mirror of
https://github.com/nushell/nushell.git
synced 2025-05-08 08:52:56 +00:00
a2873336bb
Recommend holding until after #13125 is fully digested and *possibly* until 0.96. # Description Fixes one of the issues described in #13125 The `do` signature included a `SyntaxShape:Any` as one of the possible first-positional types. This is incorrect. `do` only takes a closure as a positional. This had the result of: 1. Moving what should have been a parser error to evaluation-time ## Before ```nu > do 1 Error: nu:🐚:cant_convert × Can't convert to Closure. ╭─[entry #26:1:4] 1 │ do 1 · ┬ · ╰── can't convert int to Closure ╰──── ``` ## After ```nu > do 1 Error: nu::parser::parse_mismatch × Parse mismatch during operation. ╭─[entry #5:1:4] 1 │ do 1 · ┬ · ╰── expected block, closure or record ╰──── ``` 2. Masking a bad test in `std assert` This is a bit convoluted, but `std assert` tests included testing `assert error` to make sure it: * Asserts on bad code * Doesn't assert on good code The good-code test was broken, and was essentially bad-code (really bad-code) that wasn't getting caught due to the bad signature. Fixing this resulted in *parse time* failures on every call to `test_asserts` (not something that particular test was designed to handle. This PR also fixes the test case to properly evaluate `std assert error` against a good code path. # User-Facing Changes * Error-type returned (possible breaking change?) # Tests + Formatting - 🟢 `toolkit fmt` - 🟢 `toolkit clippy` - 🟢 `toolkit test` - 🟢 `toolkit test stdlib` # After Submitting N/A