ruvnet f5a003fc9a chore(ci): bump test timeout to 90min + split core-and-rest shard ... PR #389's first CI run after the matrix split exposed two more shards still hitting the 45-min timeout: `core-and-rest` and `Linux Benchmarks (NEON baseline)`. Two changes: 1. Test job timeout 45 → 90 min. Compute-heavy crates with full nextest test suites + doctests can legitimately need an hour; 45 min was set conservatively without measurement. 2. Hoist the known-heavy long-tail crates into a new `core-and-rest-heavy` shard (ruvllm, ruvllm-cli, ruvector-dag, ruvector-nervous-system, ruvector-math, ruvector-consciousness, prime-radiant, mcp-brain, ruvector-decompiler). Existing `core-and-rest` continues with `--workspace --exclude` for everything else; just adds these to the exclusion list. Result: 8 test shards instead of 6, each well under the 90-min cap. macOS / Linux benchmark cancellations are env-flaky and unrelated; tracking those is a separate follow-up. Co-Authored-By: claude-flow <ruv@ruv.net>		2026-04-26 01:06:25 -04:00
..
benchmarks	feat: Add Neo4j-compatible hypergraph database package (ruvector-graph)	2025-11-25 23:11:54 +00:00
workflows	chore(ci): bump test timeout to 90min + split core-and-rest shard	2026-04-26 01:06:25 -04:00