Expected vs. Actual output for failed tests unreadable due to %0A instead of line breaks

[education]

Expected vs. Actual output for failed tests unreadable due to %0A instead of line breaks

[julianvu]

I think this is a larger issue of the autograding action's not being able to check multi-line outputs with regexes. My regex tests are failing for multi-line outputs as well.

[markpatterson27]

I think this is because autograding is treating a failed match as an error.

Error messages are escaped in core. \n is replaced with %0A. I assume stdout from processes is also escaped and error messages are escaped seperately because it's assummed the process that triggered the error has exited. When the logs are displayed, the escaped charaters are substituted back.

Except, autograding hasn't exited. It's still running through the rest of the tests. So the error message gets sent to the log, where it is escaped again. Since '%' is an escaped charater, when the log is displayed, the workflow interface can't substitute %0A back to \n because it thinks it's substituting %250A back to %0A.

TL:DR; failed match messages are being escaped twice. The solution is to not throw an error when a match fails (because failed matches are an expected outcome ;).

[jwkim90]

@daveavis Hello, I'm wondering if you resolved this issue because I'm having the same problem.

[GittyBitch]

Indeed, this is still a problem