| |
|
|
| |
compile: colorQuery kept each grammar's patterns whose captures pardes colours by scanning the 29 highlights.scm byte by byte at comptime, a quarter of a full recompile of the core (and again in each of the ~15 artifacts that import it); build.zig's treeSitter now runs the same filter once over the query it reads (grammar_manifest.colorQuery, a plain function, with Syn and synFor's capture-to-class table beside it, the one copy both the build and syntax.zig read), so ts_queries' <name>_highlights is the filtered query and Spec.query_src a field read; typst's supplement is the manifest's highlights_extra, appended before the filter as before; <name>_highlights_unfiltered is the whole query, for the test that the filter keeps every colour; a package that depends on pardes gets the same through library's treeSitter, for its own grammar set; proven byte-identical before the comptime path went (29 grammars, 28,398 filtered bytes of 79,678); core-test's forced full recompile 12.7-12.9 s wall / 17.5 s CPU to 9.7-10.1 s / 14.6 s; syntax snapshots unchanged; the changelog says so
|
| |
|
|
|
|
|
|
| |
any case
Repl took only the grammar's exact name, and a file's language was found by a case-sensitive suffix of its path, apart from the aliases a code fence already had. The aliases move to grammar_manifest with byName and byPath, which syntax, Repl and langOf share, so `Repl py` and a .PY file work. The line-by-line warning is for the python language, not an id that starts with python. Joincol's doc line was sitting on Repl.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|