_join_url_parts() only checked the first segment for a scheme, so
passing an already-absolute URL (e.g. from cart_url()) through the
|host filter would join route_prefix() + absolute URL, producing
"https://host/https://host/path/".
Now detects schemes in later segments and resets the base.
Also add missing 'market' entry to _nav.html _app_slugs to match
_nav_oob.html — without it the market menu item fell through to
coop_url('/market/') instead of market_url('/').
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The checkout form action used url_for('cart_global.checkout') which
only resolves inside the cart app. Replace with cart_url() and
page_cart_url() Jinja globals that work across all apps.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Merge three alembic heads and drop:
- calendar_entries.order_id FK → orders.id
- tickets.order_id FK → orders.id
Columns kept as plain integers for glue-service bridging.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The back_populates="menu_items" referenced a relationship removed from
Post in the glue layer commit. MenuItem model kept for table preservation.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Templates: item.post.X → item.X (MenuNode has label/slug/feature_image directly)
- factory.py: add glue.models to import loop + register_glue_handlers() at startup
- alembic env.py: add glue.models to import loop
- New migration: container_relations + menu_nodes tables with backfill from existing data
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Cross-domain relationships like Product.order_items → OrderItem use
string references that SQLAlchemy resolves by class name lookup. All
model packages must be imported so every class is registered before
mapper configuration runs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Two branches diverged from e5c3f9a2b1d6 (tag_groups):
- domain_events → generic_containers (g7e5b1c3d4f8)
- tickets → page_configs → market_places → page_tracking (c3d4e5f6a7b8)
This merge migration joins them into a single head so
alembic upgrade head works without specifying a branch.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Was present on main but never carried over. Without it, alembic
upgrade head fails with 'No script_location key found'.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
shared/logging/ shadows Python's stdlib logging module, causing a
circular import when any code does `import logging`. This breaks
both the entrypoint Redis flush and Hypercorn app loading.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
requirements.txt copied from shared_lib for Docker build compatibility.
Alembic env.py now uses try/except for app-specific model imports so
upgrade head works in Docker where only one app's code is present.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
