Reduce noise of TLAi linter. (#6103)

.github/workflows/tlaplus.yml

@@ -144,7 +144,7 @@ jobs:
         ## Identify git diff: $(git diff --name-only HEAD^ | grep '.tla')
         ## Install genaiscript runtime: https://microsoft.github.io/genaiscript/reference/cli/
         ## Output LLM response in SARIF format: https://microsoft.github.io/genaiscript/reference/scripts/annotations/ (redirect other output to /dev/null for GH not to also show the annotations)
-        run: npx --yes genaiscript run scripts/TLAi-linter.genai.js $(git diff --name-only HEAD^ | grep '\.tla') --max-tokens 2000 --out-annotations results.sarif > /dev/null
+        run: npx --yes genaiscript run scripts/TLAi-linter.genai.js $(git diff --name-only HEAD^ | grep '\.tla') --max-tokens 2000 --out-annotations results.sarif > /dev/null && jq '.runs |= map(.results |= map(select(.level != "note")))' results.sarif > filtered.sarif
 
       - name: Upload SARIF file
         if: (success() || failure()) && steps.git_commit.outputs.contains_tla == 'true'
@@ -152,4 +152,4 @@ jobs:
         ## https://docs.github.com/en/code-security/code-scanning/integrating-with-code-scanning/uploading-a-sarif-file-to-github
         uses: github/codeql-action/upload-sarif@v3
         with:
-          sarif_file: results.sarif
+          sarif_file: filtered.sarif

scripts/TLAi-linter.genai.js

@@ -14,6 +14,7 @@ def(
 
 // use $ to output formatted text to the prompt
 $`You are an expert at TLA+/TLAPLUS. Your task is to check if the prose comments and their TLA+ declarations and definitions are syntactically and semantically consistent!!!
+However, you may assume that that a parser has already checked the syntax of the TLA+ code.
 Explain any consistencies and inconsistencies you may find.  Report inconsistent and consistent pairs in a single ANNOTATION section.
 
 ## TLA+ Syntax Hints
@@ -30,4 +31,7 @@ Explain any consistencies and inconsistencies you may find.  Report inconsistent
 
 ## Formal and informal math Hints
 - Take into account that humans may write informal math that is syntactically different from the formal math, yet semantically equivalent.  For example, humans may write \`N > 3T\` instead of \`N > 3 * T\`.
+
+## Natural language Hints
+- Unless a built-in TLA+ declaration or definition like an operator in the TLA+ standard library dictates otherwise, the prose comments should follow British English spelling conventions.
 `;

tla/StatsFile.tla

@@ -16,7 +16,7 @@ WriteStatsFile ==
     /\ Serialize(<<TLCGet("stats")>>, StatsFilename, [format |-> "NDJSON", charset |-> "UTF-8", openOptions |-> <<"WRITE", "CREATE", "APPEND">>])
 
 \* Append TLC coverage in ndJson format to file identified by CoverageFilename.  Create CoverageFilename if it does not exist.
-SerialiseCoverage ==
+SerializeCoverage ==
     Serialize(<<TLCGet("spec")>>, CoverageFilename, [format |-> "NDJSON", charset |-> "UTF-8", openOptions |-> <<"WRITE", "CREATE", "APPEND">>])
 
 ====

tla/actions.py

@@ -16,7 +16,7 @@ ls -tr <spec>_coverage.json | xargs cat | python3 actions.py
 
 The spec needs to have constraint like:
 
-SerialiseCoverage ==
+SerializeCoverage ==
     Serialize(<<TLCGet("spec")>>, CoverageFilename, [format |-> "NDJSON", charset |-> "UTF-8", openOptions |-> <<"WRITE", "CREATE", "APPEND">>])
 
 """

tla/consensus/MCccfraftAtomicReconfig.cfg

@@ -93,4 +93,4 @@ INVARIANTS
     RetirementCompletedAreRetirementCompletedInv
 
 _PERIODIC
-    SerialiseCoverage
+    SerializeCoverage

tla/consensus/MCccfraftWithReconfig.cfg

@@ -92,4 +92,4 @@ INVARIANTS
     RetirementCompletedAreRetirementCompletedInv
 
 _PERIODIC
-    SerialiseCoverage
+    SerializeCoverage