Open the psycopg transaction without unsupported SQLAlchemy-style options, then set the PostgreSQL isolation and read-only characteristics before the first query. Cover the exact snapshot endpoint path with a regression test.
Result validation bypassed the project-specific GIS transform and called ST_Transform on custom engineering SRIDs. Read project and map coordinates from gis.junctions so every project uses its configured publication transform.
Project selection previously relied only on membership, so inactive projects remained visible. Filter at the repository boundary and add regression coverage.
Reorganize algorithm packages by business responsibility, move orchestration into services, and keep database access behind pooled repositories.
Harden analysis API validation, remove unsafe legacy simulation endpoints, and add regression and architecture boundary coverage.
BREAKING CHANGE: legacy algorithm module paths and obsolete simulation endpoints are removed.
- make realtime replacement and analysis result writes transactional\n- consolidate SCADA repositories and remove process-global project state\n- validate SCADA batches and use indexed GIS-backed business queries\n\nBREAKING CHANGE: remove the public analysis result writer and the pipeline-health network_name query parameter.
Reorganize WNDB by responsibility and remove legacy scheme endpoints.\n\nRoute analysis and time-series access through project pools, preserve transactional realtime replacement, and refresh GIS materialized views after writes.\n\nAdd database architecture documentation, live pooling coverage, API contract updates, and executable container verification.\n\nBREAKING CHANGE: legacy scheme APIs and flat app.native.wndb module imports are removed.
Bind every scheme request to ProjectContext, keep viewer access read-only, reject concurrent optimization jobs without blocking worker threads, and cap export/update payload sizes. Run optimization and workbook work off the event loop.
jiang