Scientific paper proposal: FlakyGuard (ASE 2025), week 2 - #2935
Conversation
algomaster99
left a comment
algomaster99
left a comment
There was a problem hiding this comment.
This looks very well in scope, and the paper's abstract is quite interesting. Just dropped a minor wording change.
Co-authored-by: Aman Sharma <amansha@kth.se>
|
@algomaster99 Thank you for the suggestion, updated. |
|
Closing and reopening PR to retrigger CI. |
algomaster99
left a comment
algomaster99
left a comment
There was a problem hiding this comment.
@LeuisKen and Matyas, the proposal looks good and you can start making a presentation on it. I am approving it, but I can't merge due to canvas issues which I will fix soon. After the fix, I am happy to merge your PR.
|
@LeuisKen and Matyas, sorry for such noisy activity history, but I think I have fixed the bugs. So I am merging it and it hopefully lands in canvas too. |
|
It does! You can now officially prepare your presentation. |
|
@algomaster99 Thanks! |
… merge Testing KTH/github-canvas-integration-devops#7 against PR KTH#2935 since we lack merge access there. Revert once that PR merges to main.
|
Thank you @algomaster99! |
3 participants
Assignment Proposal
Title
FlakyGuard: Automatically Fixing Flaky Tests at Industry Scale (ASE 2025)
Names and KTH ID
Deadline
Category
Description
We want to present the ASE 2025 paper "FlakyGuard: Automatically Fixing Flaky Tests at Industry Scale" by Li, Behrang, Shi, and Liu. The authors propose a tool called FlakyGuard that automatically repairs flaky tests, and they evaluate it in production at Uber. We plan to focus on how FlakyGuard traverses a call graph to collect just enough context to resolve flakiness, discuss the reported success rate together with how many fixes actually land in production, and contrast the approach with previous work in this field.
Relevance
Flaky tests break the assumption that a failing build means the code is broken. When working with flaky tests, developers prefer to rerun jobs rather than investigate the failure, and over time the test suite loses its value. Therefore, fixing them automatically is directly relevant to the subject of test automation and CI.