roachtest: add O-rsg to random syntax tests #109589
Conversation
mgartner
requested review from
herkolategan and
renatolabs
and removed request for
a team
|
This change is |
There was a problem hiding this comment.
This seems to create a new field on the generic TestSpec struct (used by all tests across all teams) for a feature used exclusively (correct me if I'm wrong) by the SQL team.
I also wonder if the concept of a RandomSyntaxTest gets a little blurry. I'm thinking of cases where randomness plays a role in the SQL statement used (e.g., create a backup or a changefeed with certain options randomly) but that probably wouldn't be categorized as a RandomSyntaxTest as defined here, if I understand correctly.
For consideration: what if, instead, we allowed the TestSpec to include ExtraLabels to be attached to GitHub issues when the test failed? You'd still be able to achieve your goal of adding O-rsg to some tests, while allowing the feature to be used by other teams and would remove the potential ambiguity around what defines a "random syntax test".
Would also be nice to add tests for this new field (see github_test.go).
|
Can be in a separate PR, but just noting we'll also need to add this label to non-roachtests, specifically the ones in |
Great idea. Done.
Thanks for pointing this out. I'll do that in a separate PR. |
There was a problem hiding this comment.
Test is missing a datadriven file (see echotest.Require call), but otherwise
Reviewable status:
complete! 1 of 0 LGTMs obtained (waiting on @herkolategan and @rafiss)
|
TFTR! bors r+ |
|
Build failed: |
The `O-rsg` label is now applied to issues created when a test with randomized SQL syntax fails: sqlsmith, TLP, costfuzz, and unoptimized query oracle tests. Release note: None
|
bors r+ |
|
Build succeeded: |
The
O-rsglabel is now applied to issues created when a test withrandomized SQL syntax fails: sqlsmith, TLP, costfuzz, and unoptimized
query oracle tests.
Epic: None
Release note: None