Struct ProcessLinter 

pub struct ProcessLinter { /* private fields */ }

What: The linter of one invocation: the environment it runs with, its bounds, where its files go, and, once known, its program and configuration file. Why: The lookup and the file are made at most once; their failures are remembered and reported to every later candidate the same way.

In TS you’d write (pseudocode):

class ProcessLinter implements MarkdownLinter { executable?: string | null; configuration?: TemporaryFile | string }

Implementations

impl ProcessLinter

What: impl ProcessLinter { ... } attaches the two remembered preparations. Why: They are made on first use and never twice.

In TS you’d write (pseudocode):

class ProcessLinter { program(): string; configurationPath(options): string }

pub fn with_notices(self, cadence: Cadence, notices: Box<dyn Write>) -> Self

The same linter with another notice cadence and sink; tests choose a short cadence and read the notices from a buffer.

Trait Implementations

impl MarkdownLinter for ProcessLinter

What: impl MarkdownLinter for ProcessLinter starts the real linter for each request. Why: See the module comment.

In TS you’d write (pseudocode):

class ProcessLinter implements MarkdownLinter {}

fn lint(&mut self, request: &LintRequest<'_>) -> Result<LintRun, String>

What: Lint one candidate through the linter’s standard input. Why: The linter reads .lfsconfig and linked images from the worktree, so it runs from the top level and is told the candidate’s name relative to it.

In TS you’d write (pseudocode):

lint(request: LintRequest): LintRun; // throws string