PICO-8's shorthand `if (cond) stmt [else stmt]` is line-bounded, but
tree-sitter has no built-in newline awareness. Without an external
scanner ( the same mechanism tree-sitter-python uses for INDENT /
DEDENT / NEWLINE ), the grammar greedily binds `else` to the nearest
`if` and takes only one consequence statement for the shorthand body.
Token classification is unaffected, so syntax highlighting renders
identically to a correct parse; only auto-indent and semantic
selection are subtly off, in a code pattern that is uncommon in real
PICO-8 code.
New `grammars/pico-8-lua/KNOWN_LIMITATIONS.md` walks through both
incorrect cases ( the dangling-else mis-bind and the multi-statement
shorthand body ), tabulates which Zed features are and aren't
affected, and sketches the fix. README cross-links it from the
"Known limitations" block and adds it as a prerequisite to the v0.3
LSP work.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reorganize into grammars/<name>/ subdirs ( Zed's [grammars.*] supports
a `path` field, so both grammars ship from this repo without a sibling-
repo split ). Vendor tree-sitter-lua as the fork base for tree-sitter-
pico8-lua; upstream MIT license preserved at grammars/pico-8-lua/
UPSTREAM-LICENSE.md.
Dialect features added: != as ~= alias, \ integer divide, ^^ binary xor,
>>> / <<> / >>< shifts and rotates, compound-assignment statements,
memory peek prefixes @ % $ (% coexists with binary modulo), single-line
`if (cond) stmt [else stmt]` and `while (cond) stmt`, statement-level
print shorthand ?, and `#include path` directives. Identifier rule no
longer accepts ! ? @ $ ( upstream did ).
Pico-8 Lua language ( languages/pico-8-lua/, suffix .p8lua ) ships
highlights with the full ~110 PICO-8 builtins as @function.builtin.
The cart injection now hands __lua__ bodies to pico-8-lua, so .p8 carts
and bare .p8lua files share the dialect-aware grammar. Examples updated
to exercise the dialect end-to-end.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The grammar parses the .p8 cartridge container ( header, version, and
the named __lua__/__gfx__/__gff__/__label__/__map__/__sfx__/__music__
sections, plus a fallback unknown_section ). The Zed language definition
hands the __lua__ body to Zed's built-in Lua via injections.scm, so
non-dialect Lua code highlights correctly today; PICO-8-specific syntax
(?, +=, !=, single-line if, etc.) will fall back to error highlighting
in those regions only — see README roadmap for the dialect grammar fork.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
kistaro
