prisma/prisma-engines - CodSpeed

prisma-engines

Blog Docs Changelog

Performance History

Latest Results

chore: trigger an engines integration release [integration] Signed-off-by: Alexey Orlenko's AI Agent <robot@aqrln.net>

shadow-db-safety-s4

3 days ago

test(schema-engine): pin where consent stops buying a reset A thousand rows and one is more than the engine will destroy on anyone's say-so: the run stops with P3027, the message names the limit and a shadow database URL with no credentials in it, and the rows are all still there afterwards — consent bought a reset, and the reset did not happen. A thousand rows exactly is not more than a thousand, so that run goes through and leaves the shadow database empty, which is the boundary the number describes. The rows are inserted a few hundred at a time rather than generated by the database: every flavour spells generating them differently, and one of them caps a recursive query at a thousand rows — the very number these tests are about. Signed-off-by: Alexey Orlenko's AI Agent <robot@aqrln.net>

shadow-db-safety-s4

3 days ago

test(schema-engine): cover the shadow database consent gate The dirty shadow database is a table the tests put there themselves, standing in for whatever a real database holds: replaying a migration history into it is refused with P3026 on the dev path and on the diff path alike, and the table is still there afterwards, which is the whole point. A shadow database holding nothing but `_prisma_migrations` is refused too — a migration history in a database the engine did not just write is either a real database or the remains of a run that crashed. The refusal names the database and not the way in: the diff-path test gives the shadow database a login of its own and asserts that neither the user nor the password reaches the message. With consent the same setup goes through and the shadow database is left empty. SQLite gets its own test for the shape it did not have before — a file holding a table the migration history also creates is refused, and consented to, reset, and replayed into cleanly. Signed-off-by: Alexey Orlenko's AI Agent <robot@aqrln.net>

shadow-db-safety-s2

4 days ago

test(schema-engine): cover the same-database shadow database guard The diff-path proof is the incident itself: a from-migrations diff whose shadow database is the database being diffed now fails with P3025, and the table that was created before the diff is still there afterwards. Both collision shapes are covered — the datasource URL and a URL diff target — as is the differently spelled spelling of the same database, and a separate shadow database on the same server still works. On the dev path, each flavour gets the spellings of its own connection strings that leave the database identity intact: the `postgres://` alias, host case, a different `?schema=`, extra parameters, SQLite paths written with and without the `file:` scheme. The default-port spelling stays a URL-level unit test, since the test servers do not listen on their flavour's default port. Three diff tests configured the database under test as their own shadow database, which is what the guard exists to refuse; they now ask for a shadow database of their own. That also makes the one that diffs a migration history against a live URL meaningful: it used to compare the wiped database with itself. Signed-off-by: Alexey Orlenko's AI Agent <robot@aqrln.net>

shadow-db-safety

4 days ago

ci: track prisma/prisma's v7 branch for driver-adapter tests (#5850) The Prisma 7 codebase is moving from prisma/prisma’s `main` branch to a `v7` branch; prisma-next’s tree will become prisma/prisma’s new `main`. This repository clones prisma/prisma to build the driver-adapter test kit used by the query-compiler tests, and consumes the checkout through the relative package paths in `libs/driver-adapters/pnpm-workspace.yaml`. Those paths only resolve against the Prisma 7 package layout, so after the swap a clone of `main` would no longer satisfy the workspace and the whole QC suite would fail to build. Hence the default branch here is load-bearing, not cosmetic. ## Changes - `Makefile`: `PRISMA_BRANCH ?= v7`, and `ensure-prisma-present` now fetches/diffs against `origin/$(PRISMA_BRANCH)` instead of a hardcoded `main` (including the divergence warning text). - `.github/workflows/select-prisma-branch.yml`: the default is `v7`, including the guard that skips redundant overrides. - `README.md`: the driver-adapter/QC testing section and the sibling-PR merge order now refer to prisma/prisma’s `v7` branch. References to this repository’s own `main` branch are untouched. ## Notes - **Safe to merge immediately.** `v7` already exists on prisma/prisma and currently points at the same commit as `main`, so nothing changes behaviourally until the swap happens. - **The per-PR branch override is unaffected.** A branch named explicitly in the PR description (the slash command documented in the README) still wins over the default. - Companion PRs: prisma/prisma (workflow triggers for the `v7` branch) and prisma/engines-wrapper (dispatch with `ref: v7`). --------- Signed-off-by: Alexey Orlenko's AI Agent <robot@aqrln.net>

main

6 days ago

ci: pass PRISMA_BRANCH to recipes as a shell variable In CI the value can come from the per-PR branch override in the pull request description, so interpolating it into the recipe text lets make expand whatever it contains. Export the variable and reference it as a quoted shell variable so it reaches the shell as data instead of source. Signed-off-by: Alexey Orlenko's AI Agent <robot@aqrln.net>

ci/prisma-branch-v7

6 days ago

ci: track prisma/prisma's v7 branch for driver-adapter tests The Prisma 7 codebase is moving from prisma/prisma's `main` branch to a `v7` branch, and prisma-next's tree will take over as the new `main`. This repository clones prisma/prisma to build the driver-adapter test kit for the query-compiler tests, and the checkout is consumed through the relative package paths in `libs/driver-adapters/pnpm-workspace.yaml`. Those paths only resolve against the Prisma 7 package layout, so once the swap happens a clone of `main` would no longer satisfy the workspace and the entire QC suite would fail to build. Point the default at `v7` in the Makefile, in the reusable branch selection workflow, and in the README. The per-PR `/prisma-branch <name>` override in the PR description keeps working as before. Signed-off-by: Alexey Orlenko's AI Agent <robot@aqrln.net>

ci/prisma-branch-v7

6 days ago

ci: track prisma/prisma's v7 branch for driver-adapter tests The Prisma 7 codebase is moving from prisma/prisma's `main` branch to a `v7` branch, and prisma-next's tree will take over as the new `main`. This repository clones prisma/prisma to build the driver-adapter test kit for the query-compiler tests, and the checkout is consumed through the relative package paths in `libs/driver-adapters/pnpm-workspace.yaml`. Those paths only resolve against the Prisma 7 package layout, so once the swap happens a clone of `main` would no longer satisfy the workspace and the entire QC suite would fail to build. Point the default at `v7` in the Makefile, in the reusable branch selection workflow, and in the README. The per-PR `/prisma-branch <name>` override in the PR description keeps working as before. Signed-off-by: Alexey Orlenko's AI Agent <robot@aqrln.net>

ci/prisma-branch-v7

6 days ago

Latest Branches

0%

feat(schema-engine): refuse to reset a shadow database that holds real data#5853

3 days ago

36b57e9

shadow-db-safety-s4

0%

feat(schema-engine): require an external shadow database to be empty or consented before reset#5852

4 days ago

57f582c

shadow-db-safety-s2

0%

fix(schema-engine): refuse a shadow database that is the main database#5851

4 days ago

7a245d5

shadow-db-safety

© 2026 CodSpeed Technology

Home Terms Privacy Docs