Struct FailureSite 

pub struct FailureSite {
    pub span: Span,
    pub component: Symbol,
    pub action: Symbol,
}

One entry in Parser::failure_sites — everything a runtime failure needs to point BACK at the sentence a Stmt::Site marker stood in front of.

★ “CURRENT SITE = n” IS ENOUGH, AND HERE IS WHY. A trap does not need a call stack to be useful — LOGOS actions do not recurse into each other’s OWN bodies the way a general-purpose language’s functions do (an action calls another via Call <Store>'s <action>, a dispatch through data, not a nested nomination-time invocation whose OWN Sites would need distinguishing on a stack) — so “which Site did the runtime most recently pass through” already answers “which sentence trapped” unambiguously, for the same reason a program counter alone (no full backtrace) already answers it in a language with no first-class stack frames to unwind through. A future construct that DOES nest one action’s failure inside another’s turn is exactly the case that would need this reconsidered — flagging it here so nobody has to rediscover why a stack was not built.

★ NO file FIELD, ON PURPOSE. A Parser parses ONE flat token stream and has no concept of splicing at all — loader::apply_prelude_mapped assembles a user file plus its demand-imported stdlib modules into that ONE stream BEFORE the parser ever sees it, and ui_bridge.rs’s checked seam is what undoes the splice, exactly once, on whatever span leaves the compile pipeline (item 10 of this session’s own record — the double-mapping bug that fix closed). span here is in the SAME spliced coordinates every other span this parser hands out is in, for the same reason: a second, bespoke translation point for FailureSite spans specifically would be exactly the double-mapping shape that bug already was. Whoever renders a FailureSite for a person translates it through the SAME seam every other diagnostic in this pipeline already goes through.

Fields

span: Span

Where the marked sentence is, in spliced coordinates — see the field’s own note above for why there is no separate file to pair it with.

component: Symbol

The component (or store, or page) whose action/handler this sentence is written in.

action: Symbol

The action or event handler this sentence belongs to — __e3-style synthesized names for a When …: handler, same as every other place this pipeline names one.

Implementations

impl FailureSite

pub fn render(&self, interner: &Interner, message: &str) -> String

message (the runtime error text itself, unchanged) wrapped with WHICH sentence raised it — component/action, in the author’s own words wherever this site can recover them. Deliberately does NOT touch file:line:col: span is still in spliced coordinates (this struct’s own doc comment says why there is no file to pair it with), and turning a byte offset into a line/column needs the source text this crate never holds once parsing ends — that translation is ui_bridge.rs’s checked seam’s job, on the SAME span, same as every other diagnostic; this method only ever supplies the component/action half.

An event handler’s action is a synthesized __e<n> (lgx_lower.rs::action_fn_name’s own numbering) — never a name the author wrote, so it is named generically (“an event handler”) rather than surfaced verbatim. An ordinary To <name>: action’s own name IS what the author wrote, and is quoted as `To <name>` to read the same way the section that declares it does.

Trait Implementations

impl Clone for FailureSite

fn clone(&self) -> FailureSite

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for FailureSite

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Copy for FailureSite