- In shadow_read, TMDO_Hook_Bus::intercept_get() returned the flat value.
update_metadata() reads the old value through that filter after the Hook
Bus has upserted the new one, so WP saw old === new and never wrote native
meta: every update in shadow_read was lost on demotion (dev36 2026-09-26).
shadow_read now returns $check (native) and compares flat on a sample of
reads (filter tmdo_shadow_compare_sample_rate, default 0.1). aeav_only is
unchanged; the post query router still routes in shadow_read.
- Live suite shadow_write: xfail markers removed, HivePress listing case
added (new int key hp_rating_count); fixtures gain post() and a
post_type-scoped field()/flat().
- Verify: live shadow_write + serialized 8/8 PASS (old code: 4 FAIL);
unit 685 OK, integration 484 OK, phpcs (phpcs.xml) 0, stan OK
- PR: PR-056
- TMDO_Type_Caster: text / textarea values that are arrays or objects are
stored with maybe_serialize() (was (string) → the literal "Array", which
bridge-restore-native then wrote back over native meta); from_db()
unserializes them with TMDO_Safe_Unserialize. text columns are
VARCHAR(255): longer values are stored NULL (native keeps them) and
logged as text_overflow.
- Add `wp tmdo repair-array-literals [--entity] [--confirm]`
(TMDO_Array_Literal_Repair): rewrites flat "Array" values from native
meta and lists native rows that are "Array" themselves (backup needed).
- Verify: TypeCasterSerializeTest 13 OK, ArrayLiteralRepairTest 3 OK;
unit 685 OK, integration 483 OK, phpcs (phpcs.xml) 0, stan OK;
live `serialized` 3 PASS; dev36 repair dry-run lists the value corrupted
by the 2026-09-26 round trip (user #1 community-events-location)
- PR: PR-055
- TMDO_Cache_Layer::prefetch() / warm_post() (and so on_save_post /
on_loop_start) queried wp_wpdo_cold_{post_type} whenever cold fields
were registered, even with the cold module idle and the table never
created: every HivePress / WC save logged "table doesn't exist" errors.
They now return unless the module is write-active and the table exists.
- CacheLayerTest activates the module in setUp; new cases assert zero cold
queries while idle or when the table is missing.
- Verify: CacheLayerTest 12 OK; unit 671 OK, integration 480 OK,
phpcs (phpcs.xml) 0, stan OK; dev36 live: seeding listings/products/
vendors with cold_hp_listing idle → 0 wpdo_cold errors (was hundreds)
- PR: PR-054
- Add TMDO_Promotion_Guard: HivePress models read every meta key with an
empty-key get_*_meta() and WooCommerce does the same for products (and
writes stock/sales with direct SQL), so on those sites native meta must
stay authoritative. Promotion to aeav_only is refused in
TMDO_Mode_Manager::set() (covers REST, CLI, phases, auto-promoter) for
post/user/term/comment with HivePress and post/user with WooCommerce;
TMDO_Post_Migration::cleanup() and the cleanup phase refuse too.
Demotion stays open; filter tmdo_promotion_guard_allow overrides.
- Admin notice lists guarded entities already in aeav_only with the
restore-native command (no forced demotion on read).
- Verify: PromotionGuardTest 8 OK; unit 668 OK, integration 480 OK,
phpcs (phpcs.xml) 0, stan OK; dev36 live: 4 promotions + post-cleanup
refused, wpdo_bridge_modes unchanged
- PR: PR-053
- Add tests/live: a `wp eval-file` runner (PASS/FAIL/XFAIL/XPASS/SKIP) with
fixtures that override entity modes in process (pre_option) and read
native / flat values with plain SQL, so WP core meta flows run for real
- shadow_write suite reproduces the dev36 finding: in shadow_read,
update_*_meta returns false and native meta is never written
(3 XFAIL marked for PR-056); the dual_write control passes
- Move the dev36 integration harness into tests/live/tools; outputs go to
cwd / $TMDO_LIVE_OUT, site URL and path are env-configurable
- Verify: wp --path=<dev36> eval-file tests/live/run.php shadow_write
→ 1 passed, 3 xfail; unit 658 OK, integration 480 OK,
phpcs (phpcs.xml) 0, stan OK
- PR: PR-052
- CI PHPStan (level 6, baseline for pre-existing debt) failed on 35 errors in
code added on hardening/2.0: missing iterable value types (PHPDoc only), a
redundant method_exists() in uninstall.php, and the new WP-CLI commands
registered as array('Class', 'instanceMethod') — now array(new Class(), ...)
- Local runs had been masked by a PHPStan tmpDir shared with the act CI runner
(cached container pointing at a removed phar); use a private tmpDir
- Verify: phpstan (private tmpDir) no errors, phpcs clean on changed files,
phpunit Unit 658 OK; dev22 (running this branch) CLI post-status-sync /
bridge-restore-native dry runs OK
- retire() read native meta with wpdb::get_col(), which (via get_var())
returns null for an empty string, so a flat '' written back as '' was
counted as a mismatch and blocked the retirement. It now reads raw rows
with get_results().
- The integration test wpdb stub's get_var() / get_col() now read '' as
null like the real wpdb, so tests see WordPress behaviour.
- Found writing back the sso group on dev21 / dev22 (picture_url = '').
- Verify: core unit 658 / integration 480 green, phpcs clean
- PR: PR-042 follow-up
- missing_integration_addons() no longer asks for the SSO AddOn when
2meet-hub-core / 2meet-spoke-sso are active: they register their own
tables and the `sso` values belong in usermeta (benchmark: no
measurable gain from the flat table for by-user-id access).
- `sso` joins the deprecated AddOns; for hub / spoke / sso the notice says
they go without replacement and to run
`wp tmdo bridge-restore-native user --group=sso --retire` first.
- Verify: core unit 658 / integration 479 green, phpcs clean
- PR: PR-043
- TMDO_Native_Restorer::run() takes an optional group; the CLI gains
`bridge-restore-native <entity> --group=<g>`.
- New retire(): once native meta holds every non-NULL value of a group's
flat table (checked against the persisted field definitions), forget
its registry_meta row and rename the table to <table>_retired_<Ymd>, so
orphan restore no longer routes its keys to the flat table. Refused
while a plugin still registers the group or any value differs.
CLI: `--group=<g> --retire`.
- For retiring the user `sso` group (SSO AddOn deprecation).
- Verify: core unit 659 / integration 479 green, phpcs clean
- PR: PR-042
- TMDO_Member_Fields no longer registers user/sso; the definition moved
unchanged (hash 03543f19) to 2meet-data-optimizer-sso-addon.
- missing_integration_addons(): 2meet-hub-core or 2meet-spoke-sso without
the SSO AddOn is reported (admin notice + Site Health).
- The deprecated-AddOn notice points hub / spoke users to the SSO AddOn.
- Sites must activate the SSO AddOn before this ships (PR-041); aeav_only
sites without it fall back to orphan-group restore.
- Verify: core unit 659 / integration 475, SSO 6, HP 142/49, WC 10/22
green, phpcs clean
- PR: PR-040
- TMDO_Post_Query_Router drops SQL_CALC_FOUND_ROWS from routed searches
(posts_request) and answers found_posts_query from a COUNT(*) of the
same query, cached in the object cache under posts' last_changed.
- Hook Bus flat post writes and deletes move posts' last_changed: in
aeav_only they skip update_metadata()'s own hooks, so query caches
would otherwise outlive the write.
- Bench (1M): cached page 8 ms (was 694 / 363 ms for scenarios 1 / 2);
after invalidation 8 ms + COUNT 608 / 344 ms, on par with before.
- REST X-WP-Total COUNT is on the frozen Hot zone path (ADR-003): not
changed.
- Verify: core unit 659 / integration 475, HP 142/49, WC 10/22 green,
phpcs clean
- PR: PR-030
- TMDO_Post_Query_Router: when a group's post_status / post_date copies
are ready and none is NULL (one index probe per group per request), a
single-post-type search with explicit statuses and date ordering INNER
JOINs the flat table, filters the copied status, orders by the copied
post_date and skips GROUP BY. Otherwise the query keeps its old shape.
- New posts_orderby filter; bench gains an `entity+copy` variant.
- Bench (1M, with SQL_CALC_FOUND_ROWS): scenario 1 p95 3,739 -> 773 ms
(p50 670 -> 770); scenario 2 p50 356 -> 448 ms. Without found-rows the
same shape runs in 8 ms (PR-030).
- Verify: core unit 659 / integration 471 green, phpcs clean
- PR: PR-029
- TMDO_Post_Status_Sync: post group tables get post_status / post_date
columns (posts' collation) and an (post_status, post_date) index.
Brand-new empty tables get them on creation; existing tables only via
`wp tmdo post-status-sync [--dry-run] [--batch]`, never as DDL on a
populated table in a web request. Per-group state in
tmdo_post_status_columns ('columns' | 'ready').
- transition_post_status refreshes the copies; Hook Bus upserts that
create a post row fill them in (existing rows keep theirs).
- Bench (1M rows): add columns 0.05 s, index 5.3 s, batched backfill 32 s,
0 mismatches afterwards.
- Verify: core unit 659 / integration 466 green, phpcs clean
- PR: PR-028
- ADR-003 step 4 (core side): TMDO_Schema_Registry::register() raises
_doing_it_wrong() the first time each provider registers a zone=hot
field in a request (zone omitted defaults to hot). Fields still
register, so existing hot data stays readable.
- Cold / warm / archive registrations are unaffected.
- Test bootstrap's _doing_it_wrong stub now records calls.
- Verify: core unit 658 / integration 459, HP 142/49, WC 10/22 green,
phpcs clean
- PR: PR-038
- register_group() accepts $args['indexes'] (name => ordered keys); indexes on
unknown keys are dropped. Kept outside the field definition, so schema
hashes are unchanged.
- Schema_Manager::ensure_group_indexes() adds missing cidx_* indexes to the
existing flat table in place; the applied definition is remembered in
tmdo_entity_indexes so later requests skip SHOW INDEX.
- Bench (1M listings, scenario 2): Entity p50 337 ms -> 278 ms with
(hp_verified, hp_price), on par with Hot (269 ms).
- Verify: core unit 656 / integration 459 green, phpcs clean
- PR: PR-027
- Compare the Hot zone and Entity bridge (capabilities, 1M-listing benchmark, per-site usage: no 2meet-DO site uses the Hot zone; only legacy dev34 / dev10 hold Hot copies of postmeta)
- Decision (user, 2026-09-26): converge on Entity; freeze the Hot zone, skip PR-026, keep PR-027, move Hot-only HivePress fields into new Entity groups, stop Hot registrations
- PR: PR-025
- TMDO_Native_Restorer copies an entity's flat-table values into its native meta table so the value each key reads today is what native returns afterwards: the owning group (by post type for posts) wins, values are stored as WP stores them, NULL removes the native row
- Resumable per-group id cursor, catch-up pass for rows updated during the run, optional demote to dual_write; CLI wp tmdo bridge-restore-native <entity> [--dry-run] [--batch=<n>] [--demote]
- Verify: NativeRestorerTest (5, user) + NativeRestorerPostTest (2, scoped vs legacy routing); core unit 652 / integration 456, HP 148, WC 10 OK; scale estimate dev21 user ~5.1M / post ~4.0M flat rows
- PR: PR-031
- TMDO_Compatibility::missing_integration_addons() reports active HivePress / WooCommerce whose AddOn bootstrap class is absent
- Surface it as Site Health test wpdo_missing_integration_addons (also on the admin Doctor tab) and an administrator notice (TMDO_Missing_Addons_Notice)
- Verify: MissingIntegrationAddonsTest (5, class stand-ins in separate processes); core unit 651 / integration 449, HP 148, WC 10 OK; read-only check: dev20/21 HivePress+WooCommerce, dev22 HivePress, others clean
- PR: PR-035
- Add docs/PARTNER-REGISTRATION.md: TMDO_Anti_EAV_Aware bridge setup (plugins_loaded:3), table and entity-group registration with post_types scopes, frozen definitions, and the AddOn → self-registration migration checked with [COLLISION]/[ORPHAN]
- TMDO_Deprecated_Addons_Notice warns administrators while any of the 8 in-house partner AddOns is active (site or network)
- Trait register_entity_fields() forwards an optional $args (post_types) to the registry
- Verify: DeprecatedAddonsNoticeTest (4); core unit 645 / integration 449, HP 148, WC 10 OK
- PR: PR-021
- TMDO_Post_Fields::post_type_for_group() / groups_for_post_type() derive HivePress-style groups from their registry post_type scope (core keeps its own attachment / wc_product / nav_menu_item map); the three copy-pasted switches and the query router's POST_TYPE_GROUP_MAP use them
- Term/comment backfill, shadow verifier and CLI take their groups from the registry (TMDO_Term_Comment_Backfill::registered_groups()); installer drops the hp_* entries already covered by its table patterns; REST listings default post type is filterable (tmdo_rest_default_post_type, default 'post')
- Verify: NoHpMappingsTest guard + mapping/default tests; core unit 640 / integration 449, HP 146/49, WC 10/22 OK
- PR: PR-020e
- Feature flags keep only the live AddOn modules (wc_orders, latepoint) as ADDON_MODULES / addon_modules(); drop map_hpct_status() and the 7 HivePress HPCT modules whose interceptors were removed in PR-014
- Remove HPCT detection/notice (Compatibility, core), the admin HPCT import tab and status row, CLI import-hpct, the 7 advisor rules for dead modules, and the removed interceptors from COEXIST_WHITELIST
- Verify: NoHpctLegacyTest guard; core unit 635 / integration 449, HP 145/49, WC 10/22 OK; all sites have every module idle and never imported HPCT
- PR: PR-020c
- Drop hp_listing_core / hp_request_core / hp_vendor_core from TMDO_Post_Fields and hp_user from TMDO_Member_Fields; the HivePress AddOn registers identical definitions (PR-019/020a/033b)
- Tests: core asserts it registers none of them; HP-specific core tests removed (covered by the AddOn); SyncBridgeEntityGuardTest registers its own hp_price group
- Verify: core unit 633 / integration 449, HP 145/49, WC 10/22 OK
- Deploy: activate the HivePress AddOn on dev20/21/22 first
- PR: PR-020b
- TMDO_Post_Group_Reconciler copies non-NULL values that legacy routing stored in another group's table into the post type's own group (NULL/missing legacy values keep the scoped value), resumable via a cursor, with a catch-up pass for rows updated during the run, then switches on tmdo_post_scoped_routing
- CLI: wp tmdo post-reconcile-groups [--dry-run] [--batch=<n>]; registry exposes get_group_post_types()
- Verify: PostGroupReconcilerPlanTest (4), PostGroupReconcilerTest (6); core unit 635 / integration 449, HP 145/49, WC 10/22 OK; read-only dry-run: dev20 320 values / 0 overwrites, dev21/22 none
- PR: PR-034
- Entity groups can declare post_types; get_field() resolves a post's key against its own type's groups first, then unscoped ones (legacy last-wins lookup kept for callers without a post type)
- Hook Bus and the post query router pass the post type, so a listing's hp_rating / hp_status / hp_verified map to hp_listing_core instead of the vendor/request tables; core HP groups declare their post types (schema hashes unchanged)
- Behind tmdo_post_scoped_routing: on for sites without post flat rows, off where rows may sit in the wrong table until PR-034 reconciles them (dry-run: dev20–22 off, dev23/24/30/31/40 on)
- Verify: EntityRegistryScopeTest (8), PostScopedRoutingTest (4); core unit 630 / integration 443, HP 144/49, WC 10/22 OK
- PR: PR-033
- On init:1, re-register aeav_only groups whose flat table exists and whose definition is in wpdo_registry_meta but no active plugin registers them, so their data stays readable (routing only, never DDL)
- Keys owned by a live group, or with rows in native meta (newer data written while orphaned), are left alone; the native-meta check is cached for a day
- Surface restored groups via Site Health (wpdo_orphaned_entity_groups), the admin Doctor tab, and wp wpdo doctor
- Verify: EntityRegistryRestoreTest (9), OrphanGroupRestoreTest (7); core unit 622 / integration 439, HP 141/49, WC 10/22 OK; dry-run on dev20–22 restores 16/17/4 groups, keeps hp_taxonomy hp_sort_order/hp_icon on native
- PR: PR-018
- register_group() keeps the first definition but records a field diff (added/removed/changed) when the same entity/group is re-registered differently; identical re-registration stays silent
- Surface collisions via Site Health (wpdo_entity_group_collisions), the admin Doctor tab, and wp wpdo doctor; warning log throttled to once a day
- Verify: EntityRegistryCollisionTest (7); core unit 613 / integration 432, HP 141/49, WC 10/22 OK
- PR: PR-017
- GET /wpdo/v1/listings rejects unregistered or non-viewable post types (400) and joins posts so Zone A only lists published posts of that type
- WP_Query fallback now returns only show_in_rest hot columns, matching the Zone A path
- Verify: core unit 605 / integration 432 OK (new RestApi tests); 1M bench: page 7ms, total COUNT 0.12s → 1.7s (tracked in PR-030)
- PR: PR-015
- uninstall.php checks every site first and wp_die()s (keeping tables,
options and plugin files) on aeav_only modes, misc rows, or zone
modules past cutover; a failed bootstrap also refuses
- Multisite now checks and cleans every site: network activation is always
false at uninstall, so only the main site was being cleaned before
- Drop the inline fallback DROP loop, unreachable now that a missing
helper refuses instead
Verify: UninstallGuardTest (8 tests incl. running uninstall.php single and
multisite); unit 602 OK, integration 429 OK
PR: PR-008
- install() no longer switches the Hook Bus on or puts user/term/comment
into dual_write; the now-unused maybe_autoactivate_entity_bridge() is gone
- Hook Bus and Mode_Manager defaults are off/disabled; sites with stored
modes (incl. aeav_only) keep them since only unset keys use defaults
- Guard run_dbdelta()'s upgrade.php require like install_v2_tables() so
install() can run under the integration test stub
Verify: InstallerV2Test::test_install_does_not_enable_entity_bridge,
ModeManagerDefaultsTest; unit 602 OK, integration 421 OK
PR: PR-007
- Both term/comment filters now default to disabled for new installs, so
activating the plugin no longer reroutes or drops other plugins' meta
- One-time maybe_upgrade() migration keeps the misc bucket on for sites
that already hold misc rows and never set the toggle (data stays readable)
- Admin toggles show the same defaults as the filters
Verify: phpunit-integration --filter 'defaults_false|preserve_default';
unit 600 OK, integration 420 OK
PR: PR-006
- Seed a HivePress-shaped dataset three ways (native meta, entity tables,
hot table) with flat tables built by TMDO's own schema code
- Refuse any database not named *_bench before the bootstrap connects;
bootstrap now honours a pre-defined ABSPATH for the wp-admin stub
- Full scale seeds 9.5M postmeta rows in 7m25s; 0 value mismatches
Verify: TMDO_TEST_DB_NAME=wp_wpdo_bench php tests/bench/seed.php
--vendors=100000 --listings=1000000; unit 600 OK, integration 416 OK
PR: PR-004
wpdev