Repository navigation
t27c gen: open slice x[i..] emitted as x[i]; &[_]T{...} emitted as &.{ _ } #6250
Description
Found by the t27b slices lane (PR #6249, Refs #6063).
Problem
t27c gen emits wrong Zig for two slice forms, so specs that use them cannot be checked against the reference path:
- an open-ended slice
x[i..]is emitted as an indexx[i]; - a slice literal
&[_]T{ a, b }is emitted as&.{ _ }.
Evidence
The slices lane confirmed both with a probe spec and zig test on the generated Zig. t27b lowers both forms correctly.
Expected
x[i..] stays a slice to the end, and &[_]T{...} keeps its element type and elements.
Related
The x = e; shadowing in test blocks is already tracked in #4847 and #5606.
🤖 Generated with Claude Code
User Scenarios
- Given a spec containing an open-ended slice
x[i..], whent27c genemits Zig code, then the output uses a slice-to-end expression rather than an index expression. - Given a spec containing a slice literal
&[_]T{ a, b }, whent27c genemits Zig code, then the output preserves the element typeTand the elementsa, b.
Requirements
FR-001: t27c gen MUST emit Zig slice-to-end syntax for open-ended slice x[i..].
FR-002: t27c gen MUST emit Zig slice literal syntax preserving element type and elements for &[_]T{...}.
Success Criteria
zig teston generated Zig for a probe spec withx[i..]passes (exit 0)zig teston generated Zig for a probe spec with&[_]T{ a, b }passes (exit 0)- Both probe specs compile without errors in the generated Zig output
Boundary
- bootstrap/t27c.py
Activity
gHashTag commentedon Oct 5, 2026
Dogfood (#6457): 5 own specs show "'' used as an identifier without @"" syntax" as their first error at lab run 682564f: tri/encoding/{csv,msgpack}, tri/math/matrix, tri/search/aho_corasick, tri/trees/fenwick_tree (csv:28 is
&[_]CsvRow{}). Probably this issue; please confirm when fixing.fix(specs): msgpack encodes and decodes MessagePack instead of return…
gHashTag commentedon Oct 7, 2026
Six more specs now reach this lowering. They were stopped earlier by imports that did not resolve, and #7191's second slice (local 93838ef) fixes those imports. Under the #7242 build,
t27c test-reportnow stops each one at'_' used as an identifier without @"_" syntax:specs/bus/pubsub.t27specs/lsp/client.t27specs/lsp/server.t27specs/provider/adapters.t27specs/provider/transform.t27specs/sync/index.t27Measured on the Railway lab, 2026-10-07T03:10Z. The fix itself is in
bootstrap/src/compiler.rs, so it stays with the owner.🤖 Generated with Claude Code
gHashTag commentedon Oct 7, 2026
Reopened 2026-10-07: the bulk close of issues without a
## Boundarysection was reverted by the owner. Nothing to do here; this comment replaces the earlier close notice.