Performance: let debug mode serve fresh woven classes from OPcache (debug hot request 13 ms vs 5.5 ms in production)

オープン
  • 何か問題がありますか？
#814 コメント 1 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

まだ誰も着手していません。

評価

難易度
5/5
見積もり時間
1週間以上
初心者へのやさしさ
28/100
issue の種類
機能追加
明瞭さ
おおむね明確
活発さ
活発
技術スタック
php
領域
backend, performance

調査の方向性

Start by reading follow-up issue #745, the freshness work tracked in #815, and the transformation performance test named in the acceptance criteria. Compare the three proposed loading approaches, focusing on XDebug breakpoint mapping and magic constants before proposing a design. Done means an approved approach preserves woven output, cache maps, and magic constants; avoids woven-file reads and recompilation on hot debug requests; and still re-weaves changed sources.

索引モデルが issue の本文から書いたものです。

説明

Enhancement Performance To Be Discussed

Implementation note: start in plan mode. XDebug breakpoint mapping and magic constants need an approved design before any code is written.

Measured with goaop/shop-demo (PHP 8.5.12, framework b755e57). Follow-up of #745, which made fresh untransformed files opcache-cached in debug mode and deliberately kept woven files on the filter.

Problem

In debug mode every woven class is loaded through php://filter/read=go.source.transforming.loader/resource=<original>. OPcache never caches a filtered include, so on every request each woven class:

  • is checked for freshness (filemtime, filesize, isFreshSince());
  • has its cached copy read with file_get_contents();
  • is compiled from scratch.

Measurement (hot request, OPcache warm, 94 woven classes)

debug => true debug => false
request time 13.0 ms 5.5 ms
filemtime / filesize calls 323 / 239 0 / 0
file_get_contents calls 94 0
profiled share of class loading + cache metadata 20 % 6 %

(323 filemtime includes ~84 redundant re-stats of tracked resources, tracked in #815.)

Why the filter is used today

Debug mode serves woven code through the filter so that PHP reports the original path for the included file. XDebug breakpoints set in the original source then hit, and __FILE__/__DIR__ of the woven body point at the original file (#679).

Options to evaluate in the plan

  1. Opt-in flag (e.g. a Features:: bit or a debug sub-option): include fresh woven files by their cache path, like production. Breakpoints then have to be set in the cache files, or mapped by the IDE (path mappings), and magic constants are already rewritten by MagicConstantTransformer.
  2. Automatic: use the cache path when no step debugger is active (xdebug_info('mode') without debug, or extension_loaded('xdebug') === false), and the filter otherwise. This keeps the freshness checks either way.
  3. Keep the filter but make it cheaper (no file_get_contents() of a separate file, e.g. by reading the cached content through the stream directly). The compile cost remains.

Acceptance

  • Woven output, cache maps and magic constants identical with and without the new path.
  • With the new path, a hot debug request does no file_get_contents() for woven classes and no compilation (OPcache hits), measured with the shop demo or the fixture project.
  • A changed source is still re-woven on the next debug request.
  • composer test, composer analyze; composer test:performance:transformation unchanged.

Estimate

Up to −7 ms per debug request on this app (proportional to the number of woven classes loaded). High: debugger integration semantics, design decision, cross-platform paths.

Generated by Claude Code

主要言語
PHP
スター
1.7k
フォーク
166
平均マージ
1日 1時間
マージ済み PR（30日）
42

環境構築

このプロジェクトには開発コンテナ、Dockerfile、コントリビューションガイドがありません。まず README を読み、一般的な手順ははじめてのコントリビューションガイドを参照してください。

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

goaop/framework のほかの issue

goaop/framework の issue をすべて見る

似ている issue

PHP の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。