feat(issues): add next checkout after last seen - #2054
Conversation
alanpeixinho
left a comment
alanpeixinho
left a comment
There was a problem hiding this comment.
looking great so far
| """ | ||
|
|
||
| with connection.cursor() as cursor: | ||
| cursor.execute(query, params) |
There was a problem hiding this comment.
might be worth adding cache here
|
|
||
| query = """ | ||
| WITH last_seen AS ( | ||
| SELECT DISTINCT ON (IC.issue_id) |
There was a problem hiding this comment.
we might be able to be faster using a subquery to get strictly the next checkout on subquery.
This suggestion was made on cursor, so we need to validate if it is correct.
WITH last_seen AS (
SELECT DISTINCT ON (IC.issue_id)
IC.issue_id,
C.id AS last_checkout_id,
C.start_time AS last_start_time,
C.origin,
C.tree_name,
C.git_repository_url,
C.git_repository_branch
FROM
incidents IC
LEFT JOIN tests T ON IC.test_id = T.id
LEFT JOIN builds B ON (
IC.build_id = B.id
OR T.build_id = B.id
)
LEFT JOIN checkouts C ON B.checkout_id = C.id
WHERE
IC.issue_id = ANY(ARRAY['issue-id-1', 'issue-id-2']) -- %(issue_id_list)s
ORDER BY
IC.issue_id,
IC.issue_version DESC,
IC._timestamp DESC
)
SELECT
LS.issue_id,
C.id AS checkout_id,
C.start_time,
C.git_commit_hash,
C.git_commit_name,
C.git_repository_url,
C.git_repository_branch,
C.tree_name,
C.origin
FROM
last_seen LS
CROSS JOIN LATERAL (
SELECT
C.id,
C.start_time,
C.git_commit_hash,
C.git_commit_name,
C.git_repository_url,
C.git_repository_branch,
C.tree_name,
C.origin
FROM
checkouts C
WHERE
C.origin = LS.origin
AND C.tree_name IS NOT DISTINCT FROM LS.tree_name
AND C.git_repository_url IS NOT DISTINCT FROM LS.git_repository_url
AND C.git_repository_branch IS NOT DISTINCT FROM LS.git_repository_branch
AND C.start_time > LS.last_start_time
ORDER BY
C.start_time ASC,
C.id ASC
LIMIT 1
) C
WHERE
LS.last_checkout_id IS NOT NULL
AND LS.last_start_time IS NOT NULL;
In simple tests it was reasonably faster. But at the same time, the issues query is not a slow one. So fell free to ignore if it can make the endpoint more complex.
### Original
Planning:
Buffers: shared hit=53 read=1
Planning Time: 1.421 ms
Execution Time: 1499.004 ms
### Subquery
Planning:
Buffers: shared hit=54
Planning Time: 1.093 ms
Execution Time: 73.985 ms
There was a problem hiding this comment.
you are right. I don't think it makes the endpoint more complex. Done.
| }); | ||
| }, [data?.extra, formatMessage, issueId]); | ||
|
|
||
| const nextCheckoutSection: ISection | undefined = useMemo(() => { |
There was a problem hiding this comment.
Unrelated to this task. But at some point I think we need to do a full layout improvement on this page.
This presented information is already quite sparse, and could find a better way to organize it.
Again, for this specific task, we should focus on just providing the information.
There was a problem hiding this comment.
I agree this page needs improvements. I'll open an issue for it.
Expose the earliest later checkout on the same tree in issue extras, using start_time since parent-commit hierarchy is unavailable (kernelci#1958).
Add a Next Checkout section after last-seen incident data using the extras payload (kernelci#1958).
Avoid scanning every later checkout on the tree and reuse the same cache pattern as first/last seen. Signed-off-by: Felipe Bergamin <felipebergamin@profusion.mobi>
dcf3cf8 to
4b74c37
Compare
felipebergamin
changed the title
2 participants
Expose the earliest later checkout on the same tree in issue extras, using start_time since parent-commit hierarchy is unavailable (#1958).
Visual reference