Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

bazel: how do we stress tests w/ Bazel? #67165

Closed
rickystewart opened this issue Jul 1, 2021 · 0 comments · Fixed by #67674
Closed

bazel: how do we stress tests w/ Bazel? #67165

rickystewart opened this issue Jul 1, 2021 · 0 comments · Fixed by #67674
Labels
A-build-system C-enhancement Solution expected to add code/behavior + preserve backward-compat (pg compat issues are exception) T-dev-inf

Comments

@rickystewart
Copy link
Collaborator

rickystewart commented Jul 1, 2021

blocks #67164

Epic CRDB-8036

@rickystewart rickystewart added C-enhancement Solution expected to add code/behavior + preserve backward-compat (pg compat issues are exception) A-build-system T-dev-inf labels Jul 1, 2021
craig bot pushed a commit that referenced this issue Jul 15, 2021
67197: sql: improve expression index usability r=mgartner a=mgartner

#### sql: do not allow renaming inaccessible columns

Release note: None

#### sql: do not allow inaccessible columns in primary keys

Release note: None

#### sql: support expressions in unique constraints

Postgres does not allow expressions in unique constraints, but it does
allow expressions in unique indexes. For now we allow expressions in
unique constraints, because we cannot differentiate between a unique
constraint and a unique index table definition: they are both parsed
into the same struct, `tree.UniqueConstraintTableDef`.

In the long term we may want to disallow expression in unique
constraints to be consistent with Postgres. We could do this by changing
the parser so that `UNIQUE ((a + b))` does not parse successfully.
See #65825.

Release note: None

#### sql: better error message when creating a view with an inaccessible column

Previously, the error message returned when attempting to create a view
with an inaccessible column was confusing:

    unimplemented: views do not currently support * expressions

The error message now reads:

    column "x" does not exist

Release note: None

#### sql: do not allow inaccessible columns to be dropped

Release note: None


67631: dev: teach `dev` how to run benchmarks r=rail a=rickystewart

Also a little bit of refactoring, and adding some code to make the code
less stringent (e.g. now you can just do
`dev build //pkg/cmd/cockroach-short` and it will succeed).

Closes #67141.

Release note: None

67638: colexec: improve sort operator a bit r=yuzefovich a=yuzefovich

**colexec: improve sort operator a bit**

Previously, we were missing memory accounting around some slices
allocated internally by the sort operator which is now added.
Additionally, the references to those slices are now kept by the
operator which will be useful when the sorter is used by the external
sorter.

Additionally, this commit eliminates some bounds checks.

Release note: None

**colexec: add some BCE assertions for distinct and sort**

This is achieved by templating the inlined execgen functions. Note that
in some cases this templating leaves redundant `_ = true` lines, but
figuring out what's up with that is left as a TODO.

Additionally, this commit removes some redundant attempts at achieving
BCE for non-sliceable vectors (which had a negative impact in case of
JSONs) and clarifies the name of a utility function.

Release note: None

67651: sql: require placeholder types to be identical to use a cached plan r=RaduBerinde a=rafiss

fixes #67605

Release note (bug fix): Fix a bug where a prepared statement could
incorrectly reuse the query plan of a different prepared statements that
had similar, but not identical type hints.

67674: dev: teach `dev` how to `stress` r=rail a=rickystewart

Only the latest commit counts for this review, the other is from #67631.

The previous version of this code would assume that you had `stress`
installed globally -- this doesn't work in the long term, so instead if
you're stress testing pre-build the `stress` binary and pass it in to
`bazel test` invocation.

Closes #67165.

Release note: None

67675: dev: get rid of fuzzing stubs r=rail a=rickystewart

I don't think anyone actually uses fuzzing and it doesn't seem to be a
requirement for `dev`. Until such a point as we actually need it for
something, remove these stubs.

Release note: None

Co-authored-by: Marcus Gartner <[email protected]>
Co-authored-by: Ricky Stewart <[email protected]>
Co-authored-by: Yahor Yuzefovich <[email protected]>
Co-authored-by: Rafi Shamim <[email protected]>
@craig craig bot closed this as completed in fce4c8d Jul 15, 2021
pawalt pushed a commit to pawalt/cockroach that referenced this issue Jul 22, 2021
The previous version of this code would assume that you had `stress`
installed globally -- this doesn't work in the long term, so instead if
you're stress testing pre-build the `stress` binary and pass it in to
`bazel test` invocation.

Closes cockroachdb#67165.

Release note: None
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
A-build-system C-enhancement Solution expected to add code/behavior + preserve backward-compat (pg compat issues are exception) T-dev-inf
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant