Architecture changes:
- API Routes -> manager.py (thin adapter) -> services/ (business logic) -> queries/ (atomic DB ops)
- Services own session lifecycle via create_session()
- Queries accept Session as parameter, do single-table atomic operations
New app/assets/services/ layer:
- __init__.py - exports all service functions
- ingest.py - ingest_file_from_path(), register_existing_asset()
- asset_management.py - get_asset_detail(), update_asset_metadata(), delete_asset_reference(), set_asset_preview()
- tagging.py - apply_tags(), remove_tags(), list_tags()
Removed from queries/asset_info.py:
- ingest_fs_asset (moved to services/ingest.py as ingest_file_from_path)
- update_asset_info_full (moved to services/asset_management.py as update_asset_metadata)
- create_asset_info_for_existing_asset (moved to services/ingest.py as register_existing_asset)
Updated manager.py:
- Now a thin adapter that transforms API schemas to/from service calls
- Delegates all business logic to services layer
- No longer imports sqlalchemy.orm.Session or models directly
Test updates:
- Fixed test_cache_state.py import of pick_best_live_path (moved to helpers.py)
- Added comprehensive service layer tests (41 new tests)
- All 112 query + service tests pass
Amp-Thread-ID: https://ampcode.com/threads/T-019c24e2-7ae4-707f-ad19-c775ed8b82b5
Co-authored-by: Amp <amp@ampcode.com>
* model_management: lazy-cache aimdo_tensor
These tensors cosntructed from aimdo-allocations are CPU expensive to
make on the pytorch side. Add a cache version that will be valid with
signature match to fast path past whatever torch is doing.
* dynamic_vram: Minimize fast path CPU work
Move as much as possible inside the not resident if block and cache
the formed weight and bias rather than the flat intermediates. In
extreme layer weight rates this adds up.
* Fix bypass dtype/device moving
* Force offloading mode for training
* training context var
* offloading implementation in training node
* fix wrong input type
* Support bypass load lora model, correct adapter/offloading handling
This was missing the stochastic rounding required for fp8 downcast
to be consistent with model_patcher.patch_weight_to_device.
Missed in testing as I spend too much time with quantized tensors
and overlooked the simpler ones.
If there are non-trivial python objects nested in the model_options, this
causes all sorts of issues. Traverse lists and dicts so clones can safely
overide settings and BYO objects but stop there on the deepclone.
* feat(api-nodes-Kling): add new models (V3, O3)
* remove storyboard from VideoToVideo node
* added check for total duration of storyboards
* fixed other small things
* updated display name for nodes
* added "fake" seed
* revert threaded model loader change
This change was only needed to get around the pytorch 2.7 mempool bugs,
and should have been reverted along with #12260. This fixes a different
memory leak where pytorch gets confused about cache emptying.
* load non comfy weights
* MPDynamic: Pre-generate the tensors for vbars
Apparently this is an expensive operation that slows down things.
* bump to aimdo 1.8
New features:
watermark limit feature
logging enhancements
-O2 build on linux
