dangernoodle-io / TaipanMiner / 24937100203

90%

push

github

ci: per-board static check, drop redundant 4x check rounds (#204)

* ci: per-board static check, drop redundant 4x check rounds

Move `pio check --skip-packages` into each matrix shard so it runs once
for the shard's own env instead of looping all 4 envs every invocation.

Before:
- build.yml: `test` job ran `make check` (4 envs serially) before coverage
- release.yml: each of 4 matrix shards ran `make check` AND `make test`,
  for 16 cppcheck rounds and 4 redundant native test runs total

After:
- build.yml: `test` job runs only host coverage; each `build`/`debug`
  shard runs `make check-<env>` parallel with the build
- release.yml: new top-level `test` job runs host tests once; each `build`
  shard runs `make check-<env>` before the firmware build

Job names preserved (test, build, debug, summary, release) so any branch-
protection required-checks keep matching. Static-check coverage of all 4
boards is preserved (one shard each).

New Makefile target: `make check-<env>` mirrors the existing `build-<env>`
pattern.

* ci: cache tool-cppcheck globally across matrix shards

Discovered while watching #204's first build run: each matrix shard
spent ~2 min on `Tool Manager: Installing platformio/tool-cppcheck`.
`--skip-packages` skips platforms/frameworks but NOT analysis tools, and
the existing `~/.platformio` cache is per-env-keyed, so 4 shards × cold
download = the bulk of the static-analysis wall time.

Add a dedicated cache layer for `~/.platformio/packages/tool-cppcheck`
with a global key (no matrix.env in it). All shards now share the
cppcheck tool — first shard primes the cache, subsequent runs hit it
across all envs.

Same change in build.yml (build + debug matrices) and release.yml.

269 of 334 branches covered (80.54%)

Branch coverage included in aggregate %.

823 of 842 relevant lines covered (97.74%)

493927.38 hits per line