chore: update insta snapshot format by snth · Pull Request #5750 · PRQL/prql

snth · 2026-03-27T15:18:21Z

Force-update all integration test snapshots to current insta format. The newer insta (1.46.3) enforces a stricter snapshot format, causing CI to fail with 'existing value is in a legacy format'.

@r

Update inline snapshot string delimiters (@r" -> @", r### -> r#) and remove deprecated snapshot_kind: text metadata from snap files. All changes are format-only with no semantic differences. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Non-format changes from insta snapshot regeneration: - compile/fmt append_select_nulls.snap: expression field corrected to match current .prql file (stale '# duckdb:skip' / '# postgres:skip' comments were removed from the file in a prior commit but the snap metadata was never regenerated) - compileall snap files: leading blank lines added to snapshot content. Each blank line represents a dialect whose output matches the generic dialect (empty diff). The previous insta version trimmed these; the new version preserves them. The actual diff content is unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T15:48:47Z

This final commit c971aff is identical to 5f1de96 , I just split it into two commits to separate semantic changes from regenerating the snapshots to pure formatting changes. All tests passed for 5f1de96 but now some are failing for c971aff. What gives?

max-sixty · 2026-03-27T16:18:10Z

great! will look at the snapshot issue

@prql-bot are you online?!

prql-bot · 2026-03-27T16:20:00Z

I'm online! 👋

Regarding @snth's question about the CI failure on c971aff — the failure is not related to the snapshot changes. It's a Docker Hub authentication error in the test-dbs-external job:

Head "https://registry-1.docker.io/v2/library/postgres/manifests/alpine": unauthorized: incorrect username or password

This is an infrastructure/credentials issue pulling the postgres container image. All the actual Rust tests, lints, builds, and code coverage checks passed successfully. A re-run of that job should fix it (assuming Docker Hub credentials are refreshed or the rate limit clears).

Authenticate with Docker Hub before pulling images to avoid rate limits. - test-rust.yaml: login before docker compose up (postgres, mysql, clickhouse) - build-devcontainer.yaml: login before image builds as defensive measure Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The Docker Hub login step in test-rust and build-devcontainer workflows needs access to DOCKERHUB_USERNAME and DOCKERHUB_TOKEN secrets. These are only available via 'secrets: inherit' on workflow_call invocations. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T17:44:13Z

@max-sixty I set up the Docker Hub account and added the details to the repo secrets. Should be good to merge now.

max-sixty · 2026-03-27T19:44:09Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

max-sixty · 2026-03-27T19:44:14Z

thanks @snth !

max-sixty · 2026-03-28T20:54:24Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

@prql-bot ?

Skip empty diffs (where dialect output matches generic) instead of including them, which produced leading/intermediate blank lines in the snapshot files. Closes follow-up from #5750. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

prql-bot · 2026-03-28T21:12:09Z

Done — opened #5752 to fix this at the source. The compileall test was collecting empty diffs (where a dialect's output matches generic) and joining them with newlines, producing the leading blank lines. The fix skips empty diffs entirely.

@snth

## Problem The `tend-mention` workflow's `verify` job has a concurrency group keyed by PR number with `cancel-in-progress: true`. When two comments arrive on the same PR within seconds, the second run cancels the first — even if the first was an explicit `@bot` mention. This is exactly what happened at [PRQL/prql#5750](PRQL/prql#5750 (comment)): 1. **19:44:09** — `@prql-bot can you do a follow-up...` → triggered run [23664429862](https://github.com/PRQL/prql/actions/runs/23664429862) 2. **19:44:14** — `thanks @snth !` (5 seconds later) → triggered run [23664431397](https://github.com/PRQL/prql/actions/runs/23664431397) 3. Run 2 **cancelled** run 1 via the concurrency group 4. Run 2 processed "thanks @snth" (no bot mention) → bot exited silently. The explicit `@prql-bot` request was lost. ## Solution - **Remove concurrency from the `verify` job** — it's lightweight and fast; parallel verify jobs for the same PR are harmless. - **Change `handle`'s `cancel-in-progress` to `false`** — jobs queue instead of cancelling. The first handle completes (processes the mention), then the second runs (exits silently for non-mention). No mention gets dropped. ## Testing - Added `test_mention_verify_no_concurrency` — verifies the verify job has no concurrency group - Added `test_mention_handle_job_queues_not_cancels` — verifies handle uses `cancel-in-progress: false` - Both tests fail before the fix, pass after - Full suite: 110 tests pass, all pre-commit lints pass --- Closes#93 — automated triage --------- Co-authored-by: tend-agent <270458913+tend-agent@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>

snth · 2026-03-27T15:18:21Z

Force-update all integration test snapshots to current insta format. The newer insta (1.46.3) enforces a stricter snapshot format, causing CI to fail with 'existing value is in a legacy format'.

@r

Update inline snapshot string delimiters (@r" -> @", r### -> r#) and remove deprecated snapshot_kind: text metadata from snap files. All changes are format-only with no semantic differences. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Non-format changes from insta snapshot regeneration: - compile/fmt append_select_nulls.snap: expression field corrected to match current .prql file (stale '# duckdb:skip' / '# postgres:skip' comments were removed from the file in a prior commit but the snap metadata was never regenerated) - compileall snap files: leading blank lines added to snapshot content. Each blank line represents a dialect whose output matches the generic dialect (empty diff). The previous insta version trimmed these; the new version preserves them. The actual diff content is unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T15:48:47Z

This final commit c971aff is identical to 5f1de96 , I just split it into two commits to separate semantic changes from regenerating the snapshots to pure formatting changes. All tests passed for 5f1de96 but now some are failing for c971aff. What gives?

max-sixty · 2026-03-27T16:18:10Z

great! will look at the snapshot issue

@prql-bot are you online?!

prql-bot · 2026-03-27T16:20:00Z

I'm online! 👋

Regarding @snth's question about the CI failure on c971aff — the failure is not related to the snapshot changes. It's a Docker Hub authentication error in the test-dbs-external job:

Head "https://registry-1.docker.io/v2/library/postgres/manifests/alpine": unauthorized: incorrect username or password

This is an infrastructure/credentials issue pulling the postgres container image. All the actual Rust tests, lints, builds, and code coverage checks passed successfully. A re-run of that job should fix it (assuming Docker Hub credentials are refreshed or the rate limit clears).

Authenticate with Docker Hub before pulling images to avoid rate limits. - test-rust.yaml: login before docker compose up (postgres, mysql, clickhouse) - build-devcontainer.yaml: login before image builds as defensive measure Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The Docker Hub login step in test-rust and build-devcontainer workflows needs access to DOCKERHUB_USERNAME and DOCKERHUB_TOKEN secrets. These are only available via 'secrets: inherit' on workflow_call invocations. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T17:44:13Z

@max-sixty I set up the Docker Hub account and added the details to the repo secrets. Should be good to merge now.

max-sixty · 2026-03-27T19:44:09Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

max-sixty · 2026-03-27T19:44:14Z

thanks @snth !

max-sixty · 2026-03-28T20:54:24Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

@prql-bot ?

Skip empty diffs (where dialect output matches generic) instead of including them, which produced leading/intermediate blank lines in the snapshot files. Closes follow-up from #5750. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

prql-bot · 2026-03-28T21:12:09Z

Done — opened #5752 to fix this at the source. The compileall test was collecting empty diffs (where a dialect's output matches generic) and joining them with newlines, producing the leading blank lines. The fix skips empty diffs entirely.

@snth

## Problem The `tend-mention` workflow's `verify` job has a concurrency group keyed by PR number with `cancel-in-progress: true`. When two comments arrive on the same PR within seconds, the second run cancels the first — even if the first was an explicit `@bot` mention. This is exactly what happened at [PRQL/prql#5750](PRQL/prql#5750 (comment)): 1. **19:44:09** — `@prql-bot can you do a follow-up...` → triggered run [23664429862](https://github.com/PRQL/prql/actions/runs/23664429862) 2. **19:44:14** — `thanks @snth !` (5 seconds later) → triggered run [23664431397](https://github.com/PRQL/prql/actions/runs/23664431397) 3. Run 2 **cancelled** run 1 via the concurrency group 4. Run 2 processed "thanks @snth" (no bot mention) → bot exited silently. The explicit `@prql-bot` request was lost. ## Solution - **Remove concurrency from the `verify` job** — it's lightweight and fast; parallel verify jobs for the same PR are harmless. - **Change `handle`'s `cancel-in-progress` to `false`** — jobs queue instead of cancelling. The first handle completes (processes the mention), then the second runs (exits silently for non-mention). No mention gets dropped. ## Testing - Added `test_mention_verify_no_concurrency` — verifies the verify job has no concurrency group - Added `test_mention_handle_job_queues_not_cancels` — verifies handle uses `cancel-in-progress: false` - Both tests fail before the fix, pass after - Full suite: 110 tests pass, all pre-commit lints pass --- Closes#93 — automated triage --------- Co-authored-by: tend-agent <270458913+tend-agent@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>

snth · 2026-03-27T15:18:21Z

Force-update all integration test snapshots to current insta format. The newer insta (1.46.3) enforces a stricter snapshot format, causing CI to fail with 'existing value is in a legacy format'.

@r

Update inline snapshot string delimiters (@r" -> @", r### -> r#) and remove deprecated snapshot_kind: text metadata from snap files. All changes are format-only with no semantic differences. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Non-format changes from insta snapshot regeneration: - compile/fmt append_select_nulls.snap: expression field corrected to match current .prql file (stale '# duckdb:skip' / '# postgres:skip' comments were removed from the file in a prior commit but the snap metadata was never regenerated) - compileall snap files: leading blank lines added to snapshot content. Each blank line represents a dialect whose output matches the generic dialect (empty diff). The previous insta version trimmed these; the new version preserves them. The actual diff content is unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T15:48:47Z

This final commit c971aff is identical to 5f1de96 , I just split it into two commits to separate semantic changes from regenerating the snapshots to pure formatting changes. All tests passed for 5f1de96 but now some are failing for c971aff. What gives?

max-sixty · 2026-03-27T16:18:10Z

great! will look at the snapshot issue

@prql-bot are you online?!

prql-bot · 2026-03-27T16:20:00Z

I'm online! 👋

Regarding @snth's question about the CI failure on c971aff — the failure is not related to the snapshot changes. It's a Docker Hub authentication error in the test-dbs-external job:

Head "https://registry-1.docker.io/v2/library/postgres/manifests/alpine": unauthorized: incorrect username or password

This is an infrastructure/credentials issue pulling the postgres container image. All the actual Rust tests, lints, builds, and code coverage checks passed successfully. A re-run of that job should fix it (assuming Docker Hub credentials are refreshed or the rate limit clears).

Authenticate with Docker Hub before pulling images to avoid rate limits. - test-rust.yaml: login before docker compose up (postgres, mysql, clickhouse) - build-devcontainer.yaml: login before image builds as defensive measure Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The Docker Hub login step in test-rust and build-devcontainer workflows needs access to DOCKERHUB_USERNAME and DOCKERHUB_TOKEN secrets. These are only available via 'secrets: inherit' on workflow_call invocations. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T17:44:13Z

@max-sixty I set up the Docker Hub account and added the details to the repo secrets. Should be good to merge now.

max-sixty · 2026-03-27T19:44:09Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

max-sixty · 2026-03-27T19:44:14Z

thanks @snth !

max-sixty · 2026-03-28T20:54:24Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

@prql-bot ?

Skip empty diffs (where dialect output matches generic) instead of including them, which produced leading/intermediate blank lines in the snapshot files. Closes follow-up from #5750. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

prql-bot · 2026-03-28T21:12:09Z

Done — opened #5752 to fix this at the source. The compileall test was collecting empty diffs (where a dialect's output matches generic) and joining them with newlines, producing the leading blank lines. The fix skips empty diffs entirely.

@snth

## Problem The `tend-mention` workflow's `verify` job has a concurrency group keyed by PR number with `cancel-in-progress: true`. When two comments arrive on the same PR within seconds, the second run cancels the first — even if the first was an explicit `@bot` mention. This is exactly what happened at [PRQL/prql#5750](PRQL/prql#5750 (comment)): 1. **19:44:09** — `@prql-bot can you do a follow-up...` → triggered run [23664429862](https://github.com/PRQL/prql/actions/runs/23664429862) 2. **19:44:14** — `thanks @snth !` (5 seconds later) → triggered run [23664431397](https://github.com/PRQL/prql/actions/runs/23664431397) 3. Run 2 **cancelled** run 1 via the concurrency group 4. Run 2 processed "thanks @snth" (no bot mention) → bot exited silently. The explicit `@prql-bot` request was lost. ## Solution - **Remove concurrency from the `verify` job** — it's lightweight and fast; parallel verify jobs for the same PR are harmless. - **Change `handle`'s `cancel-in-progress` to `false`** — jobs queue instead of cancelling. The first handle completes (processes the mention), then the second runs (exits silently for non-mention). No mention gets dropped. ## Testing - Added `test_mention_verify_no_concurrency` — verifies the verify job has no concurrency group - Added `test_mention_handle_job_queues_not_cancels` — verifies handle uses `cancel-in-progress: false` - Both tests fail before the fix, pass after - Full suite: 110 tests pass, all pre-commit lints pass --- Closes#93 — automated triage --------- Co-authored-by: tend-agent <270458913+tend-agent@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>

snth · 2026-03-27T15:18:21Z

Force-update all integration test snapshots to current insta format. The newer insta (1.46.3) enforces a stricter snapshot format, causing CI to fail with 'existing value is in a legacy format'.

@r

Update inline snapshot string delimiters (@r" -> @", r### -> r#) and remove deprecated snapshot_kind: text metadata from snap files. All changes are format-only with no semantic differences. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Non-format changes from insta snapshot regeneration: - compile/fmt append_select_nulls.snap: expression field corrected to match current .prql file (stale '# duckdb:skip' / '# postgres:skip' comments were removed from the file in a prior commit but the snap metadata was never regenerated) - compileall snap files: leading blank lines added to snapshot content. Each blank line represents a dialect whose output matches the generic dialect (empty diff). The previous insta version trimmed these; the new version preserves them. The actual diff content is unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T15:48:47Z

This final commit c971aff is identical to 5f1de96 , I just split it into two commits to separate semantic changes from regenerating the snapshots to pure formatting changes. All tests passed for 5f1de96 but now some are failing for c971aff. What gives?

max-sixty · 2026-03-27T16:18:10Z

great! will look at the snapshot issue

@prql-bot are you online?!

prql-bot · 2026-03-27T16:20:00Z

I'm online! 👋

Regarding @snth's question about the CI failure on c971aff — the failure is not related to the snapshot changes. It's a Docker Hub authentication error in the test-dbs-external job:

Head "https://registry-1.docker.io/v2/library/postgres/manifests/alpine": unauthorized: incorrect username or password

This is an infrastructure/credentials issue pulling the postgres container image. All the actual Rust tests, lints, builds, and code coverage checks passed successfully. A re-run of that job should fix it (assuming Docker Hub credentials are refreshed or the rate limit clears).

Authenticate with Docker Hub before pulling images to avoid rate limits. - test-rust.yaml: login before docker compose up (postgres, mysql, clickhouse) - build-devcontainer.yaml: login before image builds as defensive measure Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The Docker Hub login step in test-rust and build-devcontainer workflows needs access to DOCKERHUB_USERNAME and DOCKERHUB_TOKEN secrets. These are only available via 'secrets: inherit' on workflow_call invocations. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T17:44:13Z

@max-sixty I set up the Docker Hub account and added the details to the repo secrets. Should be good to merge now.

max-sixty · 2026-03-27T19:44:09Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

max-sixty · 2026-03-27T19:44:14Z

thanks @snth !

max-sixty · 2026-03-28T20:54:24Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

@prql-bot ?

Skip empty diffs (where dialect output matches generic) instead of including them, which produced leading/intermediate blank lines in the snapshot files. Closes follow-up from #5750. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

prql-bot · 2026-03-28T21:12:09Z

Done — opened #5752 to fix this at the source. The compileall test was collecting empty diffs (where a dialect's output matches generic) and joining them with newlines, producing the leading blank lines. The fix skips empty diffs entirely.

@snth

## Problem The `tend-mention` workflow's `verify` job has a concurrency group keyed by PR number with `cancel-in-progress: true`. When two comments arrive on the same PR within seconds, the second run cancels the first — even if the first was an explicit `@bot` mention. This is exactly what happened at [PRQL/prql#5750](PRQL/prql#5750 (comment)): 1. **19:44:09** — `@prql-bot can you do a follow-up...` → triggered run [23664429862](https://github.com/PRQL/prql/actions/runs/23664429862) 2. **19:44:14** — `thanks @snth !` (5 seconds later) → triggered run [23664431397](https://github.com/PRQL/prql/actions/runs/23664431397) 3. Run 2 **cancelled** run 1 via the concurrency group 4. Run 2 processed "thanks @snth" (no bot mention) → bot exited silently. The explicit `@prql-bot` request was lost. ## Solution - **Remove concurrency from the `verify` job** — it's lightweight and fast; parallel verify jobs for the same PR are harmless. - **Change `handle`'s `cancel-in-progress` to `false`** — jobs queue instead of cancelling. The first handle completes (processes the mention), then the second runs (exits silently for non-mention). No mention gets dropped. ## Testing - Added `test_mention_verify_no_concurrency` — verifies the verify job has no concurrency group - Added `test_mention_handle_job_queues_not_cancels` — verifies handle uses `cancel-in-progress: false` - Both tests fail before the fix, pass after - Full suite: 110 tests pass, all pre-commit lints pass --- Closes#93 — automated triage --------- Co-authored-by: tend-agent <270458913+tend-agent@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>

snth · 2026-03-27T15:18:21Z

Force-update all integration test snapshots to current insta format. The newer insta (1.46.3) enforces a stricter snapshot format, causing CI to fail with 'existing value is in a legacy format'.

@r

Update inline snapshot string delimiters (@r" -> @", r### -> r#) and remove deprecated snapshot_kind: text metadata from snap files. All changes are format-only with no semantic differences. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Non-format changes from insta snapshot regeneration: - compile/fmt append_select_nulls.snap: expression field corrected to match current .prql file (stale '# duckdb:skip' / '# postgres:skip' comments were removed from the file in a prior commit but the snap metadata was never regenerated) - compileall snap files: leading blank lines added to snapshot content. Each blank line represents a dialect whose output matches the generic dialect (empty diff). The previous insta version trimmed these; the new version preserves them. The actual diff content is unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T15:48:47Z

This final commit c971aff is identical to 5f1de96 , I just split it into two commits to separate semantic changes from regenerating the snapshots to pure formatting changes. All tests passed for 5f1de96 but now some are failing for c971aff. What gives?

max-sixty · 2026-03-27T16:18:10Z

great! will look at the snapshot issue

@prql-bot are you online?!

prql-bot · 2026-03-27T16:20:00Z

I'm online! 👋

Regarding @snth's question about the CI failure on c971aff — the failure is not related to the snapshot changes. It's a Docker Hub authentication error in the test-dbs-external job:

Head "https://registry-1.docker.io/v2/library/postgres/manifests/alpine": unauthorized: incorrect username or password

This is an infrastructure/credentials issue pulling the postgres container image. All the actual Rust tests, lints, builds, and code coverage checks passed successfully. A re-run of that job should fix it (assuming Docker Hub credentials are refreshed or the rate limit clears).

Authenticate with Docker Hub before pulling images to avoid rate limits. - test-rust.yaml: login before docker compose up (postgres, mysql, clickhouse) - build-devcontainer.yaml: login before image builds as defensive measure Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The Docker Hub login step in test-rust and build-devcontainer workflows needs access to DOCKERHUB_USERNAME and DOCKERHUB_TOKEN secrets. These are only available via 'secrets: inherit' on workflow_call invocations. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T17:44:13Z

@max-sixty I set up the Docker Hub account and added the details to the repo secrets. Should be good to merge now.

max-sixty · 2026-03-27T19:44:09Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

max-sixty · 2026-03-27T19:44:14Z

thanks @snth !

max-sixty · 2026-03-28T20:54:24Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

@prql-bot ?

Skip empty diffs (where dialect output matches generic) instead of including them, which produced leading/intermediate blank lines in the snapshot files. Closes follow-up from #5750. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

prql-bot · 2026-03-28T21:12:09Z

Done — opened #5752 to fix this at the source. The compileall test was collecting empty diffs (where a dialect's output matches generic) and joining them with newlines, producing the leading blank lines. The fix skips empty diffs entirely.

@snth

## Problem The `tend-mention` workflow's `verify` job has a concurrency group keyed by PR number with `cancel-in-progress: true`. When two comments arrive on the same PR within seconds, the second run cancels the first — even if the first was an explicit `@bot` mention. This is exactly what happened at [PRQL/prql#5750](PRQL/prql#5750 (comment)): 1. **19:44:09** — `@prql-bot can you do a follow-up...` → triggered run [23664429862](https://github.com/PRQL/prql/actions/runs/23664429862) 2. **19:44:14** — `thanks @snth !` (5 seconds later) → triggered run [23664431397](https://github.com/PRQL/prql/actions/runs/23664431397) 3. Run 2 **cancelled** run 1 via the concurrency group 4. Run 2 processed "thanks @snth" (no bot mention) → bot exited silently. The explicit `@prql-bot` request was lost. ## Solution - **Remove concurrency from the `verify` job** — it's lightweight and fast; parallel verify jobs for the same PR are harmless. - **Change `handle`'s `cancel-in-progress` to `false`** — jobs queue instead of cancelling. The first handle completes (processes the mention), then the second runs (exits silently for non-mention). No mention gets dropped. ## Testing - Added `test_mention_verify_no_concurrency` — verifies the verify job has no concurrency group - Added `test_mention_handle_job_queues_not_cancels` — verifies handle uses `cancel-in-progress: false` - Both tests fail before the fix, pass after - Full suite: 110 tests pass, all pre-commit lints pass --- Closes#93 — automated triage --------- Co-authored-by: tend-agent <270458913+tend-agent@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>

snth · 2026-03-27T15:18:21Z

Force-update all integration test snapshots to current insta format. The newer insta (1.46.3) enforces a stricter snapshot format, causing CI to fail with 'existing value is in a legacy format'.

@r

Update inline snapshot string delimiters (@r" -> @", r### -> r#) and remove deprecated snapshot_kind: text metadata from snap files. All changes are format-only with no semantic differences. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Non-format changes from insta snapshot regeneration: - compile/fmt append_select_nulls.snap: expression field corrected to match current .prql file (stale '# duckdb:skip' / '# postgres:skip' comments were removed from the file in a prior commit but the snap metadata was never regenerated) - compileall snap files: leading blank lines added to snapshot content. Each blank line represents a dialect whose output matches the generic dialect (empty diff). The previous insta version trimmed these; the new version preserves them. The actual diff content is unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T15:48:47Z

This final commit c971aff is identical to 5f1de96 , I just split it into two commits to separate semantic changes from regenerating the snapshots to pure formatting changes. All tests passed for 5f1de96 but now some are failing for c971aff. What gives?

max-sixty · 2026-03-27T16:18:10Z

great! will look at the snapshot issue

@prql-bot are you online?!

prql-bot · 2026-03-27T16:20:00Z

I'm online! 👋

Regarding @snth's question about the CI failure on c971aff — the failure is not related to the snapshot changes. It's a Docker Hub authentication error in the test-dbs-external job:

Head "https://registry-1.docker.io/v2/library/postgres/manifests/alpine": unauthorized: incorrect username or password

This is an infrastructure/credentials issue pulling the postgres container image. All the actual Rust tests, lints, builds, and code coverage checks passed successfully. A re-run of that job should fix it (assuming Docker Hub credentials are refreshed or the rate limit clears).

Authenticate with Docker Hub before pulling images to avoid rate limits. - test-rust.yaml: login before docker compose up (postgres, mysql, clickhouse) - build-devcontainer.yaml: login before image builds as defensive measure Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The Docker Hub login step in test-rust and build-devcontainer workflows needs access to DOCKERHUB_USERNAME and DOCKERHUB_TOKEN secrets. These are only available via 'secrets: inherit' on workflow_call invocations. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T17:44:13Z

@max-sixty I set up the Docker Hub account and added the details to the repo secrets. Should be good to merge now.

max-sixty · 2026-03-27T19:44:09Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

max-sixty · 2026-03-27T19:44:14Z

thanks @snth !

max-sixty · 2026-03-28T20:54:24Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

@prql-bot ?

Skip empty diffs (where dialect output matches generic) instead of including them, which produced leading/intermediate blank lines in the snapshot files. Closes follow-up from #5750. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

prql-bot · 2026-03-28T21:12:09Z

Done — opened #5752 to fix this at the source. The compileall test was collecting empty diffs (where a dialect's output matches generic) and joining them with newlines, producing the leading blank lines. The fix skips empty diffs entirely.

@snth

## Problem The `tend-mention` workflow's `verify` job has a concurrency group keyed by PR number with `cancel-in-progress: true`. When two comments arrive on the same PR within seconds, the second run cancels the first — even if the first was an explicit `@bot` mention. This is exactly what happened at [PRQL/prql#5750](PRQL/prql#5750 (comment)): 1. **19:44:09** — `@prql-bot can you do a follow-up...` → triggered run [23664429862](https://github.com/PRQL/prql/actions/runs/23664429862) 2. **19:44:14** — `thanks @snth !` (5 seconds later) → triggered run [23664431397](https://github.com/PRQL/prql/actions/runs/23664431397) 3. Run 2 **cancelled** run 1 via the concurrency group 4. Run 2 processed "thanks @snth" (no bot mention) → bot exited silently. The explicit `@prql-bot` request was lost. ## Solution - **Remove concurrency from the `verify` job** — it's lightweight and fast; parallel verify jobs for the same PR are harmless. - **Change `handle`'s `cancel-in-progress` to `false`** — jobs queue instead of cancelling. The first handle completes (processes the mention), then the second runs (exits silently for non-mention). No mention gets dropped. ## Testing - Added `test_mention_verify_no_concurrency` — verifies the verify job has no concurrency group - Added `test_mention_handle_job_queues_not_cancels` — verifies handle uses `cancel-in-progress: false` - Both tests fail before the fix, pass after - Full suite: 110 tests pass, all pre-commit lints pass --- Closes#93 — automated triage --------- Co-authored-by: tend-agent <270458913+tend-agent@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>

snth · 2026-03-27T15:18:21Z

Force-update all integration test snapshots to current insta format. The newer insta (1.46.3) enforces a stricter snapshot format, causing CI to fail with 'existing value is in a legacy format'.

@r

Update inline snapshot string delimiters (@r" -> @", r### -> r#) and remove deprecated snapshot_kind: text metadata from snap files. All changes are format-only with no semantic differences. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Non-format changes from insta snapshot regeneration: - compile/fmt append_select_nulls.snap: expression field corrected to match current .prql file (stale '# duckdb:skip' / '# postgres:skip' comments were removed from the file in a prior commit but the snap metadata was never regenerated) - compileall snap files: leading blank lines added to snapshot content. Each blank line represents a dialect whose output matches the generic dialect (empty diff). The previous insta version trimmed these; the new version preserves them. The actual diff content is unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T15:48:47Z

This final commit c971aff is identical to 5f1de96 , I just split it into two commits to separate semantic changes from regenerating the snapshots to pure formatting changes. All tests passed for 5f1de96 but now some are failing for c971aff. What gives?

max-sixty · 2026-03-27T16:18:10Z

great! will look at the snapshot issue

@prql-bot are you online?!

prql-bot · 2026-03-27T16:20:00Z

I'm online! 👋

Regarding @snth's question about the CI failure on c971aff — the failure is not related to the snapshot changes. It's a Docker Hub authentication error in the test-dbs-external job:

Head "https://registry-1.docker.io/v2/library/postgres/manifests/alpine": unauthorized: incorrect username or password

This is an infrastructure/credentials issue pulling the postgres container image. All the actual Rust tests, lints, builds, and code coverage checks passed successfully. A re-run of that job should fix it (assuming Docker Hub credentials are refreshed or the rate limit clears).

Authenticate with Docker Hub before pulling images to avoid rate limits. - test-rust.yaml: login before docker compose up (postgres, mysql, clickhouse) - build-devcontainer.yaml: login before image builds as defensive measure Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The Docker Hub login step in test-rust and build-devcontainer workflows needs access to DOCKERHUB_USERNAME and DOCKERHUB_TOKEN secrets. These are only available via 'secrets: inherit' on workflow_call invocations. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T17:44:13Z

@max-sixty I set up the Docker Hub account and added the details to the repo secrets. Should be good to merge now.

max-sixty · 2026-03-27T19:44:09Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

max-sixty · 2026-03-27T19:44:14Z

thanks @snth !

max-sixty · 2026-03-28T20:54:24Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

@prql-bot ?

Skip empty diffs (where dialect output matches generic) instead of including them, which produced leading/intermediate blank lines in the snapshot files. Closes follow-up from #5750. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

prql-bot · 2026-03-28T21:12:09Z

Done — opened #5752 to fix this at the source. The compileall test was collecting empty diffs (where a dialect's output matches generic) and joining them with newlines, producing the leading blank lines. The fix skips empty diffs entirely.

@snth

## Problem The `tend-mention` workflow's `verify` job has a concurrency group keyed by PR number with `cancel-in-progress: true`. When two comments arrive on the same PR within seconds, the second run cancels the first — even if the first was an explicit `@bot` mention. This is exactly what happened at [PRQL/prql#5750](PRQL/prql#5750 (comment)): 1. **19:44:09** — `@prql-bot can you do a follow-up...` → triggered run [23664429862](https://github.com/PRQL/prql/actions/runs/23664429862) 2. **19:44:14** — `thanks @snth !` (5 seconds later) → triggered run [23664431397](https://github.com/PRQL/prql/actions/runs/23664431397) 3. Run 2 **cancelled** run 1 via the concurrency group 4. Run 2 processed "thanks @snth" (no bot mention) → bot exited silently. The explicit `@prql-bot` request was lost. ## Solution - **Remove concurrency from the `verify` job** — it's lightweight and fast; parallel verify jobs for the same PR are harmless. - **Change `handle`'s `cancel-in-progress` to `false`** — jobs queue instead of cancelling. The first handle completes (processes the mention), then the second runs (exits silently for non-mention). No mention gets dropped. ## Testing - Added `test_mention_verify_no_concurrency` — verifies the verify job has no concurrency group - Added `test_mention_handle_job_queues_not_cancels` — verifies handle uses `cancel-in-progress: false` - Both tests fail before the fix, pass after - Full suite: 110 tests pass, all pre-commit lints pass --- Closes#93 — automated triage --------- Co-authored-by: tend-agent <270458913+tend-agent@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>

snth · 2026-03-27T15:18:21Z

Force-update all integration test snapshots to current insta format. The newer insta (1.46.3) enforces a stricter snapshot format, causing CI to fail with 'existing value is in a legacy format'.

@r

Update inline snapshot string delimiters (@r" -> @", r### -> r#) and remove deprecated snapshot_kind: text metadata from snap files. All changes are format-only with no semantic differences. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Non-format changes from insta snapshot regeneration: - compile/fmt append_select_nulls.snap: expression field corrected to match current .prql file (stale '# duckdb:skip' / '# postgres:skip' comments were removed from the file in a prior commit but the snap metadata was never regenerated) - compileall snap files: leading blank lines added to snapshot content. Each blank line represents a dialect whose output matches the generic dialect (empty diff). The previous insta version trimmed these; the new version preserves them. The actual diff content is unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T15:48:47Z

This final commit c971aff is identical to 5f1de96 , I just split it into two commits to separate semantic changes from regenerating the snapshots to pure formatting changes. All tests passed for 5f1de96 but now some are failing for c971aff. What gives?

max-sixty · 2026-03-27T16:18:10Z

great! will look at the snapshot issue

@prql-bot are you online?!

prql-bot · 2026-03-27T16:20:00Z

I'm online! 👋

Regarding @snth's question about the CI failure on c971aff — the failure is not related to the snapshot changes. It's a Docker Hub authentication error in the test-dbs-external job:

Head "https://registry-1.docker.io/v2/library/postgres/manifests/alpine": unauthorized: incorrect username or password

This is an infrastructure/credentials issue pulling the postgres container image. All the actual Rust tests, lints, builds, and code coverage checks passed successfully. A re-run of that job should fix it (assuming Docker Hub credentials are refreshed or the rate limit clears).

Authenticate with Docker Hub before pulling images to avoid rate limits. - test-rust.yaml: login before docker compose up (postgres, mysql, clickhouse) - build-devcontainer.yaml: login before image builds as defensive measure Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The Docker Hub login step in test-rust and build-devcontainer workflows needs access to DOCKERHUB_USERNAME and DOCKERHUB_TOKEN secrets. These are only available via 'secrets: inherit' on workflow_call invocations. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

snth · 2026-03-27T17:44:13Z

@max-sixty I set up the Docker Hub account and added the details to the repo secrets. Should be good to merge now.

max-sixty · 2026-03-27T19:44:09Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

max-sixty · 2026-03-27T19:44:14Z

thanks @snth !

max-sixty · 2026-03-28T20:54:24Z

@prql-bot can you do a follow-up to change the code that produces those snapshots to avoid producing the blank lines at the top? fix at the source please

@prql-bot ?

Skip empty diffs (where dialect output matches generic) instead of including them, which produced leading/intermediate blank lines in the snapshot files. Closes follow-up from #5750. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

prql-bot · 2026-03-28T21:12:09Z

Done — opened #5752 to fix this at the source. The compileall test was collecting empty diffs (where a dialect's output matches generic) and joining them with newlines, producing the leading blank lines. The fix skips empty diffs entirely.

@snth

## Problem The `tend-mention` workflow's `verify` job has a concurrency group keyed by PR number with `cancel-in-progress: true`. When two comments arrive on the same PR within seconds, the second run cancels the first — even if the first was an explicit `@bot` mention. This is exactly what happened at [PRQL/prql#5750](PRQL/prql#5750 (comment)): 1. **19:44:09** — `@prql-bot can you do a follow-up...` → triggered run [23664429862](https://github.com/PRQL/prql/actions/runs/23664429862) 2. **19:44:14** — `thanks @snth !` (5 seconds later) → triggered run [23664431397](https://github.com/PRQL/prql/actions/runs/23664431397) 3. Run 2 **cancelled** run 1 via the concurrency group 4. Run 2 processed "thanks @snth" (no bot mention) → bot exited silently. The explicit `@prql-bot` request was lost. ## Solution - **Remove concurrency from the `verify` job** — it's lightweight and fast; parallel verify jobs for the same PR are harmless. - **Change `handle`'s `cancel-in-progress` to `false`** — jobs queue instead of cancelling. The first handle completes (processes the mention), then the second runs (exits silently for non-mention). No mention gets dropped. ## Testing - Added `test_mention_verify_no_concurrency` — verifies the verify job has no concurrency group - Added `test_mention_handle_job_queues_not_cancels` — verifies handle uses `cancel-in-progress: false` - Both tests fail before the fix, pass after - Full suite: 110 tests pass, all pre-commit lints pass --- Closes#93 — automated triage --------- Co-authored-by: tend-agent <270458913+tend-agent@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>

snth marked this pull request as draft March 27, 2026 15:20

snthand others added 2 commits March 27, 2026 17:37

snth force-pushed the insta-update branch from 5f1de96 to c971affCompare March 27, 2026 15:42

snthand others added 2 commits March 27, 2026 19:01

snth marked this pull request as ready for review March 27, 2026 17:43

snth requested a review from max-sixty March 27, 2026 17:43

max-sixty merged commit a799b27 into mainMar 27, 2026
82 checks passed

max-sixty deleted the insta-update branch March 27, 2026 19:44

max-sixty mentioned this pull request Mar 28, 2026
bot miss at PRQL max-sixty/tend#93
Closed

tend-agent mentioned this pull request Mar 28, 2026
fix(mention): don't cancel verify job on rapid comments max-sixty/tend#95
Merged

prql-bot mentioned this pull request Mar 28, 2026
fix: remove leading blank lines from compileall snapshots #5752
Merged
3 tasks

snth marked this pull request as draft March 27, 2026 15:20

snthand others added 2 commits March 27, 2026 17:37

snth force-pushed the insta-update branch from 5f1de96 to c971affCompare March 27, 2026 15:42

snthand others added 2 commits March 27, 2026 19:01

snth marked this pull request as ready for review March 27, 2026 17:43

snth requested a review from max-sixty March 27, 2026 17:43

max-sixty merged commit a799b27 into mainMar 27, 2026
82 checks passed

max-sixty deleted the insta-update branch March 27, 2026 19:44

max-sixty mentioned this pull request Mar 28, 2026
bot miss at PRQL max-sixty/tend#93
Closed

tend-agent mentioned this pull request Mar 28, 2026
fix(mention): don't cancel verify job on rapid comments max-sixty/tend#95
Merged

prql-bot mentioned this pull request Mar 28, 2026
fix: remove leading blank lines from compileall snapshots #5752
Merged
3 tasks

snth marked this pull request as draft March 27, 2026 15:20

snthand others added 2 commits March 27, 2026 17:37

snth force-pushed the insta-update branch from 5f1de96 to c971affCompare March 27, 2026 15:42

snthand others added 2 commits March 27, 2026 19:01

snth marked this pull request as ready for review March 27, 2026 17:43

snth requested a review from max-sixty March 27, 2026 17:43

max-sixty merged commit a799b27 into mainMar 27, 2026
82 checks passed

max-sixty deleted the insta-update branch March 27, 2026 19:44

max-sixty mentioned this pull request Mar 28, 2026
bot miss at PRQL max-sixty/tend#93
Closed

tend-agent mentioned this pull request Mar 28, 2026
fix(mention): don't cancel verify job on rapid comments max-sixty/tend#95
Merged

prql-bot mentioned this pull request Mar 28, 2026
fix: remove leading blank lines from compileall snapshots #5752
Merged
3 tasks

snth marked this pull request as draft March 27, 2026 15:20

snthand others added 2 commits March 27, 2026 17:37

snth force-pushed the insta-update branch from 5f1de96 to c971affCompare March 27, 2026 15:42

snthand others added 2 commits March 27, 2026 19:01

snth marked this pull request as ready for review March 27, 2026 17:43

snth requested a review from max-sixty March 27, 2026 17:43

max-sixty merged commit a799b27 into mainMar 27, 2026
82 checks passed

max-sixty deleted the insta-update branch March 27, 2026 19:44

max-sixty mentioned this pull request Mar 28, 2026
bot miss at PRQL max-sixty/tend#93
Closed

tend-agent mentioned this pull request Mar 28, 2026
fix(mention): don't cancel verify job on rapid comments max-sixty/tend#95
Merged

prql-bot mentioned this pull request Mar 28, 2026
fix: remove leading blank lines from compileall snapshots #5752
Merged
3 tasks

snth marked this pull request as draft March 27, 2026 15:20

snthand others added 2 commits March 27, 2026 17:37

snth force-pushed the insta-update branch from 5f1de96 to c971affCompare March 27, 2026 15:42

snthand others added 2 commits March 27, 2026 19:01

snth marked this pull request as ready for review March 27, 2026 17:43

snth requested a review from max-sixty March 27, 2026 17:43

max-sixty merged commit a799b27 into mainMar 27, 2026
82 checks passed

max-sixty deleted the insta-update branch March 27, 2026 19:44

max-sixty mentioned this pull request Mar 28, 2026
bot miss at PRQL max-sixty/tend#93
Closed

tend-agent mentioned this pull request Mar 28, 2026
fix(mention): don't cancel verify job on rapid comments max-sixty/tend#95
Merged

prql-bot mentioned this pull request Mar 28, 2026
fix: remove leading blank lines from compileall snapshots #5752
Merged
3 tasks

snth marked this pull request as draft March 27, 2026 15:20

snthand others added 2 commits March 27, 2026 17:37

snth force-pushed the insta-update branch from 5f1de96 to c971affCompare March 27, 2026 15:42

snthand others added 2 commits March 27, 2026 19:01

snth marked this pull request as ready for review March 27, 2026 17:43

snth requested a review from max-sixty March 27, 2026 17:43

max-sixty merged commit a799b27 into mainMar 27, 2026
82 checks passed

max-sixty deleted the insta-update branch March 27, 2026 19:44

max-sixty mentioned this pull request Mar 28, 2026
bot miss at PRQL max-sixty/tend#93
Closed

tend-agent mentioned this pull request Mar 28, 2026
fix(mention): don't cancel verify job on rapid comments max-sixty/tend#95
Merged

prql-bot mentioned this pull request Mar 28, 2026
fix: remove leading blank lines from compileall snapshots #5752
Merged
3 tasks

snth marked this pull request as draft March 27, 2026 15:20

snthand others added 2 commits March 27, 2026 17:37

snth force-pushed the insta-update branch from 5f1de96 to c971affCompare March 27, 2026 15:42

snthand others added 2 commits March 27, 2026 19:01

snth marked this pull request as ready for review March 27, 2026 17:43

snth requested a review from max-sixty March 27, 2026 17:43

max-sixty merged commit a799b27 into mainMar 27, 2026
82 checks passed

max-sixty deleted the insta-update branch March 27, 2026 19:44

max-sixty mentioned this pull request Mar 28, 2026
bot miss at PRQL max-sixty/tend#93
Closed

tend-agent mentioned this pull request Mar 28, 2026
fix(mention): don't cancel verify job on rapid comments max-sixty/tend#95
Merged

prql-bot mentioned this pull request Mar 28, 2026
fix: remove leading blank lines from compileall snapshots #5752
Merged
3 tasks

snth marked this pull request as draft March 27, 2026 15:20

snthand others added 2 commits March 27, 2026 17:37

snth force-pushed the insta-update branch from 5f1de96 to c971affCompare March 27, 2026 15:42

snthand others added 2 commits March 27, 2026 19:01

snth marked this pull request as ready for review March 27, 2026 17:43

snth requested a review from max-sixty March 27, 2026 17:43

max-sixty merged commit a799b27 into mainMar 27, 2026
82 checks passed

max-sixty deleted the insta-update branch March 27, 2026 19:44

max-sixty mentioned this pull request Mar 28, 2026
bot miss at PRQL max-sixty/tend#93
Closed

tend-agent mentioned this pull request Mar 28, 2026
fix(mention): don't cancel verify job on rapid comments max-sixty/tend#95
Merged

prql-bot mentioned this pull request Mar 28, 2026
fix: remove leading blank lines from compileall snapshots #5752
Merged
3 tasks

Uh oh!

Conversation

snth commented Mar 27, 2026

Uh oh!

snth commented Mar 27, 2026

Uh oh!

max-sixty commented Mar 27, 2026

Uh oh!

prql-bot commented Mar 27, 2026

Uh oh!

snth commented Mar 27, 2026

Uh oh!

max-sixty commented Mar 27, 2026

Uh oh!

max-sixty commented Mar 27, 2026

Uh oh!

Uh oh!

max-sixty commented Mar 28, 2026

Uh oh!

prql-bot commented Mar 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

snth commented Mar 27, 2026

Uh oh!

snth commented Mar 27, 2026

Uh oh!

max-sixty commented Mar 27, 2026

Uh oh!

prql-bot commented Mar 27, 2026

Uh oh!

snth commented Mar 27, 2026

Uh oh!

max-sixty commented Mar 27, 2026

Uh oh!

max-sixty commented Mar 27, 2026

Uh oh!

Uh oh!

max-sixty commented Mar 28, 2026

Uh oh!

prql-bot commented Mar 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

snth commented Mar 27, 2026

Uh oh!

snth commented Mar 27, 2026

Uh oh!

max-sixty commented Mar 27, 2026

Uh oh!

prql-bot commented Mar 27, 2026

Uh oh!

snth commented Mar 27, 2026

Uh oh!

max-sixty commented Mar 27, 2026

Uh oh!

max-sixty commented Mar 27, 2026

Uh oh!

Uh oh!

max-sixty commented Mar 28, 2026

Uh oh!

prql-bot commented Mar 28, 2026

Uh oh!

Reviewers