Remove panics from xml writing #1075
Conversation
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| @@ -334,7 +351,7 @@ mod tests { | |||
| #[tokio::test] | |||
| async fn test_no_history() { | |||
| let (tx, rx) = async_channel::unbounded(); | |||
| let mut child_rx = pin!(HydratedInfo::build_transformer(rx)); | |||
| let mut child_rx = std::pin::pin!(HydratedInfo::build_transformer(rx)); | |||
There was a problem hiding this comment.
use std::pin::pin gives an unused warning... but if I remove it, then it won't compile.
| cfg.description.clone(), | ||
| )), | ||
| build_event_stream::build_event::Payload::Aborted(cfg) => Some( | ||
| match build_event_stream::aborted::AbortReason::try_from(cfg.reason) { |
There was a problem hiding this comment.
addresses a warning
| } | ||
| } | ||
| Err(e) => | ||
| println!("could not access metadata for test result {} at file {}.\nError {}", |
There was a problem hiding this comment.
Does anything in here bubble back up? Should it?
There was a problem hiding this comment.
So this error means that when you went to access the size of the file, the filesystem operation failed.
It's for an intermediate test result that is reported by the Bazel event protocol log to exist.
We could halt the program here. I doubt this will happen. The cases we have seen look more like we can't write to the disk (maybe it's full, maybe we have some permission issue). Although if it is a permission issue, maybe we can't read.
I would say leave it as is now and we'll see how these spurious errors chance.
What do you think?
occasionally we see panics when trying to generate the junit xml. The junit code has a number of
expectcalls.This pr changes the code to pass the
Resultalong, aggregate all the xml writing failures, and list those at the very end.It should only improve matters.