Fix native PHP sites loading assets from remote siteurl/home (STU-1925) - #3988
Conversation
| // PHP's built-in server ignores the auto_prepend_file ini directive when a | ||
| // router script is in use, so apply it ourselves before dispatching. This is | ||
| // how Studio's pre-boot prepend (local WP_HOME/WP_SITEURL) and reprint's | ||
| // runtime.php for imported sites get a chance to run. |
There was a problem hiding this comment.
Yep, I see the same thing. Some PHP 8.4 release seems to have changed this behavior.
require_once seems to work really well in this context. With this change, auto_prepend_file as a config directive just keeps working, regardless of PHP version. Nice 👍
📊 Performance Test ResultsComparing 7f11e26 vs trunk app-size
site-editor
site-startup
Results are median values from multiple test runs. Legend: 🟢 Improvement (faster) | 🔴 Regression (slower) | ⚪ No change (<50ms diff) |
fredrikekelund
left a comment
There was a problem hiding this comment.
Good catch @shaunandrews and nice work, @bcotrim 👍 This is a nice and targeted fix, and I'm glad we found the auto_prepend_file discrepancy between PHP versions.
Let's generate the siteurl / home URL deterministically (since we already know which value we want). Other than that – this LGTM
| $studio_is_https = ( ! empty( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] !== 'off' ) | ||
| || ( ! empty( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) && stripos( $_SERVER['HTTP_X_FORWARDED_PROTO'], 'https' ) !== false ); |
There was a problem hiding this comment.
The PHP child processes never use HTTPS. Only the HTTPS proxy does that in in Studio. Might as well remove this part
| if ( PHP_SAPI !== 'cli' && ! empty( $_SERVER['HTTP_HOST'] ) ) { | ||
| $studio_is_https = ( ! empty( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] !== 'off' ) | ||
| || ( ! empty( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) && stripos( $_SERVER['HTTP_X_FORWARDED_PROTO'], 'https' ) !== false ); | ||
| $studio_local_url = ( $studio_is_https ? 'https' : 'http' ) . '://' . $_SERVER['HTTP_HOST']; |
There was a problem hiding this comment.
I mentioned this on Slack, too, but I think we should generate this URL deterministically instead of looking at the Host header. There's just no reason to use a dynamic value here, since we already know the complete URL we want for siteurl and home
| const hash = crypto.createHash( 'sha1' ).update( sitePath ).digest( 'hex' ).slice( 0, 16 ); | ||
| const prependPath = path.join( os.tmpdir(), `studio-siteurl-prepend-${ hash }.php` ); |
There was a problem hiding this comment.
This is fine, but we could use fs.mktempd for a similar effect, instead of generating a hash ourselves https://nodejs.org/docs/latest/api/fs.html#fspromisesmkdtempprefix-options
| // Define WP_HOME/WP_SITEURL before WordPress boots so the site serves from | ||
| // the local URL regardless of the siteurl/home in the DB (e.g. after pulling | ||
| // a remote site — STU-1925). Pre-boot so derived URLs (WP_CONTENT_URL, etc.) | ||
| // resolve locally too. The URL is the site's configured local URL (custom | ||
| // domain or http://localhost:PORT), matching the Playground runtime's | ||
| // --site-url, rather than the request Host header. |
There was a problem hiding this comment.
I think we could slim this down, or indeed even remove it. Some of this is encoding the AI agent session history
fredrikekelund
left a comment
There was a problem hiding this comment.
Confirming that the latest changes look good and test well 👍
… (#3988) ## Related issues - Fixes STU-1925 ## How AI was used in this PR Claude Code investigated the root cause, implemented the fix, and verified it end-to-end against a pulled site (cotrim.dev). I reviewed the diff and tested manually via the CLI. ## Proposed Changes When a site pulled from a remote (e.g. a WordPress.com staging site) runs on the native PHP runtime, it kept loading assets from the remote URL because the database's `siteurl`/`home` were served as-is. Native PHP now defines `WP_HOME`/`WP_SITEURL` before WordPress boots — derived from the request, so it survives dynamic ports and custom domains and never mutates the database (a push still sends the original remote URL). This also fixes a latent issue: PHP's built-in server ignores `auto_prepend_file` when a router script is used, so reprint's `runtime.php` never ran for imported native sites — the router now applies it explicitly. ## Testing Instructions 1. On the **native PHP** runtime, use a site whose database holds a remote `siteurl` (pull a WP.com staging site, or set `siteurl`/`home` to a remote URL in the DB). 2. Start the site, open it, and check DevTools → Network. 3. **Before:** theme/plugin assets (e.g. `style.css`) load from the remote domain. **After:** all assets load from `http://localhost:<port>`, and `wp-json` reports the local `url`/`home`. 4. `wp-cli` still reports the stored URL (intentionally untouched). ## Pre-merge Checklist - [x] Have you checked for TypeScript, React or other console errors?
## Related issues <!-- Link a related issue to this PR. If the PR does not immediately resolve the issue, for example, it requires a separate deployment to production, avoid using the "Fixes" keyword and use "Related to" instead. --> - N/A ## How AI was used in this PR Claude compared previous release-note sections and release-note PR descriptions, and drafted the curated 1.12.0 wording. The wording was reviewed in-thread before updating the PR. ## Proposed Changes - Replace the generated `1.12.0` release-note block with a shorter, user-facing summary. - Added the two PRs cherry-picked onto the release branch after code freeze: #3974 and #3988 - Folded all dependency/dependabot bumps into a single "Updated multiple dependencies" line. - Omitted internal-only PRs (test/CI infra, `AGENTS.md`/`STUDIO.md` docs, pure refactors, and the experimental Studio Web / "hosted" groundwork). ## Testing Instructions - Read the `1.12.0` section of `RELEASE-NOTES.txt` - Confirm the groupings and wording read well and are accurate - Cross-check against the merged-PR list for anything important that should be surfaced or reworded ## Pre-merge Checklist - [ ] Have you checked for TypeScript, React or other console errors?
…per-command CLI shutdown (#4082) ## Related issues - AINFRA-2588 (Investigate Studio Windows E2E hangs in Buildkite) - Supersedes the split verification work in #4075; overlaps with #4070 (included as-is) and #4061 (its tilde fix is included; its daemon isolation is not) ## Proposed Changes Windows E2E has been broken since June 29 by two independent regressions that merged the same day. This PR carries the minimal combination that fixes both, plus CI reporting fixes so failures can't hide: **1. Per-command CLI shutdown handling (from #4070).** #3954's shared `killAll()` on `will-quit` runs one listener after the quit handler that spawns `site stop --all`, killing the stop command before it stops any site. Sites leaked into the machine-global process-manager daemon until its capacity cap was exhausted — the 3-hour hangs. Restoring per-child quit handlers lets the quit-time stop survive; verified on CI: quit-time stops complete in under a second (previously a 20-second timeout on every session) and zero capacity errors. **2. PHP INI tilde fixes.** #3988 passed the site-url prepend file (reprint's runtime: constants + SQLite loader) to PHP as an unquoted `-d auto_prepend_file=` value. On machines where the temp path contains a Windows 8.3 short name (e.g. `C:\Users\BUILDK~1\...` — any username over 8 characters), PHP's INI parser fails on the `~` (`syntax error, unexpected '~'`), keeps only the prefix, and every request dies with `Fatal error: Failed opening required 'C:\Users\BUILDK'` before WordPress boots. This broke every page load of every native-PHP site on affected machines and caused the ~25 Windows E2E failures. Fixed at both ends: - `auto_prepend_file` is now quoted and backslash-normalized via the existing `toPhpIniPath()`, like every other path directive. - `getPhpSafeTmpDir()` resolves the Windows short name to its long form for every temp path handed to PHP (opcache dir, phpMyAdmin config/sessions, site-url prepend dir). **3. Honest CI reporting.** The mac/Windows/Linux E2E jobs all posted to the same "E2E Tests" GitHub status and the last writer won, so a fast green mac job masked a failing or still-running Windows job. The notify now lives on the E2E group: one status, pending until every platform finishes. Also: `run-e2e-tests.sh` traps termination so a canceled/timed-out job can't record exit status 0 and turn the build green (observed in build 18744). **4. Windows E2E re-enabled** with a 100-minute job cap. Prior verification of these fixes together (#4075, build 18789): 47 passed / 0 failed / 26 minutes — the first green Windows E2E since June 29. This PR's own CI re-verifies the combination as extracted here. Deliberately not included, pending their own review: #4041 (bounded daemon socket requests, daemon force-settle, leaked-daemon reaping) and #4061's per-home daemon isolation. This PR's CI run doubles as the experiment showing whether they are required for green E2E or are hardening. ## Testing Instructions - **CI**: all three E2E platforms should pass; the "E2E Tests" GitHub status stays pending until mac, Windows, and Linux all finish, then reports one combined result. The Windows job should show no `syntax error, unexpected '~'` in daemon logs, no `site stop --all command timed out` lines, and no `CAPACITY_LIMIT_REACHED` errors. - **Unit**: `npm test -- apps/cli/tests/ apps/studio/src/tests/` passes. - **Manual (Windows)**: on a machine whose user profile path gets 8.3-mangled (username over 8 characters), create and open a native-PHP site — pages render instead of a PHP fatal. ## Pre-merge Checklist - [x] Have you checked for TypeScript, React or other console errors? --------- Co-authored-by: Gergely Csecsey <gergely.csecsey@automattic.com>



Related issues
How AI was used in this PR
Claude Code investigated the root cause, implemented the fix, and verified it end-to-end against a pulled site (cotrim.dev). I reviewed the diff and tested manually via the CLI.
Proposed Changes
When a site pulled from a remote (e.g. a WordPress.com staging site) runs on the native PHP runtime, it kept loading assets from the remote URL because the database's
siteurl/homewere served as-is. Native PHP now definesWP_HOME/WP_SITEURLbefore WordPress boots — derived from the request, so it survives dynamic ports and custom domains and never mutates the database (a push still sends the original remote URL). This also fixes a latent issue: PHP's built-in server ignoresauto_prepend_filewhen a router script is used, so reprint'sruntime.phpnever ran for imported native sites — the router now applies it explicitly.Testing Instructions
siteurl(pull a WP.com staging site, or setsiteurl/hometo a remote URL in the DB).style.css) load from the remote domain. After: all assets load fromhttp://localhost:<port>, andwp-jsonreports the localurl/home.wp-clistill reports the stored URL (intentionally untouched).Pre-merge Checklist