the compute_admin copy keys * Add the compute decide flow * Add the Compute admin tab and badge * Add quarter and gateway view helpers * Add the admin Compute page and request rows * Guard the admin Compute tab sources * Seed the admin Compute world in the local backend * Trim dialog labels and keep flags on one line * Document the admin Compute tab * Spell the compute copy keys out for the consumer rule * Reword compute_admin copy and add error keys * Give decisions a 25 s deadline and move sentences to copy * Tell the admin to check email after a lost decision answer * Uncache the Compute page, hand the opener to the dialog, say when cut off * American spelling, escape unknown job states, order CI notes * Widen the Compute source scan to the libs * Update the copy key count in the handoff * Use E2's job state and source lists, fix the shared drift extractors * Read errors with the shared compute reader * Use jobBadge, keep admin libs out of the dashboard scan * Note what the admin tab shares with E2
* Capture the compute CLI help as fixtures * Add the compute dialog instruction builder * Pin the compute dialog to the CLI help * Replace the dataset page Compute dialog * Guard the Compute dialog wiring * Top-align the multi-line copy button * Format the Compute dialog tests * Document the Compute dialog (ADR 0030, map) * Carry the page fragment through sign-in * Lead the access step with the shared lede * Scan every rendered dialog string * Refresh the captured CLI help
* Capture job page fixtures from the backend * Transcribe job page vocabulary with drift test * Add job page clients to the compute API * Make job logs safe to show * Word the job failure codes * Share one charge rule * Add job page view models and polling rules * Forward the truncated header; skip cache for compute * Add the compute job page * Fix local time wording and back link target * Document the job page * Give job read and cancel longer deadlines * Test the held result links as a pure state * Write out direction marks in job logs * Scan job files for navigation and beacons * Fix job page review findings in the browser code * Update job page docs after review
* Update admin badge and card after a decision * Place empty chart text between gridlines * Hide empty Actions card and stray download line * Backdate seeded job records to their start * Fix doubled period in decision announcement
…epic-compute-web # Conflicts: # .context/decisions/README.md # .context/handoff.md
Deploying nemar-website with
|
| Latest commit: |
0e06e24
|
| Status: | ✅ Deploy successful! |
| Preview URL: | https://8c43b36c.nemar-website.pages.dev |
| Branch Preview URL: | https://feature-issue-493-epic-compu.nemar-website.pages.dev |
Also put the drift-test numbering comments in order.
|
Review fixes pushed to this branch (
Gates: |
Mirror nemar-cli's COMPUTE_TERMS_PUBLISHED in the gap table, tag Affiliation only where requests are open, and drift-test the constant.
|
Follow-up pushed: no compute nudge from the site's own gap table while the compute terms are unpublished (owner's ruling; nemar-cli
Deliberately not changed, please say if the owner wants it: the inert Settings card ("not open yet", state Gates: lint and typecheck clean; |
Epic #493: the website's compute surfaces. Draft until nemar-cli's Epic D (nemarOrg/nemar-cli#2116) is on
dev; do not merge before it.What lands
stagingis merged in (ADR 0031 and the authorization-request form exemption; both ADR index lines and both handoff entries kept).Gates run on this head
bun run lint,bun run typecheck(0 errors): pass.vitest: 5,470 pass, 0 fail, 0 skipped when the drift tests read the nemar-cli Epic D checkout (the same file mapping CI uses); without a checkout, 5,436 pass and the 5 drift files skip.Order and the red checks
dev, so they are red here until Epic D (#2116) merges intodev. In the other direction, nemar-cli's account-copy parity test is red on #2116 until this epic's keys are on the website. That cycle is why #2116 carries the one-time parity exception and merges first.GET /compute/me; a site talking to an API without it shows one "not available" notice (ADR 0030).