batch function lines to speed up processing of many tiny functions

2 files changed, 25 insertions, 13 deletions

diff --git a/src/intertyper.js b/src/intertyper.js
index 03fbf072..417f6588 100644
--- a/src/intertyper.js
+++ b/src/intertyper.js
@@ -23,9 +23,9 @@ var NSW_NUW = set('nsw', 'nuw');
 
 // Intertyper
 
-function intertyper(data, sidePass, baseLineNum) {
+function intertyper(data, sidePass, baseLineNums) {
   var mainPass = !sidePass;
-  baseLineNum = baseLineNum || 0;
+  baseLineNums = baseLineNums || [[0,0]]; // each pair [#0,#1] means "starting from line #0, the base line num is #1"
 
   dprint('framework', 'Big picture: Starting intertyper, main pass=' + mainPass);
 
@@ -67,10 +67,15 @@ function intertyper(data, sidePass, baseLineNum) {
         };
         unparsedBundles.push(unparsedGlobals);
       }
+      var baseLineNumPosition = 0;
       for (var i = 0; i < lines.length; i++) {
         var line = lines[i];
         lines[i] = null; // lines may be very very large. Allow GCing to occur in the loop by releasing refs here
 
+        while (baseLineNumPosition < baseLineNums.length-1 && i >= baseLineNums[baseLineNumPosition+1][0]) {
+          baseLineNumPosition++;
+        }
+
         if (mainPass && (line[0] == '%' || line[0] == '@')) {
           // If this isn't a type, it's a global variable, make a note of the information now, we will need it later
           var testType = /[@%\w\d\.\" ]+ = type .*/.exec(line);
@@ -104,7 +109,7 @@ function intertyper(data, sidePass, baseLineNum) {
           } else {
             ret.push({
               lineText: line,
-              lineNum: i + 1 + baseLineNum
+              lineNum: i + 1 + baseLineNums[baseLineNumPosition][1]
             });
             if (/^\ +switch\ .*/.test(line)) {
               // beginning of llvm switch
diff --git a/src/jsifier.js b/src/jsifier.js
index df2ae43f..b6434455 100644
--- a/src/jsifier.js
+++ b/src/jsifier.js
@@ -108,17 +108,24 @@ function JSify(data, functionsOnly, givenFunctions) {
     }
   });
 
-  // TODO: batch small functions
-  for (var i = 0; i < data.unparsedFunctions.length; i++) {
-    var func = data.unparsedFunctions[i];
-    // We don't need to save anything past here
-    data.unparsedFunctions[i] = null;
-    dprint('unparsedFunctions', '====================\n// Processing |' + func.ident + '|, ' + i + '/' + data.unparsedFunctions.length);
-    if (DEBUG_MEMORY) MemoryDebugger.tick('pre-func ' + func.ident);
-    JSify(analyzer(intertyper(func.lines, true, func.lineNum-1)), true, Functions);
-    if (DEBUG_MEMORY) MemoryDebugger.tick('func ' + func.ident);
+
+  var MAX_BATCH_FUNC_LINES = 1000;
+  while (data.unparsedFunctions.length > 0) {
+    var currFuncLines = [];
+    var currBaseLineNums = [];
+    while (currFuncLines.length == 0 ||
+           (data.unparsedFunctions.length > 0 && currFuncLines.length + data.unparsedFunctions[0].lines.length <= MAX_BATCH_FUNC_LINES)) {
+      currBaseLineNums.push([currFuncLines.length, data.unparsedFunctions[0].lineNum-1]);
+      currFuncLines = currFuncLines.concat(data.unparsedFunctions[0].lines); // for first one, assign, do not concat?
+      data.unparsedFunctions.shift();
+    }
+    dprint('unparsedFunctions','====================\n// Processing function batch of ' + currBaseLineNums.length +
+                               ' functions, ' + currFuncLines.length + ' lines, functions left: ' + data.unparsedFunctions.length);
+    if (DEBUG_MEMORY) MemoryDebugger.tick('pre-func');
+    JSify(analyzer(intertyper(currFuncLines, true, currBaseLineNums)), true, Functions);
+    if (DEBUG_MEMORY) MemoryDebugger.tick('post-func');
   }
-  func = null; // Do not hold on to anything from inside that loop (JS function scoping..)
+  currFuncLines = currBaseLineNums = null; // Do not hold on to anything from inside that loop (JS function scoping..)
   data.unparsedFunctions = null;
 
   // Actors