Update Combinatory.newt, fix parse error

newt-vscode/src/abbrev.ts

@@ -1,4 +1,5 @@
-// This needs to be flexhed out a lot, I've been adding them as needed
+// This needs to be fleshed out a lot, I've been adding them as needed
+// There is a mix of agda, lean, and my own shortcuts here
 export const ABBREV: Record<string, string> = {
   "\\x": "×",
   "\\r": "→",
@@ -8,20 +9,26 @@ export const ABBREV: Record<string, string> = {
   "\\circ": "∘",
   "\\oplus": "⊕",
   "\\otimes": "⊗",
+  // lean
   "\\1": "₁",
   "\\2": "₂",
   "\\<": "⟨",
   "\\>": "⟩",
+  // agda
   "\\_0": "₀",
   "\\_1": "₁",
   "\\_2": "₂",
   "\\_3": "₃",
+  // lean has \n here, which is a royal pain
   "\\neg": "¬",
   "\\bN": "ℕ",
   "\\bZ": "ℤ",
-  "\\gi": "ι",
-  "\\gs": "σ",
-  "\\gt": "τ",
-  "\\gD": "Δ",
-  "\\gG": "Γ"
+  "\\GG": "Γ",
+  "\\Gi": "ι",
+  "\\Gl": "λ",
+  "\\Gs": "σ",
+  "\\Gt": "τ",
+  "\\GD": "Δ",
+  "\\[[": "⟦",
+  "\\]]": "⟧",
 };

newt-vscode/src/extension.ts

@@ -30,10 +30,14 @@ export function activate(context: vscode.ExtensionContext) {
     const changes = event.contentChanges;
     if (changes.length === 0) return;
 
+    // TODO - agda input mode does the replacement as soon as possible
+    // but if the sequence is a prefix, it will change for subsequent characters
+    // The latter would require keeping state, but if we don't allow sequences to prefix
+    // each other, we could activate quicker.
     const lastChange = changes[changes.length - 1];
     const text = lastChange.text;
     // Check if the last change is a potential shortcut trigger
-    if (!text || !( " ')\\".includes(text) || text.startsWith('\n'))) return;
+    if (!text || !( " ')\\_".includes(text) || text.startsWith('\n'))) return;
 
     const document = editor.document;
     const position = lastChange.range.end;