You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Dynamic mode: confirm inferred types before table creation, with a provenance flag #195
api/services/dynamic_loader.py creates a table named csv_<sha256[:16]> in uploads_schema(). This is already separate from the twelve T&E tables.
Column types are inferred from the data. The inference is not confirmed by a person before the table is created. I have not reproduced the mis-typed clearance and date examples from the feedback.
Proposed
Return the inferred types from the upload preview, and create the table only after the user confirms them or changes them.
Store a provenance value on each dynamic table (for example inferred_unconfirmed, user_confirmed) and show it in the API response and the UI.
Keep dynamic-mode tables out of any T&E import path. Assert this in a test.
Done when
A test shows that no table exists before confirmation, and that confirmation creates it with the confirmed types.
A test shows that the provenance value is stored and returned.
The T&E import path rejects a dynamic-mode table as a source (test run attached).
Activity
feat: record whether a dynamic table's column types were confirmed (#195
feat: record whether a dynamic table's column types were confirmed (#195
feat: record whether a dynamic table's column types were confirmed (#195
Merge main into #195/#172/#184 branch
Merge main into #195/#172/#184 branch (picks up #215)
Merge main into #195/#172/#184 branch (picks up #221 and #222) and pl…
Merge main into #195/#172/#184 branch (picks up #225)
Merge pull request #213 from amar-python/claude/csv-loading-postgres-…
Merge main into the M5 branch (#109)
test: steadier browser tests; matrix updated for M5-M7 and #229
Merge origin/main; keep only the test additions still needed (#191, #192