feat(relevant warnings) Predict relevant warnings #1882
Conversation
| } | ||
| headers = get_codecov_auth_header(request) | ||
| requests.post( | ||
| url="https://overwatch.codecov.dev/api/ai/seer/relevant-warnings", |
There was a problem hiding this comment.
At the very end, we can point this at prod overwatch (not deployed yet), which sounds like it'll end up at overwatch.codecov.io
|
|
||
|
|
||
| # Copied from https://github.com/codecov/bug-prediction-research/blob/main/src/core/typings.py | ||
| class Location(BaseModel): |
There was a problem hiding this comment.
Depending on the tool we can have the full code region (start_line, start_column, end_line, end_column)
Would we be interested in that?
There was a problem hiding this comment.
hmm good question. i don't think start_column or end_column are particularly useful rn
i realized it would prolly simplify things if the encoded_location string (which matches the regex in Location) was instead a:
from pydantic import BaseModel
class Location(BaseModel):
filename: str
start_line: str
end_line: strso a warning in the overwatch request has location={"filename": ..., "start_line": ..., "end_line": ...} instead of encoded_location. can also include start_column, end_column keys since pydantic will just ignore them if they're not in the model
low priority tho, so up to you if you wanna make this change in overwatch
Co-authored-by: Jenn Mueng <30991498+jennmueng@users.noreply.github.com>
Co-authored-by: Jenn Mueng <30991498+jennmueng@users.noreply.github.com>
Co-authored-by: Jenn Mueng <30991498+jennmueng@users.noreply.github.com>
Needs getsentry/sentry#85133 to go in first
Background: tech spec
Currently, the tests consist of mocks and patches. In the future, can store real data and VCR tests like autofix
demo