Skip to content

storage: test the persist sink's pipeline end to end - #38936

Draft
martykulma wants to merge 1 commit into
maz-sink-group-snapshotfrom
maz-sink-pipeline-tests
Draft

martykulma wants to merge 1 commit into
maz-sink-group-snapshotfrom
maz-sink-pipeline-tests

Conversation

@martykulma

@martykulma martykulma commented Sep 19, 2026

Copy link
Copy Markdown
Contributor

write_batches had a harness of its own, driven by a script standing in for
what the minter would send. That can't say whether the two agree, and the
grouping a pinned snapshot gets is a property of the pair.

run_sink_pipeline wires the real minter into the real writer, so the ordering
between a committed ceiling and the data it covers is the operators' own.
run_mint_batch_descriptions covers the minter alone, which had no test at all.

Both pinned scripts write one batch, the interleaved one included, which is
where routing by description used to close and reopen a builder on every flip
between the snapshot's time and the replication rows above it.

🤖 Generated with Claude Code

@martykulma
martykulma added this pull request to stack #38937 September 19, 2026 00:30
@martykulma martykulma changed the title maz sink pipeline tests storage: test the persist sink's pipeline end to end Sep 19, 2026
The writer had a harness of its own, driven by a script that stands in for what
the minter would send. That cannot say whether the two agree, and the grouping a
pinned snapshot gets is a property of the pair rather than of either operator.

run_sink_pipeline wires the real minter into the real writer, so the ordering
between a committed ceiling and the data it covers is the operators' own.
run_mint_batch_descriptions covers the minter alone, which had no test at all.

Both pinned scripts write one batch, the interleaved one included, which is
where routing by description used to close and reopen a builder on every flip
between the snapshot's time and the replication rows above it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@martykulma
martykulma force-pushed the maz-sink-pipeline-tests branch from 48729da to 8573f77 Compare September 19, 2026 01:00
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant