Deadfinder::Options

Raw --accept-status / --dead-status values are kept for help and error messages; the parsed ranges are what the hot path consults. Both are set once during CLI parsing, before any worker fiber exists, so the workers only ever read them.

Opt-in: verify that #fragment link targets actually exist in the linked document. Off by default because it needs the response body, which the status-only link check never reads.

Opt-in: verify that #fragment link targets actually exist in the linked document. Off by default because it needs the response body, which the status-only link check never reads.

Minimum interval, in milliseconds, between two requests to the same host. Per-host on purpose: one slow host must not stall the others.

Minimum interval, in milliseconds, between two requests to the same host. Per-host on purpose: one slow host must not stall the others.

Opt-in CI gate. The v1 CLI contract always exits 0, so this stays off by default and only --fail-on-dead turns findings into a non-zero exit.

Opt-in CI gate. The v1 CLI contract always exits 0, so this stays off by default and only --fail-on-dead turns findings into a non-zero exit.

HTTP method used for link status checks only; documents (the scan target page, a sitemap) are always fetched with GET because their body is the point. See HttpClient::METHOD_AUTO for the auto fallback rule.

HTTP method used for link status checks only; documents (the scan target page, a sitemap) are always fetched with GET because their body is the point. See HttpClient::METHOD_AUTO for the auto fallback rule.

Extra attempts after the first for a transient failure (connection error, timeout, 429, 5xx). A 404 is never retried.

Extra attempts after the first for a transient failure (connection error, timeout, 429, 5xx). A 404 is never retried.

How many targets are scanned at once. This does not widen the request budget: concurrency stays the global cap on in-flight HTTP requests, and target concurrency only decides how many pages compete for it. Without it -c only parallelized the links within a single page, so a 5000-URL sitemap paid 5000 serial round trips before any of that concurrency helped.

How many targets are scanned at once. This does not widen the request budget: concurrency stays the global cap on in-flight HTTP requests, and target concurrency only decides how many pages compete for it. Without it -c only parallelized the links within a single page, so a 5000-URL sitemap paid 5000 serial round trips before any of that concurrency helped.