Some of the changes to improve --build memory footprint (microsoft#2604)

Author: sheetalkamat

internal/compiler/program.go

@@ -1405,9 +1405,10 @@ func (p *Program) Emit(ctx context.Context, options EmitOptions) *EmitResult {
 		}
 	}
 
+	newLine := p.Options().NewLine.GetNewLineCharacter()
 	writerPool := &sync.Pool{
 		New: func() any {
-			return printer.NewTextWriter(p.Options().NewLine.GetNewLineCharacter())
+			return printer.NewTextWriter(newLine)
 		},
 	}
 	wg := core.NewWorkGroup(p.SingleThreaded())

internal/diagnostics/diagnostics_generated.go

@@ -39,10 +39,6 @@
         "category": "Message",
         "code": 100009
     },
-    "all, unless --singleThreaded is passed.": {
-        "category": "Message",
-        "code": 100010
-    },
     "Non-relative paths are not allowed. Did you forget a leading './'?": {
         "category": "Error",
         "code": 5090

internal/diagnostics/extraDiagnosticMessages.json

@@ -200,10 +200,6 @@ func (t *BuildTask) updateDownstream(orchestrator *Orchestrator, path tspath.Pat
 }
 
 func (t *BuildTask) compileAndEmit(orchestrator *Orchestrator, path tspath.Path) {
-	if orchestrator.buildSemaphore != nil {
-		orchestrator.buildSemaphore <- struct{}{}        // acquire slot
-		defer func() { <-orchestrator.buildSemaphore }() // release slot
-	}
 	t.errors = nil
 	if orchestrator.opts.Command.BuildOptions.Verbose.IsTrue() {
 		t.result.reportStatus(ast.NewCompilerDiagnostic(diagnostics.Building_project_0, orchestrator.relativeFileName(t.config)))

internal/execute/build/buildtask.go

@@ -51,14 +51,15 @@ func (h *host) Trace(msg *diagnostics.Message, args ...any) {
 }
 
 func (h *host) GetSourceFile(opts ast.SourceFileParseOptions) *ast.SourceFile {
-	// Cache dts and json files as they will be reused
-	return h.sourceFiles.loadOrStoreNewIf(opts, h.host.GetSourceFile, func(value *ast.SourceFile) bool {
-		return value != nil && (tspath.IsDeclarationFileName(opts.FileName) || tspath.FileExtensionIs(opts.FileName, tspath.ExtensionJson))
-	})
+	if tspath.IsDeclarationFileName(opts.FileName) || tspath.FileExtensionIs(opts.FileName, tspath.ExtensionJson) {
+		// Cache dts and json files as they will be reused
+		return h.sourceFiles.loadOrStore(opts, h.host.GetSourceFile, false /* allowZero */)
+	}
+	return h.host.GetSourceFile(opts)
 }
 
 func (h *host) GetResolvedProjectReference(fileName string, path tspath.Path) *tsoptions.ParsedCommandLine {
-	return h.resolvedReferences.loadOrStoreNew(path, func(path tspath.Path) *tsoptions.ParsedCommandLine {
+	return h.resolvedReferences.loadOrStore(path, func(path tspath.Path) *tsoptions.ParsedCommandLine {
 		configStart := h.orchestrator.opts.Sys.Now()
 		// Wrap command line options in "compilerOptions" key to match tsconfig.json structure
 		var commandLineRaw *collections.OrderedMap[string, any]
@@ -71,7 +72,7 @@ func (h *host) GetResolvedProjectReference(fileName string, path tspath.Path) *t
 		configTime := h.orchestrator.opts.Sys.Now().Sub(configStart)
 		h.configTimes.Store(path, configTime)
 		return commandLine
-	})
+	}, true /* allowZero */)
 }
 
 func (h *host) ReadBuildInfo(config *tsoptions.ParsedCommandLine) *incremental.BuildInfo {

internal/execute/build/host.go

@@ -61,8 +61,6 @@ type Orchestrator struct {
 	tasks  *collections.SyncMap[tspath.Path, *BuildTask]
 	order  []string
 	errors []*ast.Diagnostic
-	// Semaphore to limit concurrent builds
-	buildSemaphore chan struct{}
 
 	errorSummaryReporter tsc.DiagnosticsReporter
 	watchStatusReporter  tsc.DiagnosticReporter
@@ -240,14 +238,9 @@ func (o *Orchestrator) Watch() {
 func (o *Orchestrator) updateWatch() {
 	oldCache := o.host.mTimes
 	o.host.mTimes = &collections.SyncMap[tspath.Path, time.Time]{}
-	wg := core.NewWorkGroup(o.opts.Command.CompilerOptions.SingleThreaded.IsTrue())
-	o.tasks.Range(func(path tspath.Path, task *BuildTask) bool {
-		wg.Queue(func() {
-			task.updateWatch(o, oldCache)
-		})
-		return true
+	o.rangeTask(func(path tspath.Path, task *BuildTask) {
+		task.updateWatch(o, oldCache)
 	})
-	wg.RunAndWait()
 }
 
 func (o *Orchestrator) resetCaches() {
@@ -263,20 +256,14 @@ func (o *Orchestrator) DoCycle() {
 	var needsConfigUpdate atomic.Bool
 	var needsUpdate atomic.Bool
 	mTimes := o.host.mTimes.Clone()
-	wg := core.NewWorkGroup(o.opts.Command.CompilerOptions.SingleThreaded.IsTrue())
-	o.tasks.Range(func(path tspath.Path, task *BuildTask) bool {
-		wg.Queue(func() {
-			if updateKind := task.hasUpdate(o, path); updateKind != updateKindNone {
-				needsUpdate.Store(true)
-				if updateKind == updateKindConfig {
-					needsConfigUpdate.Store(true)
-				}
+	o.rangeTask(func(path tspath.Path, task *BuildTask) {
+		if updateKind := task.hasUpdate(o, path); updateKind != updateKindNone {
+			needsUpdate.Store(true)
+			if updateKind == updateKindConfig {
+				needsConfigUpdate.Store(true)
 			}
-		})
-		// Watch for file changes
-		return true
+		}
 	})
-	wg.RunAndWait()
 
 	if !needsUpdate.Load() {
 		o.host.mTimes = mTimes
@@ -307,11 +294,9 @@ func (o *Orchestrator) buildOrClean() tsc.CommandLineResult {
 	var buildResult orchestratorResult
 	if len(o.errors) == 0 {
 		buildResult.statistics.Projects = len(o.Order())
-		if o.opts.Command.CompilerOptions.SingleThreaded.IsTrue() {
-			o.singleThreadedBuildOrClean(&buildResult)
-		} else {
-			o.multiThreadedBuildOrClean(&buildResult)
-		}
+		o.rangeTask(func(path tspath.Path, task *BuildTask) {
+			o.buildOrCleanProject(task, path, &buildResult)
+		})
 	} else {
 		// Circularity errors prevent any project from being built
 		buildResult.result.Status = tsc.ExitStatusProjectReferenceCycle_OutputsSkipped
@@ -325,25 +310,40 @@ func (o *Orchestrator) buildOrClean() tsc.CommandLineResult {
 	return buildResult.result
 }
 
-func (o *Orchestrator) singleThreadedBuildOrClean(buildResult *orchestratorResult) {
-	// Go in the order since only one project can be built at a time so that random order isnt picked by work group creating deadlock
-	for _, config := range o.Order() {
+func (o *Orchestrator) rangeTask(f func(path tspath.Path, task *BuildTask)) {
+	numRoutines := 4
+	if o.opts.Command.CompilerOptions.SingleThreaded.IsTrue() {
+		numRoutines = 1
+	} else if builders := o.opts.Command.BuildOptions.Builders; builders != nil {
+		numRoutines = *builders
+	}
+
+	var currentTaskIndex atomic.Int64
+	getNextTask := func() (tspath.Path, *BuildTask, bool) {
+		index := int(currentTaskIndex.Add(1) - 1)
+		if index >= len(o.order) {
+			return "", nil, false
+		}
+		config := o.order[index]
 		path := o.toPath(config)
 		task := o.getTask(path)
-		o.buildOrCleanProject(task, path, buildResult)
+		return path, task, true
+	}
+	runTask := func() {
+		for path, task, ok := getNextTask(); ok; path, task, ok = getNextTask() {
+			f(path, task)
+		}
 	}
-}
 
-func (o *Orchestrator) multiThreadedBuildOrClean(buildResult *orchestratorResult) {
-	// Spin off the threads with waiting on upstream to build before actual project build
-	wg := core.NewWorkGroup(false)
-	o.tasks.Range(func(path tspath.Path, task *BuildTask) bool {
-		wg.Queue(func() {
-			o.buildOrCleanProject(task, path, buildResult)
-		})
-		return true
-	})
-	wg.RunAndWait()
+	if numRoutines == 1 {
+		runTask()
+	} else {
+		wg := core.NewWorkGroup(false)
+		for range numRoutines {
+			wg.Queue(runTask)
+		}
+		wg.RunAndWait()
+	}
 }
 
 func (o *Orchestrator) buildOrCleanProject(task *BuildTask, path tspath.Path, buildResult *orchestratorResult) {
@@ -398,9 +398,5 @@ func NewOrchestrator(opts Options) *Orchestrator {
 	} else {
 		orchestrator.errorSummaryReporter = tsc.CreateReportErrorSummary(opts.Sys, opts.Command.Locale(), opts.Command.CompilerOptions)
 	}
-	// If we want to build more than one project at a time, create a semaphore to limit concurrency
-	if builders := opts.Command.BuildOptions.Builders; builders != nil {
-		orchestrator.buildSemaphore = make(chan struct{}, *builders)
-	}
 	return orchestrator
 }

internal/execute/build/orchestrator.go

@@ -6,27 +6,23 @@ import (
 	"github.com/microsoft/typescript-go/internal/collections"
 )
 
-type parseCacheEntry[V any] struct {
+type parseCacheEntry[V comparable] struct {
 	value V
 	mu    sync.Mutex
 }
 
-type parseCache[K comparable, V any] struct {
+type parseCache[K comparable, V comparable] struct {
 	entries collections.SyncMap[K, *parseCacheEntry[V]]
 }
 
-func (c *parseCache[K, V]) loadOrStoreNew(key K, parse func(K) V) V {
-	return c.loadOrStoreNewIf(key, parse, func(value V) bool { return true })
-}
-
-func (c *parseCache[K, V]) loadOrStoreNewIf(key K, parse func(K) V, canCacheValue func(V) bool) V {
+func (c *parseCache[K, V]) loadOrStore(key K, parse func(K) V, allowZero bool) V {
 	newEntry := &parseCacheEntry[V]{}
 	newEntry.mu.Lock()
 	defer newEntry.mu.Unlock()
 	if entry, loaded := c.entries.LoadOrStore(key, newEntry); loaded {
 		entry.mu.Lock()
 		defer entry.mu.Unlock()
-		if canCacheValue(entry.value) {
+		if allowZero || entry.value != *new(V) {
 			return entry.value
 		}
 		newEntry = entry

internal/execute/build/parseCache.go

@@ -54,7 +54,7 @@ var OptionsForBuild = []*CommandLineOption{
 		Kind:                    CommandLineOptionTypeNumber,
 		Category:                diagnostics.Command_line_Options,
 		Description:             diagnostics.Set_the_number_of_projects_to_build_concurrently,
-		DefaultValueDescription: diagnostics.X_all_unless_singleThreaded_is_passed,
+		DefaultValueDescription: diagnostics.X_4_unless_singleThreaded_is_passed,
 		minValue:                1,
 	},
 	{