Merge pull request #3406 from demergent-labs/fix_benchmarks

fix benchmark recording

Author: lastmjs

.github/scripts/analyze_benchmarks/extractor.ts

@@ -1,9 +1,9 @@
 import { readFile } from 'fs/promises';
 
 export type BenchmarkEntry = {
-    instructions: { __bigint__: string };
+    instructions: string;
     method_name: string;
-    timestamp: { __bigint__: string };
+    timestamp: string;
 };
 
 type VersionBenchmarks = {

.github/scripts/analyze_benchmarks/index.ts

@@ -49,19 +49,34 @@ async function analyzeBenchmarksForVersion(
 /**
  * Runs the benchmark analysis for a specified version or current version
  * @param specifiedVersion Optional version to analyze. If not provided, uses current package version
+ * @param debugMode If true, throws errors instead of writing them to markdown
  */
-async function runBenchmarkAnalysis(specifiedVersion?: string): Promise<void> {
+async function runBenchmarkAnalysis(
+    specifiedVersion?: string,
+    debugMode: boolean = false
+): Promise<void> {
     const versionToAnalyze = specifiedVersion ?? currentAzleVersion;
     console.info('Analyzing benchmarks...');
     try {
         const statistics = await analyzeBenchmarksForVersion(versionToAnalyze);
         await reportResults(statistics, versionToAnalyze);
     } catch (error) {
-        const errorMessage =
-            error instanceof Error ? error.message : String(error);
-        await reportErrorResult(errorMessage, versionToAnalyze);
+        if (debugMode) {
+            // In debug mode, throw the error to get stack traces and immediate feedback
+            throw error;
+        } else {
+            // In normal mode, write error to markdown for PR visibility
+            const errorMessage =
+                error instanceof Error ? error.message : String(error);
+            await reportErrorResult(errorMessage, versionToAnalyze);
+        }
     }
     console.info(`Report generated at ${MARKDOWN_FILE}`);
 }
 
-runBenchmarkAnalysis(process.argv[2]);
+// Parse command line arguments
+const args = process.argv.slice(2);
+const debugMode = args.includes('--debug') === true;
+const versionArg = args.find((arg) => arg.startsWith('--') === false);
+
+runBenchmarkAnalysis(versionArg, debugMode);

.github/scripts/analyze_benchmarks/markdown.ts

@@ -16,12 +16,24 @@ export async function generateMarkdownReport(): Promise<string> {
         })
     );
 
-    return `# Benchmarks
+    return `<!--
+This file is auto-generated by .github/scripts/analyze_benchmarks/markdown.ts
+Do not edit this file manually. Changes will be overwritten.
+-->
+
+# Benchmarks
+
+## Purpose and Usage
+
+This document tracks high-level performance changes across Azle versions for comparison between releases.
+
+**For detailed benchmarks showing how different Candid data types perform, see the [benchmarks example](examples/stable/test/end_to_end/candid_rpc/benchmarks/benchmarks.md).** This is the best resource for understanding day-to-day performance in your Azle applications.
 
 ${generateVersionTables(sortedBenchmarks)}
 
 ---
-*Report generated automatically by Azle*`;
+
+_Report generated automatically by Azle_`;
 }
 
 function generateVersionTables(

.github/scripts/analyze_benchmarks/statistics.ts

@@ -23,11 +23,13 @@ const BASELINE_EFFICIENCY_WEIGHTS = {
  * @returns Statistical analysis of the benchmark data
  */
 export function calculateVersionStatistics(
-    entries: BenchmarkEntry[]
+    entries: BenchmarkEntry[] | undefined
 ): Statistics {
-    const instructions = entries.map((entry) =>
-        Number(entry.instructions.__bigint__)
-    );
+    if (entries === undefined || entries.length === 0) {
+        throw new Error('No benchmark entries found for this version');
+    }
+
+    const instructions = entries.map((entry) => Number(entry.instructions));
 
     return calculateStatistics(instructions);
 }

benchmarks.md

@@ -1,5 +1,11 @@
 # Benchmarks
 
+## Purpose and Usage
+
+This document tracks high-level performance changes across Azle versions for comparison between releases.
+
+**For detailed benchmarks showing how different Candid data types perform, see the [benchmarks example](examples/stable/test/end_to_end/candid_rpc/benchmarks/benchmarks.md).** This is the best resource for understanding day-to-day performance in your Azle applications.
+
 ## Version `0.30.0`
 
 <table>

src/stable/build/commands/build/javascript.ts

@@ -223,7 +223,7 @@ function experimentalMessage(importName: string): string {
 function handleBenchmarking(): string {
     return /*TS*/ `
         if (globalThis.process !== undefined && globalThis.process.env.AZLE_RECORD_BENCHMARKS === 'true') {
-            const methodMeta = canisterClassMeta.methodMeta;
+            const methodMeta = globalThis._azleCanisterClassMeta.methodMeta;
 
             const canisterMethodNames = Object.entries(methodMeta).reduce((acc, [key, value]) => {
                 if (value === undefined) {