Sync hot-patch-6.15 into development (Phase 1 of 2)#1655
Merged
jakejackson1 merged 123 commits intoMay 18, 2026
Conversation
(cherry picked from commit 996e527)
(cherry picked from commit cf026bf)
Gravity Forms 2.9.4 set their minimum requirements to PHP7.4+ (cherry picked from commit e84e591)
Make compatible with library updates included in Gravity Forms 2.9.7 and 2.9.8. Backwards compatible with < GF 2.9.7 too. See gravityforms/gravityforms#3208 See gravityforms/gravityforms#3196 (cherry picked from commit 062f3e4)
…ound processing library Rather than dispatch a bunch of individual queues, traditionally, Gravity PDF has treated a queue as a bunch of mini queues. In a new GF update, retry support has been added if a queue returns anything besides `false`, which caused big delays processing the mini queues in Gravity PDF. Rewriting Gravity PDF to convert the mini queues into individual queues would be a breaking change and a lot of development time. Instead, we'll merge all the mini queues together to be processed sequentially. Along with this adjustment, the retry logic has been reduced from 3 to 1. This allows for faster processing of a queue when an error does occur, since there will only be one extra timeout period instead of three. The unrecoverable error has also been removed, as this would prevent any other PDFs from generating or Notification emails being sent. The benefit to this is the Notification emails will still be sent out, even if a PDF isn't attached. The deprecation notice from GF 2.9.7 was addressed in 062f3e4 (cherry picked from commit 8a4bc03)
… doesn't already exist (cherry picked from commit 1fdde0a)
(cherry picked from commit 927bd8e)
(cherry picked from commit 7e169df)
(cherry picked from commit 46b4a3e)
Fixes display issues and PHP notice due to array key issues. (cherry picked from commit eb56ad3)
1. Move Gravity PDF info to bottom of report for quick viewing in support tickets 2. Don't translate labels in exported report 3. Don't translate outdated template export value 4. Display "unlimited" WordPress Memory in export value if PHP memory set to -1 Fixes #1597 (cherry picked from commit 6606e87)
- Replace Cache class with version that better accounts for directory permissions of the parent folder - Remove creation of unnecessary ttfontdata folder - Move override files/classes to new folder for easier development
Mpdf enhancements
Update coding-standards, deployment, javascript-tests, and phpunit workflows to current GitHub Actions patterns; add an automated job that tracks new WordPress core releases. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replace the legacy TestCafe runner and its bash-driven install helpers with a Playwright-based test suite under tests/playwright/, including shared fixtures and helpers in tools/playwright/. The new GitHub Actions workflow (playwright-e2e.yml) replaces end-to-end-tests.yml and runs Playwright sharded across multiple wp-env instances. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Relocate PHPUnit bootstrap, fixtures, and config under tools/phpunit, introduce dedicated wp-env profiles (development/integration/e2e) under tools/wp-env, refresh PHP and JS test suites for the current toolchain, and update Jest config and supporting mu-plugins so that tests run cleanly under WordPress 7.0 and PHP 8.5. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Rewrite the licensing layer end-to-end: refresh the EDD plugin updater integration, lift license state and helpers into Helper_Abstract_Addon so add-ons can share behavior, expose the updater on the data class for downstream code, and restructure the license settings UI and Model_Settings/Controller_Settings handlers to match. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add a Redact_Processor that masks license keys, download links, and other sensitive values before they reach the log file, and register it in Logger::pushProcessor alongside the existing handlers. Adjusts the Logger docblock to credit the move from Helper_Logger. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Introduce Helper_Interface_Extension_Uninstaller so add-ons can opt into a unified uninstall pipeline, expand Model_Uninstall to drive it, clear the new license/update cron hooks on deactivation, and add an upgrade routine that removes the legacy update cache during 6.14 → 6.15 upgrades. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
…y handling Tighten the system report to surface temp directory health and related configuration more clearly, and align View_System_Report with the new model output. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Route custom and core font validation through the new mPDF cache, align template management with the cleaner cache lifecycle, and harden Helper_Misc::cleanup_dir so it only deletes paths within resolvable, allow-listed folders. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Sweep src/assets/js/react and src/assets/js/admin to current React patterns and ES module conventions, and refresh the legacy entries bundle. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Move the bundled translation catalogs out of src/assets/languages into the WordPress-standard languages/ directory (with .l10n.php companions), regenerate the .po/.mo pairs for the shipped locales, and add a tools/potomatic helper script plus dictionary that drives the new translation workflow. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Apply a sweep of settings/field/view-layer cleanups across Helper_ classes, fields, models, views, bundled template configs, and the matching SCSS partials: translator-comment coverage, narrow visibility adjustments, and the remaining WordPress 7.0 / PHP 8.5 compatibility tweaks that the larger feature commits did not absorb. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Point the plugin bootstrap at the new build/ asset paths, load gravity-pdf-updater.php from the entry file, move the WordPress text-domain to /languages, switch the API URL to api.gravitypdf.com, adopt the WP 6.3+ defer script strategy, add a CLAUDE.md project guide, and bump the version metadata in pdf.php and api.php. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phase 1 of the hot-patch ↔ development reconciliation. Brings every
production fix and structural change from hot-patch-6.15 (releases
6.12.5 → 6.14.x) into development, which had been frozen since
2024-11-13. After this commit, hot-patch is retired and development
becomes the single trunk; future hot-fixes branch off development.
Conflict resolution
-------------------
* `.github/workflows/deployment.yml` — took hot-patch (drops the WP.org
deploy job entirely; dev's rename of that job was now obsolete).
* `src/Controller/Controller_Upgrade_Routines.php` — kept dev's 6.12.0
cleanup_tmp_dir hook clearing alongside hot-patch's 6.13.2
fix_tmp_folder_permissions and 6.15.0 remove_legacy_update_cache
routines (non-overlapping versions).
* `src/Helper/Helper_Options_Fields.php` — took hot-patch for every
conflicted hunk (translator comments, esc_html__ escaping, refreshed
doc URLs). Dev's auto-merged REST schema additions (validate /
sanitize callbacks for REST Form Settings) are preserved. Removed the
duplicate `desc` key that resulted from hot-patch inserting an
escaped variant near dev's pre-existing entry on the
logged_out_timeout setting.
* `src/Helper/Helper_PDF.php` — kept both imports
(GFPDF\Statics\Cache from dev + GFPDF\Helper\Mpdf\Request from
hot-patch).
* `src/Model/Model_PDF.php` — took hot-patch's three changes (refreshed
middleware doc URL, two-bucket cleanup_tmp_dir with separate mPDF /
template ages, more specific GP_Populate_Anything method_exists guard).
* `src/bootstrap.php` — took hot-patch (adds the GF async /
background-process class alias shim).
* `src/assets/js/admin/helper/spinner.js` — took hot-patch's formatting.
* `src/assets/js/admin/settings/common/setupDynamicTemplateFields.js`
— kept dev's actionToolbar refactor (renamed module + selector
`.form-action-toolbar-2` + 4th `true` param to insertAfter), and
layered hot-patch's `handleMergeTags` import and call after
`doMergetags()`.
* `src/assets/js/react/bootstrap/templateBootstrap.js` — took
hot-patch (adds `chosen:updated` trigger).
* `src/assets/js/react/gfpdf-main.js` — kept dev's preview /
unsaved-changes wiring (`previewButton`, `actionToolbar`,
`unsavedChangesWarning` imports + calls). Normalised the rest to
hot-patch's tabs/semicolons formatting.
* `src/assets/js/react/utilities/PdfSettings/actionToolbar.js` — kept
dev's renamed module (was hot-patch's `addEditButton.js`) with the
full `form-action-toolbar` clone behaviour.
* `src/assets/scss/Pages/_pdf-settings.scss` — kept dev's `//` comment
style and added hot-patch's
`.gform-settings-input__container--with-merge-tag` rule.
* `src/assets/scss/gfpdf-styles.scss` — kept dev's new
`MediaQueries/min-782` import; normalised every `@use` to
double-quoted paths.
Test infrastructure
-------------------
* Accepted hot-patch's relocation of test data into
`tools/phpunit/data/{fonts,entries,forms,pdf}/`. Dev's parallel move
to `tests/phpunit/data/{fonts,json}/` was reverted (the partial
restructure was now obsolete) and the legacy
`tests/phpunit/unit-tests/{fonts,json}/` tree was confirmed deleted on
both sides.
* Took hot-patch's versions of `tools/phpunit/bootstrap.php`,
`tools/phpunit/gravityforms-factory.php`, and the
`tests/phpunit/unit-tests/test-*.php` + `Test_Controller_Custom_Fonts.php`
files (they reference the new `tools/phpunit/data/` constants).
* Accepted hot-patch's deletion of the legacy TestCafe-style E2E suite
in `tests/e2e/` (dev had modified some of those files; hot-patch
replaced the whole layer with the Playwright tests under
`tests/playwright/` that were carried in automatically).
Deferred to Phase 2
-------------------
* `src/Controller/Controller_Pdf_Queue.php` — kept dev's queue
architecture (flattened queue + 7.0-style notification dispatch).
* `tests/phpunit/unit-tests/Controller/Test_Controller_Pdf_Queue.php`
— kept dev's matching tests.
The following hot-patch commits touched the queue stack and must be
re-applied on top of dev's queue refactor in a Phase 2 follow-up PR:
8a7914d Update PDF Background Processing to be compatible with the
updated background processing library (+ phpunit alias).
dc772da Update Background Processing Queue.
389ae6c Add queue cleanup when Background Processing toggled.
3fe1c1e Ensure background queue uses correct entry when resending
notifications.
Known follow-ups
----------------
* `composer.lock` and `package-lock.json` will need a refresh against
hot-patch's dependency manifests in a separate commit.
* Some i18n entries in `Helper_Options_Fields.php` still use plain
`__()` where hot-patch had `esc_html__()` (dev's auto-merged
regressions in non-conflict hunks). Not a behaviour regression vs
development today, but worth normalising in a small follow-up.
* `src/Helper/Log/MonoLoggerPsrLog2And3.php` requires PHP 8 syntax to
parse and is loaded conditionally via `Helper/Log/Logger.php`
(carried over from hot-patch unchanged).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
These five files (added on development for the REST API / PDF Preview / unsaved-changes feature) used the older Prettier config — 2-space soft indents, no semicolons. Hot-patch standardised on tabs + semicolons via wp-scripts. Auto-applied 'yarn lint:js --fix' after the merge to normalise. No behaviour change. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## development #1655 +/- ##
================================================
+ Coverage 77.74% 92.85% +15.10%
================================================
Files 255 58 -197
Lines 13925 1511 -12414
Branches 370 419 +49
================================================
- Hits 10826 1403 -9423
+ Misses 3091 103 -2988
+ Partials 8 5 -3 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
When resolving src/bootstrap.php in 607337e I took hot-patch wholesale (--theirs), which dropped dev's wiring for the new REST API controllers that were carried in for the 7.0 release: - $this->rest_api() call in Router::init() between gf_form_settings() and pdf() - rest_api() method that instantiates Rest_Form_Settings and Rest_Pdf_Preview, adds them to the singleton, and registers a rest_request_after_callbacks logger for /gravity-pdf/* routes Without this wiring the Rest classes existed on disk but never registered any routes, so the PDF Preview button and every Form-PDF CRUD endpoint silently failed. Re-applied dev's additions verbatim; all other hot-patch changes to bootstrap.php (GF async/background process alias shim, gfpdf_settings_sanitize toggle hook, copyright bump) are preserved. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three more issues found by auditing files I took with --theirs in the
Phase 1 merge:
1. test-slow-pdf-processes.php — when I took hot-patch's version
wholesale I lost two dev-added test methods that exercise the
end-to-end PDF generation pipeline:
* test_process_legacy_pdf_endpoint (covers Controller_PDF's
legacy gf_pdf endpoint, including its setExpectedIncorrectUsage
deprecation notices)
* test_process_pdf_endpoint (covers the modern endpoint via
$GLOBALS['wp']->query_vars['gpdf'])
Restored dev's full file, then patched the font-glob path to use
PDF_PLUGIN_DIR . '/tools/phpunit/data/fonts/' (hot-patch's
relocation) and bumped the copyright to 2026.
2. tests/phpunit/unit-tests/Rest/Test_Rest_Pdf_Preview.php — this
file was auto-merged in (dev added it new) but referenced
__DIR__ . '/../../data/fonts/' which was dev's relocation that
we reverted. Fixed to PDF_PLUGIN_DIR . '/tools/phpunit/data/fonts/'
so the mPDF fontDir resolves under hot-patch's data layout.
3. tests/e2e/advanced-checks/pdf-preview.test.js — orphaned TestCafe
test that survived the merge as an A-side add from dev. Imports
from the now-deleted tests/e2e/utilities/page-model/helpers/pdf
module and pulls in 'testcafe' which is no longer a dependency
on hot-patch. Deleted the file (and its now-empty parent dirs).
Test intent remains in git history; rewriting it as a Playwright
spec under tests/playwright/core/ is Phase 2 work.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
…t debt
PHPUnit found Test_PDF::test_cleanup_tmp_dir failing because the test
ages on dev side (1.01hr) were auto-merged in alongside hot-patch's
cleanup_tmp_dir implementation (which now uses a 12hr threshold for
template tmp + 1hr for mPDF tmp). Reverted the test ages to the merge
base / hot-patch values so file age vs threshold lines up:
test1=11.5hr (keep, <12hr)
test2=12.01hr (delete, >12hr)
test3=12.5hr (delete, >12hr)
test6=5hr (keep, <12hr)
Also closed the 10 pre-existing JS lint errors in dev's new 7.0 files
that CI would otherwise flag:
preview.js
- jsdoc/no-undefined-types: fixed 'int' -> 'number'
- jsdoc/require-returns-description: added one-line descriptions
- no-console: two console.error calls suppressed with reason
('surface fetch errors to the browser console for diagnostics')
formSettings.js
- jsdoc/require-param-type: typed 'schema' and 'formId'
- jsdoc/require-returns-description: added FormData description
- no-unused-vars: gform.addFilter is variadic and supplies the
schema arg even when unused; documented with disable + reason
actionToolbar.js
- no-shadow: renamed inner 'actionToolbar' to 'originalToolbar'
so it doesn't shadow the exported function
PHPUnit after these changes: 1118 tests, 4080 assertions, 1 failure
(Test_Controller_Pdf_Queue::test_failed_queue_tasks - expected Phase 2
follow-up), 8 skipped, 1 risky (pre-existing output buffer warning).
yarn lint:js: clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Both workflows gate themselves on the 'run-tests' label via the check_for_string_in_pull_request_comment job, but their pull_request trigger uses the default event types (opened, reopened, synchronize) — so adding the label after the PR was opened does NOT re-run them. playwright-e2e.yml already lists 'labeled' in its types; copying that pattern to phpunit.tests.yml and coding-standards.yml so the same label-flow works for all three test suites. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three classes of PHPCS errors surfaced after CI ran with the
'run-tests' label:
1. Auto-fixable formatting (26 of 28 errors), applied via phpcbf:
* api.php: 'static::' -> 'self::' on 18 scope-resolutions inside
the final GPDFAPI class (Universal.CodeAnalysis.StaticInFinalClass).
* api.php + Helper_Misc.php + Helper_Options_Fields.php: missing
space between 'function' keyword and '(' on 7 anonymous functions
(Squiz.Functions.MultiLineFunctionDeclaration.SpaceAfterFunction).
2. Reserved-keyword parameter names regressed in dev's edits to
GPDFAPI methods (the merge-base + hot-patch both used non-reserved
names; dev had quietly renamed them). Restored the non-reserved
names so PHP 8 named-arg callers and the lint sniff are both happy:
get_plugin_option(): $default -> $default_value
product_table(): $return -> $should_return
likert_table(): $return -> $should_return
Helper_Misc::flatten_array(): $array -> $items (dev's new method,
so no API stability concern)
(Universal.NamingConventions.NoReservedKeywordParameterNames.*)
3. Two false-positive nonce errors in GPDFAPI::add_pdf_font(). The
helper swaps $_FILES out for a synthetic payload built from its own
$font argument (programmatic API, not an unverified form submission),
so a nonce check is irrelevant. Targeted // phpcs:ignore comments
with a one-line reason on each (WordPress.Security.NonceVerification.Missing).
Note: 'composer phpcs:lint .' run from inside the worktree currently
reports 'no violations' because the ruleset excludes /.claude/* and
the worktree lives at .claude/worktrees/sync-development-from-hot-patch/.
CI checks out at /home/runner/work/gravity-pdf/gravity-pdf/ so the
exclude doesn't fire there - which is why these violations only
surfaced after the workflow ran. Re-running phpcs against a copy of
the files outside .claude/ now reports 0 errors, 1 pre-existing
readfile() warning in Model_PDF.php that hot-patch CI already
tolerates.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
…ry budget Helper_Pdf_Queue::task() retries a failing callback exactly once, so on a callback that always throws the mock is invoked twice (initial attempt + 1 retry). Hot-patch had already corrected this expectation in its branch; dev still had 'exactly(3)' which doesn't match the helper's behaviour on either branch. I deliberately deferred the *Controller_Pdf_Queue* file and its test to Phase 2 (architectural reconciliation of dev's flattened queue vs hot-patch's queue-update changes), but test_failed_queue_tasks exercises Helper_Pdf_Queue::task() — which is byte-identical on both branches — so it's not actually queue-architecture work; it's a straight test correction. PHPUnit was reporting: Tests: 1118, Assertions: 4080, Failures: 1, Skipped: 8, Risky: 1 across PHP 7.4 / 8.3 / 8.5. The 1 failure was this test; the 8 skips (EDD_SL_Plugin_Updater x6 + multisite x2) and 1 risky (output-buffer warning on Test_Slow_PDF_Processes::test_generate_pdf) are pre-existing on hot-patch and tolerated. Updated the @SInCE docblock to describe the actual behaviour. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
…ow_PDF_Processes risky
PHPCS (11 errors auto-fixed by phpcbf in dev's surviving 7.0 files —
all Squiz.Functions.MultiLineFunctionDeclaration.SpaceAfterFunction
on anonymous `function( ... )` declarations):
src/Rest/Rest_Form_Settings.php - 4 errors (lines 176, 972, 1019, 1084)
src/Statics/Cache.php - 1 error (line 94)
src/Statics/Debug.php - 1 error (line 52)
src/Controller/Controller_PDF.php - 5 errors (lines 121, 148, 200, 217, 224)
These were missed by my earlier phpcbf sweep because I only piped the
files that CI's *first* failed run named; CI then picked the next
batch of files once those cleared. The fixes are byte-identical to
what phpcbf produces, so style-equivalent to hot-patch's existing JS
files.
Test_Slow_PDF_Processes::test_generate_pdf risky:
beStrictAboutOutputDuringTests=true in tools/phpunit/config.xml fails
any test that leaves the output-buffer stack different from where it
started. Helper_PDF::pre_stream_actions() (run before mPDF streams)
drains *every* active output buffer:
while ( ob_get_level() > 0 ) {
ob_end_clean();
}
So by the time generate_pdf() trips the wp_die() error path and the
WPDieException bubbles up to the test, PHPUnit's own buffers have
already been torn down. The risky warning then fires on test teardown.
Snapshot the buffer depth before the call and re-establish it
afterwards in both directions: close any extras above the snapshot,
then open empties below it. Local PHPUnit run is now clean:
'OK (1 test, 2 assertions)' with no risky note. Hot-patch did not
need this guard because it lacked the dev-side setExpectedIncorrectUsage
call that interacted with the buffer state.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
CI was exiting with code 1 on two warnings: 1) src/Model/Model_PDF.php:2694 - readfile() The line already had a phpcs:ignore comment but it referenced the *old* WPCS sniff slug (file_system_read_readfile). The sniff was renamed to file_system_operations_readfile when WordPressCS was refactored, so the existing ignore was a no-op. Updated the slug and added a reason: we're streaming the generated PDF straight to the response and WP_Filesystem has no streaming equivalent. 2) src/Rest/Rest_Form_Settings.php:972 - unused $item param This is a closure bound to the 'gfpdf_template_for_current_page' filter. The filter passes its current value as the first arg, but we intentionally ignore it and force-return the captured $template so the REST schema can be regenerated for an arbitrary template. Added a phpcs:ignore for Generic.CodeAnalysis.UnusedFunctionParameter.Found with a one-line reason rather than renaming the param (the filter signature is fixed). PHPCS exit code is now 0 against a copy of these files outside .claude/. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
…op dead ob_end_clean loop * tools/phpcs/config.xml: add Generic.CodeAnalysis.UnusedFunctionParameter.Found to the WordPress-Extra <exclude> list (alongside its existing ...FoundAfterLastUsed sibling). PHP closures bound to WP filters / REST callbacks routinely have to accept args dictated by the caller signature, and the per-line phpcs:ignore noise wasn't paying for itself. Reverted the targeted ignore on src/Rest/Rest_Form_Settings.php::get_template_schema() now that the sniff is off. * tests/phpunit/unit-tests/test-slow-pdf-processes.php: drop the 'while ob_get_level() > $initial_ob_level' guard. The only thing generate_pdf() does to the buffer stack via pre_stream_actions() is drain it, so the buffer depth can only land at or below the initial snapshot - the open-up loop is sufficient on its own. Kept the ob_start() loop and reworded the comment. PHPCS exit 0; phpunit --filter test_generate_pdf: 'OK (1 test, 2 assertions)' - no risky warning. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
…ommits Review pass — trimmed AI-slip in the post-merge fix-up commits: * tests/phpunit/unit-tests/test-slow-pdf-processes.php — collapsed the 5-line block comment above the ob_get_level() snapshot to one line. The body says how; the comment now just says why. * tests/phpunit/unit-tests/Controller/Test_Controller_Pdf_Queue.php — collapsed parenthetical-narration docblock back to one line. * src/assets/js/react/api/preview.js — trimmed JSDoc @return text and dropped paragraph-length 'surface … to the browser console' justifications from two eslint-disable-next-line no-console comments. * src/assets/js/react/utilities/PdfSettings/formSettings.js — trimmed JSDoc descriptions and shortened the two eslint-disable-next-line no-unused-vars reasons. * api.php — shortened the two phpcs:ignore reasons on the swap inside add_pdf_font(). * src/Model/Model_PDF.php — shortened the readfile() phpcs:ignore reason. PHPCS exit 0; yarn lint:js clean; phpunit --filter 'test_generate_pdf|test_failed_queue_tasks' passes. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Phase 1 of the hot-patch ↔ development reconciliation. Brings every production fix and structural change from
hot-patch-6.15(releases 6.12.5 → 6.14.x, 101 commits) intodevelopment, which had been frozen since 2024-11-13. After this lands, hot-patch is intended to be retired and development becomes the single trunk — future hot-fixes branch off development.composer install,yarn install --frozen-lockfile,yarn build,yarn lint:css,yarn test:jsall green.yarn lint:jsshows 10 pre-existing errors (verified on both source branches —actionToolbarself-shadow + JSDoc gaps in dev's new files); the 217 prettier diffs were auto-fixed in commitd2baecec.src/Controller/Controller_Pdf_Queue.phpand its test. See Phase 2 follow-up below.Conflict resolution (per file)
.github/workflows/deployment.yml— took hot-patch (drops the WP.org deploy job; dev's rename was obsolete).src/Controller/Controller_Upgrade_Routines.php— kept dev's 6.12.0cleanup_tmp_dirhook clearing alongside hot-patch's 6.13.2fix_tmp_folder_permissionsand 6.15.0remove_legacy_update_cacheroutines.src/Helper/Helper_Options_Fields.php— took hot-patch for every conflicted hunk (translator comments,esc_html__escaping, refreshed doc URLs). Dev's auto-merged REST schema additions (validate/sanitize callbacks for REST Form Settings) are preserved. Removed the duplicatedesckey that resulted from hot-patch inserting an escaped variant near dev's pre-existing entry on thelogged_out_timeoutsetting.src/Helper/Helper_PDF.php— kept both imports (GFPDF\Statics\Cachefrom dev +GFPDF\Helper\Mpdf\Requestfrom hot-patch).src/Model/Model_PDF.php— took hot-patch's three changes (refreshed middleware doc URL, two-bucketcleanup_tmp_dirwith separate mPDF / template ages, more specificGP_Populate_Anythingmethod_existsguard).src/bootstrap.php— took hot-patch (adds the GF async / background-process class alias shim).src/assets/js/admin/helper/spinner.js— took hot-patch's formatting.src/assets/js/admin/settings/common/setupDynamicTemplateFields.js— kept dev'sactionToolbarrefactor (renamed module + selector.form-action-toolbar-2+ 4thtrueparam toinsertAfter), and layered hot-patch'shandleMergeTagsimport + call afterdoMergetags().src/assets/js/react/bootstrap/templateBootstrap.js— took hot-patch (addschosen:updatedtrigger).src/assets/js/react/gfpdf-main.js— kept dev's preview / unsaved-changes wiring (previewButton,actionToolbar,unsavedChangesWarningimports + calls). Normalised to hot-patch's tabs/semicolons formatting.src/assets/js/react/utilities/PdfSettings/actionToolbar.js— kept dev's renamed module (was hot-patch'saddEditButton.js) with the fullform-action-toolbarclone behaviour.src/assets/scss/Pages/_pdf-settings.scss— kept dev's//comment style and added hot-patch's.gform-settings-input__container--with-merge-tagrule.src/assets/scss/gfpdf-styles.scss— kept dev's newMediaQueries/min-782import; normalised every@useto double-quoted paths.Test infrastructure
tools/phpunit/data/{fonts,entries,forms,pdf}/. Dev's parallel move totests/phpunit/data/{fonts,json}/was reverted (the partial restructure was now obsolete); the legacytests/phpunit/unit-tests/{fonts,json}/tree was confirmed deleted on both sides.tools/phpunit/bootstrap.php,tools/phpunit/gravityforms-factory.php, and thetests/phpunit/unit-tests/test-*.php+Test_Controller_Custom_Fonts.phpfiles (they reference the newtools/phpunit/data/constants).tests/e2e/(dev had modified some of those files; hot-patch replaced the whole layer with the Playwright tests undertests/playwright/that were carried in automatically).Phase 2 follow-up
The following hot-patch commits touched the queue stack and were deliberately not carried in this PR — they must be re-applied on top of dev's queue refactor in a separate PR:
8a7914dddc772da0389ae6ce3fe1c1e7Known follow-ups (smaller)
Helper_Options_Fields.phpstill use plain__()where hot-patch hadesc_html__()(dev's auto-merged regressions in non-conflict hunks). Not a behaviour regression vs current development, but worth normalising.src/Helper/Log/MonoLoggerPsrLog2And3.phprequires PHP 8 syntax to parse and is loaded conditionally viaHelper/Log/Logger.php— carried over from hot-patch unchanged.Test plan
71bf214c)yarn buildproduces validbuild/assetstests/playwright/runs againsttools/wp-env/e2e.json51d85354) and verifygfpdf-*.logfor surprises from the PSR Log refactor9137e349left the GitHub release workflow producing a valid plugin zip🤖 Generated with Claude Code