Eight PRs this week just to match how psql wraps text, reports errors, and formats partitions.
Passing psql's regression tests means matching behavior nobody documented — including ASCII-wrapped column edge cases (#809) and the exact error location prefix format (#803). @NikolayS landed a combined alignment and describe fix in #788, but only after reverting a previous attempt in #786. Even the test harness needed work: bash 3.2 compatibility (#801) and WAL timing variance normalization (#802) because CI doesn't care about your feelings. The partition relation titles alone needed a PG14-17 normalization pass (#800).
Three upstream PRs pending: a protocol crash, a missing OID accessor, and a stray import.
The tokio-postgres copy_in implementation could fire a double-Sync, which Postgres interprets as two transaction boundaries — not ideal when you meant one. @NikolayS's #1337 fixes the protocol state machine. Also pending: exposing type OIDs on SimpleColumn (#1338) so callers can actually inspect what they're getting, and guarding a PathBuf import that only makes sense on Unix (#1336).
A pending fix for a protocol state mismatch that left connections hung.
When a server enters COPY IN mode, it stops sending ReadyForQuery — but pgdog's connection state didn't get the memo, leaving it waiting for a message that would never arrive. @NikolayS's #885 drops the stale expectation so the pooler can actually forward the COPY stream.
Activity on the local rust-postgres fork without merged PRs to show for it.
Three commits landed in @NikolayS's rust-postgres fork this week. The upstream PRs (#1337, #1338, #1336) are still pending review — the fork is where the work happens before it goes anywhere.
The newspaper generator saw some activity, but no PRs to dissect.
One commit to the repo that turns GitHub activity into weekly newspapers. What changed? The commit data doesn't say, and neither will we.
The generator's output repo saw minor activity.
A single commit landed in gitzette-dispatch, the repo that actually produces these weekly editions. Presumably it helped. No PR to cite.
The stateless query router saw a lone commit this week.
One commit to spqr with no associated PR data. The query router remains stateless; so does this write-up.
No commits, no PRs, no releases.
DBLab took the week off. Thin clones remain thin. Branches remain branchable.
The docs stayed exactly as they were.
Zero commits to the PostgresAI documentation. Either everything was already perfectly explained, or nobody tried to explain anything new.
| pgdogdev/pgdog | ★★★★★★★★★★ | 4,279 |
| rust-postgres/rust-postgres | ★★★★★★★★★☆ | 3,925 |
| postgres-ai/database-lab-engine | ★★★★★★☆☆☆☆ | 2,444 |
| rpg | ★☆☆☆☆☆☆☆☆☆ | 222 |
| gitzette | ★☆☆☆☆☆☆☆☆☆ | 4 |
| gitzette-dispatch | ★☆☆☆☆☆☆☆☆☆ | 1 |
The generator's output repo saw minor activity.
A single commit landed in gitzette-dispatch, the repo that actually produces these weekly editions. Presumably it helped. No PR to cite.
The stateless query router saw a lone commit this week.
One commit to spqr with no associated PR data. The query router remains stateless; so does this write-up.
No commits, no PRs, no releases.
DBLab took the week off. Thin clones remain thin. Branches remain branchable.
The docs stayed exactly as they were.
Zero commits to the PostgresAI documentation. Either everything was already perfectly explained, or nobody tried to explain anything new.
| pgdogdev/pgdog | ★★★★★★★★★★ | 4,279 |
| rust-postgres/rust-postgres | ★★★★★★★★★☆ | 3,925 |
| postgres-ai/database-lab-engine | ★★★★★★☆☆☆☆ | 2,444 |
| rpg | ★☆☆☆☆☆☆☆☆☆ | 222 |
| gitzette | ★☆☆☆☆☆☆☆☆☆ | 4 |
| gitzette-dispatch | ★☆☆☆☆☆☆☆☆☆ | 1 |
Your GitHub week, turned into something worth reading.
Generate your dispatch →