Github user aarondav commented on the pull request:
https://github.com/apache/spark/pull/7028#issuecomment-120777385
@squito, I have updated the PR and simplified it greatly, thus removing
most of the magic. Now we simply append the current stack trace inside
DAGScheduler, with no special marker. Only someone looking very closely at the
trace would observe something amiss, and they would see the comment pointing to
the associated JIRA.
I think we should go with this approach over the wrapping approach for the
simple reason that it is less likely to break callers. Some callers may fail to
print the cause of the exception, which would have been fine before with no
cause but now would not work. Some callers may inspect the message directly,
which would be different (or we would have to duplicate). Some callers may
display the exception to users, which would look significantly different if
wrapped (rather than just made longer).
Please let me know if you have further concerns; I have run into this
problem sufficiently often now to want to push for this to go in (and the
workaround is really annoying -- adding prints or try-catches to every line
which may have caused the issue!).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
