Suppressing diagnostics

Sometimes a diagnostic is a false alarm, or a warning you have decided to live with. You can silence any diagnostic, and the language server never hides one silently: you can always see what a suppression is doing.

Inline, one place at a time

Put a ; @suppress <code> comment on the line above a diagnostic, or at the end of its own line. It silences that code on both the comment's line and the next.

; @suppress symbol.possibleTypo
!(cdr-atomm 1)

!(car-atomm 2) ; @suppress symbol.possibleTypo

; @suppress-file <code> silences a code for the whole file; a bare ; @suppress-file silences everything. A code-less ; @suppress silences every code on the covered lines.

Every diagnostic also offers a Suppress <code> on this line quick-fix (the lightbulb) that inserts the comment for you.

Code as data, across a whole project

A lint.metta can carry a suppression written as a MeTTa pattern:

; Silence symbol.possibleTypo on anything inside a (legacy …) call:
(suppress (legacy $$$) symbol.possibleTypo)

; Silence every code inside an (experimental …) block:
(suppress (experimental $$$))

The pattern is matched structurally, exactly like a lint rule, so the suppression is data, not a function that runs. List codes after the pattern to narrow it, or leave them off to cover every code. This is more robust than line-based suppression: it follows the code as you move it.

Never silent

A suppression you cannot see is a trap. So:

• Hover a ; @suppress directive to see the diagnostics it is silencing, with their messages, or a note that it is silencing nothing and is unused.
• From the command line, metta-lsp check <file> --show-suppressed lists every silenced diagnostic and the exact directive or rule that hid it.

metta-lsp check app.metta --show-suppressed

What can be suppressed

Everything: the core diagnostics (undefined symbols, arity, type mismatches, imports, …), the semantic-lint rules, the host-bridge checks, and the syntactic lint rules. See the diagnostics catalogue for the codes.