Implementing transaction_timeout

session summary

Nik demonstrated a browser-based workflow for reviewing the proposed transaction_timeout patch: mirror the mailing-list attachment, open a preconfigured Gitpod workspace, compile Postgres, and test the feature. The timeout correctly canceled a long transaction. Andrey then moved timer startup from transaction begin to assignment of a real XID, exploring a narrower write-transaction/Xmin-horizon timeout. They concluded this changed the patch's intended semantics and needed separate user discussion; isolation tests also exposed timing complexity and Windows instability.

timecodes

1. 03:30 Why Postgres needs transaction_timeout
2. 05:30 Turning a mailing-list patch into a browser review
3. 09:00 Building patched Postgres in Gitpod
4. 17:30 Testing transaction_timeout from psql
5. 19:00 Virtual, real, and prepared transactions
6. 20:30 Starting the timer only after XID assignment
7. 28:30 Transaction duration versus Xmin-horizon protection
8. 32:30 Isolation-test failures and unresolved semantics