Conversation
The not-found placeholder waited 8000ms before rendering, leaving a blank content area for 8 full seconds on any nonexistent or slow-to-load entity (/team/99999, /event/2026zzzzz, etc.). The debounce exists only to avoid a flash of the not-found message before data arrives; 1.5s is ample for that while no longer looking like a hung page. (The comment already described the intent as 'one second'; the 8000 value grew from 1000 over prior commits.)
On team pages multiple components concurrently request the same event/blob resources. Each caller awaits getWithExpiry(), all miss IndexedDB (nothing is written until a fetch completes), and all issue their own network fetch, so every blob/API resource was downloaded twice per page load. Add a module-level in-flight promise map keyed by storageKey: concurrent callers for the same key now share one fetch, and the entry is cleared once it settles.
CockroachDB orders NULLs first under ORDER BY ... DESC, so a match whose clean score is null on both alliances (a fully-DQed / placeholder match, e.g. 2026txmca_sf6m1) sorted to the top of every noteworthy list -- ranking a 0/null result as the highest clean score. greatest()/sum() over the no_foul columns yields null when both alliances lack a clean result. Add .nullslast() to every noteworthy order_by so these matches fall to the bottom (out of the top 30) and real high-scoring matches rank first, while legitimate single-DQ matches keep ranking by their scoring alliance.
chondl
changed the title
Owner
Author
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Three fixes from an adversarial QA pass on staging. One commit per fix, no tests on this branch. Verified against a local rig (full 2026 seed, backend on :8000).
1. Not-found page blank for 8 seconds
Symptom: any nonexistent or slow-to-load entity (
/team/99999,/event/2026zzzzz, ...) showed a blank content area for 8 full seconds before the "not found" message — looks like a hung page.Root cause:
frontend/src/pagesContent/shared/notFound.tsx:14—setTimeout(..., 8000). The debounce exists only to avoid flashing the not-found message before data arrives; the in-file comment describes it as "one second". The value grew 1000 → 5000 → 8000 over prior upstream commits, so 8000 is deliberate rather than a typo — but 8s is far longer than the anti-flash intent needs.Fix: reduce to 1500ms. Enough to suppress the flash, short enough to no longer read as broken.
2. Every blob/API resource fetched twice on team pages
Symptom: on
/team/*, each event/team blob was downloaded exactly twice per page load (observed on the production build, so not a StrictMode artifact).Root cause:
frontend/src/api/storage.tsxquery()had no in-flight dedup. Several components request the same resource concurrently; each awaitsgetWithExpiry(), all miss IndexedDB (nothing is written until a fetch resolves), and all issue their own fetch.Fix: add a module-level in-flight promise map keyed by
storageKey. Concurrent callers for the same key share one fetch; the entry clears once it settles. The fetch body is extracted intofetchAndStoreunchanged.3. Noteworthy matches rank a null/placeholder match #1
Symptom:
/v3/site/noteworthy_matches/2026returned a 0/null-score, fully-DQed placeholder match (2026txmca_sf6m1) at the top of "Highest Clean Scores" (and the other lists), ahead of real high scores.Root cause:
backend/src/db/functions/noteworthy_matches.py— the listsorder_by(desc(...))without specifying null placement. CockroachDB orders NULLs first underDESC. For 2016+ the sort uses theno_foulcolumns; a match with no clean result on either alliance yieldsgreatest(...) = NULL(andsum = NULLfor combined), so it sorts to #1. (The value only surfaces on a NULLS-FIRST-defaulting DB, which is why it appears on staging.)Fix: add
.nullslast()to every noteworthyorder_by. Null-result matches fall past the top-30 cutoff; real high-scoring matches rank first; legitimate single-alliance-DQ matches still rank by their scoring alliance.Verification (rig, before/after): simulating staging's NULLS-FIRST ordering,
2026txmca_sf6m1/_sf9m1ranked #1–#2 above the real 964-point match; with the patched query,2026dal_f1m1(964) is #1 across all six lists and the placeholder is gone.yarn lintclean; backendflake8/black/isortclean.Cross-track note (only if the blob-store stack is also taken)
Fix #2 above (
storage.tsxin-flight dedup) overlaps with Track 2 PR#2(bucket-first serving), which rewritesstorage.tsxfor bucket-first serving. Merging both produces one conflict infrontend/src/api/storage.tsx. Resolution: take #2's rewrite — itsbucketInFlightdedup (keyed by logical path) already eliminates the double blob fetch this PR targets. If you also want this PR's query-level dedup for the API/IndexedDB path, re-wrap #2'squery()in theinFlight[storageKey]pattern from this PR. If you take only Track 1, there is no conflict.