Conversation
There was a problem hiding this comment.
Code Review
This pull request introduces a new Bash script, scripts/create_issue.sh, designed to automate the creation of GitHub issues when CI workflows fail. The script utilizes the GitHub CLI to gather context about the failure and the associated pull request. Reviewers suggested several improvements to enhance the script's robustness and flexibility, including adopting set -o pipefail, using modern Bash conditional syntax [[ ... ]], and replacing hardcoded branch names with environment variables. Additionally, a recommendation was made to include logic that prevents the creation of duplicate issues for repeated failures.
| gh issue create \ | ||
| --title "$TITLE" \ | ||
| --body "$BODY" \ | ||
| --label "$LABEL_NAME" |
There was a problem hiding this comment.
This script creates a new issue every time a workflow fails. For scheduled workflows that might fail repeatedly, this can lead to many duplicate issues. Consider adding logic to check if an open issue with the same label already exists using gh issue list --label "$LABEL_NAME" --state open before creating a new one.
There was a problem hiding this comment.
This is a reasonable enhancement. What do you think? I'd check for more than the label, though: I'd check for one with the label and an identical title.
There was a problem hiding this comment.
The script is already checking for label. Added check for title. Thank you!
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
| # - catch regressions introduced by changes in environment | ||
| # - have more data for observing dynamics of performance degradation | ||
| schedule: | ||
| - cron: "0 * * * *" # hourly |
There was a problem hiding this comment.
I don't think this is a good use of our AI budget. The next time we submit something that triggers evals, it will catch any updated environment changes anyhow, so this just feels like a waste.
It also seems way too fast. If we did this, I'd run it at most once a day, since otherwise a single change in environment could result in 48 issues being filed over a weekend.
There was a problem hiding this comment.
Switched to daily. Thanks.
| # Do not run on forked branches, | ||
| # because the test does not have access to secrets in forks. | ||
| if: github.repository == 'google/a2ui' | ||
| if: github.repository == 'flutter/genui' |
There was a problem hiding this comment.
Whoa, so I guess it wasn't running?
There was a problem hiding this comment.
nope
we are in A2UI
reverted
| actions: read | ||
|
|
||
| if: github.repository == 'google/A2UI' | ||
| if: github.repository == 'flutter/genui' |
There was a problem hiding this comment.
Where do our docs end up? Is it published anywhere?
There was a problem hiding this comment.
oh, I missed it
reverted
thank you
| gh issue create \ | ||
| --title "$TITLE" \ | ||
| --body "$BODY" \ | ||
| --label "$LABEL_NAME" |
There was a problem hiding this comment.
This is a reasonable enhancement. What do you think? I'd check for more than the label, though: I'd check for one with the label and an identical title.
Fixes #1440