Conversation
|
Thanks for making this! Were we actually able to figure out a minimal reproduction where the stacktrace became like this? it sounds like maybe we weren’t — I’m not super-confident that it’ll fix it as a result, and I’m wondering what we’ve already tried. Ideally, there’s a test that would have failed before this change and now passes! I see you have an interesting hypothesis in your notes:
Can we somehow simulate that by creating a test queue and a job, and then running it in that? (We do have working stacktraces from jobs, so I’m a bit skeptical of this hypothesis too, and we may want to message/list other hypotheses first, and then go through them systematically until you have a minimal reproduction) Also, a reminder to link to the Notion task from the GitHub PR/MR! It’s here: Make sure stacktrace shows up for query on MapsPlaceReviewLocalized (likely due to findAllChunked) |
Motivation
(Added: Notion task: Make sure stacktrace shows up for query on MapsPlaceReviewLocalized (likely due to findAllChunked))
When looking through our logs, one of our stacktrace comments came back like this: "/* stacktrace=\n"
It's likely that we accidentally truncated the stacktrace at -1, which included just the final newline, when trying to find a good truncation point for the stacktrace. Instead, we should:
Changes
makeMinimalUsefulStacktraceto take in a string to make it easier to testTesting
Added new automated test for a string without "node_modules" nor "at" in it. Ran
yarn mocha test/util.test.js