Open
Description
Currently, none of the CI docker builders have backtraces enabled. This can make debugging a failed PR very time consuming - since the panic message itself is often useless, it can be necessary to run the entire CI script locally to generate a proper backtrace.
If all of the CI images had backtraces enabled, it would be possible to get a backtrace directly from the logs, saving PR authors a significant amount of time.
I wasn't sure the best way to go about this - I didn't want to duplicate RUST_BACKTRACE
across ever single Dockerfile
, but there doesn't seem to be any infrastructure in place for sharing common thigns between the various Dockerfile
s.
This came up in #60026 (comment)