diff --git a/src/librustdoc/html/render/mod.rs b/src/librustdoc/html/render/mod.rs index ea466c639c3..36d15ec3b86 100644 --- a/src/librustdoc/html/render/mod.rs +++ b/src/librustdoc/html/render/mod.rs @@ -2957,14 +2957,15 @@ fn render_call_locations(w: &mut Buffer, cx: &mut Context<'_>, item: &clean::Ite // The call locations are output in sequence, so that sequence needs to be determined. // Ideally the most "relevant" examples would be shown first, but there's no general algorithm - // for determining relevance. We instead make a proxy for relevance with the following heuristics: + // for determining relevance. We instead proxy relevance with the following heuristics: // 1. Code written to be an example is better than code not written to be an example, e.g. - // a snippet from examples/foo.rs is better than src/lib.rs. We don't know the Cargo directory - // structure in Rustdoc, so we proxy this by prioriting code that comes from a --crate-type bin. - // 2. Smaller examples are better than large examples. So we prioritize snippets that have the - // smallest line span for their enclosing item. - // 3. Finally we sort by the displayed file name, which is arbitrary but prevents the ordering - // of examples from randomly changing between Rustdoc invocations. + // a snippet from examples/foo.rs is better than src/lib.rs. We don't know the Cargo + // directory structure in Rustdoc, so we proxy this by prioritizing code that comes from + // a --crate-type bin. + // 2. Smaller examples are better than large examples. So we prioritize snippets that have + // the smallest number of lines in their enclosing item. + // 3. Finally we sort by the displayed file name, which is arbitrary but prevents the + // ordering of examples from randomly changing between Rustdoc invocations. let ordered_locations = { fn sort_criterion<'a>( (_, call_data): &(&PathBuf, &'a CallData), diff --git a/src/librustdoc/html/static/scrape-examples-help.md b/src/librustdoc/html/static/scrape-examples-help.md index 035b2e18b00..002d19ec9b6 100644 --- a/src/librustdoc/html/static/scrape-examples-help.md +++ b/src/librustdoc/html/static/scrape-examples-help.md @@ -1,4 +1,4 @@ -Rustdoc will automatically scrape examples of documented items from the `examples/` directory of a project. These examples will be included within the generated documentation for that item. For example, if your library contains a public function: +Rustdoc will automatically scrape examples of documented items from a project's source code. These examples will be included within the generated documentation for that item. For example, if your library contains a public function: ```rust // src/lib.rs @@ -16,6 +16,7 @@ fn main() { Then this code snippet will be included in the documentation for `a_func`. + ## How to read scraped examples Scraped examples are shown as blocks of code from a given file. The relevant item will be highlighted. If the file is larger than a couple lines, only a small window will be shown which you can expand by clicking ↕ in the top-right. If a file contains multiple instances of an item, you can use the ≺ and ≻ buttons to toggle through each instance. @@ -25,7 +26,7 @@ If there is more than one file that contains examples, then you should click "Mo ## How Rustdoc scrapes examples -When you run `cargo doc`, Rustdoc will analyze all the crates that match Cargo's `--examples` filter for instances of items that occur in the crates being documented. Then Rustdoc will include the source code of these instances in the generated documentation. +When you run `cargo doc -Zunstable-options -Zrustdoc-scrape-examples`, Rustdoc will analyze all the documented crates for uses of documented items. Then Rustdoc will include the source code of these instances in the generated documentation. Rustdoc has a few techniques to ensure this doesn't overwhelm documentation readers, and that it doesn't blow up the page size: