v0.2.7
Show a POSIX path in the convention notice, and make the run readable
The notice printed path.relative, which is tests\bootstrap.ts on Windows —
the one platform where the assertion checking it ran, and where it failed. The
path is read by a person and pasted back into a command, so it is shown with
forward slashes everywhere, and the test now refuses a backslash outright.
Same shape as the failure in rover this morning, and the same second half: the
job ran vitest with the JSON reporter alone, so the log held an exit code and
not one word about which test failed. The default reporter runs alongside it,
with the JSON one still feeding the gate.
Release 0.2.7
Say when a bootstrap was picked up by convention
helix probes helix.config.* and nothing else, which is right — it is
framework-agnostic and a framework's rc file is not its business. But a Ream
project keeps its answer in reamrc.ts, so the same project gives two different
bootstraps depending on which command runs the tests, and neither said so.
The reported shape: a tests/bootstrap.ts that starts the application, picked up
silently, so a unit suite that needed no server opened one with a connection
pool behind it. The workaround was to not create the file — which means the
first person to write a legitimate one walks into it.
The convention still applies; it just announces itself, and only where there is
an rc file that could disagree, so a plain helix project gets no lecture. helix
still never reads that file — it checks whether one exists, by name.
Format the Rust crates, and gate it so they stay formatted
Twelve of the thirteen crates had drifted — 949 differences in all, atlas
alone 345, and build.rs files that had never been through the formatter.
None of their workflows checked, so nothing ever said so; the drift only
surfaced when it took a publish job down.
cargo fmt applied throughout, and a cargo fmt --check step added to each
workflow so this cannot happen again. Formatting only: the Rust tests pass
unchanged in every crate.
One spot in atlas needed a real edit rather than the formatter: cargo fmt
rewrote a return Err(format!(…)) arm back to the inline form on every run
while --check kept asking for the block form, so the file could never
converge. The message is bound to a name, which fits the line budget and
settles it.
Ignore a relocated cargo target
target/ matches a directory only. When the build output is moved elsewhere
and a symlink named target is left in its place, that pattern does not catch
it — it shows up untracked, and a stray git add -A commits a path that only
resolves on one machine.
Changes since v0.2.6.