From 2e416b49270481368a1e0fb387e2dc5ac81a11a0 Mon Sep 17 00:00:00 2001 From: Caleb Jones Date: Fri, 30 Dec 2016 17:46:32 -0500 Subject: [PATCH] Add documentation for Span --- src/span.rs | 25 ++++++++++++++++++++++++- 1 file changed, 24 insertions(+), 1 deletion(-) diff --git a/src/span.rs b/src/span.rs index a2e182c..3af56ac 100644 --- a/src/span.rs +++ b/src/span.rs @@ -1,13 +1,36 @@ //! A `Span` represents a location in source file. - +//! +//! Spans are useful for reporting error messages. They're used during parsing +//! and are carried around in the `Sexp` type so that errors can be reported at +//! a higher level than the s-expressions, for example reporting semantic errors +//! in a Lisp implementation. + +/// Represents a way of talking about a region in the source. +/// +/// All spans begin at a point in the source and contain some (possibly empty) +/// amount of text. pub trait Span { + /// The `Begin` type represents the beginning point of a span. type Begin; + /// Moves a `Span` forward by `begin`. + /// + /// useful for moving a relatively positioned span to start at a specific + /// absolute position. fn offset(&self, begin: Self::Begin) -> Self; + /// Returns the point where a span begins. fn begin(&self) -> Self::Begin; + /// Combines two spans into a single span that contains both of their + /// contents. fn union(&self, other: &Self) -> Self; } +/// `ByteSpan` is a very simple span, and the default for all of the parsers in +/// Ess. +/// +/// It represents a half-open range of bytes, the first element is the first +/// byte in the span and the second element is the first by after the span. This +/// means that `&text[span.0..span.1]` works exactly as we would expect it to. pub type ByteSpan = (usize, usize); impl Span for ByteSpan { -- 2.43.2