michalheld opened a new issue, #15163:
URL: https://github.com/apache/iceberg/issues/15163
### Feature Request / Improvement
Introduce a **session-level configuration** that automatically applies
time-travel to all table references in a session, enabling **time-freeze** and
**deterministic query executions** without requiring manual query
modifications.
Currently, time-travel is only available by explicitly modifying queries
(e.g., specifying `AS OF` / snapshot IDs / timestamps). A session-level
configuration would allow consistent, repeatable query executions within a
session **without any changes to existing code**.
The session-level configuration will only support **time-based**
time-travel; there will be **no support for snapshot-id–based** session
configuration.
---
### Semantics & Behavior
#### 1. Precedence of Time-Travel Configurations
If both:
- a **session-level** time-travel configuration, and
- an **explicit table-level** time-travel specification (e.g., `FOR
SYSTEM_TIME AS OF`, snapshot ID, timestamp)
are provided, then the **explicit table-level configuration takes
precedence** over the session-level setting.
This ensures that existing explicit time-travel usage remains fully honored
and backward compatible.
---
#### 2. Snapshot Expiration Behavior
The proposed enhancement **does not change** the existing behavior for
snapshot expiration.
- If a query (via session-level time-travel) requests a timestamp or
snapshot for which the underlying snapshot has been expired/removed, the query
will **fail with the existing exception**:
- `java.lang.IllegalArgumentException`
---
#### 3. Impact on Writes
Session-level time-travel applies **only to read operations**.
- **Reads**: Resolved against the configured historical snapshot/timestamp
for the session (unless overridden at the table level).
- **Writes**: Always operate on the **current/latest table snapshot** and
are **not affected** by the session-level time-travel configuration.
This avoids surprising behavior for mutations and keeps write semantics
aligned with current Iceberg expectations.
### Query engine
Spark
### Willingness to contribute
- [x] I can contribute this improvement/feature independently
- [x] I would be willing to contribute this improvement/feature with
guidance from the Iceberg community
- [ ] I cannot contribute this improvement/feature at this time
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
