You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
server: avoid learning the same job multiple times
The current learning implementation was designed with Travis CI in mind,
where a single check was sent when the job ended. Because of that, when
a successful build was detected all the successful job outputs were
learned. This assumption doesn't hold anymore for GitHub Actions though.
GHA sends a check for every job in the matrix, and that causes RLA to
learn a successful job every time one of the following jobs in the same
build finishes, learning it multiple times. For example, if a build
contains 60 jobs, the first job to finish will be learned 60 times.
This fixes the problem by remembering which job IDs we learned before,
and skipping the learning process if the ID was previously seen. Along
with that the code to remember IDs was refactored.
let job = match build.jobs().iter().find(|j| j.outcome().is_failed()){
186
168
Some(job) => *job,
@@ -301,6 +283,9 @@ impl Worker {
301
283
[I'm a bot](https://github.com/rust-lang/rust-log-analyzer)! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact **`@rust-lang/infra`**. ([Feature Requests](https://github.com/rust-lang/rust-log-analyzer/issues?q=is%3Aopen+is%3Aissue+label%3Afeature-request))
0 commit comments