- First commit to issue fork.
- Merge request !11546Issue #2821507: Handle exceptions gracefully when saving node forms → (Open) created by acbramley
- 🇦🇺Australia acbramley
I'm not sure, the way I read it was that the current solution is fine for this. I haven't looked for another issue but that doesn't need to block this.
- 🇨🇭Switzerland berdir Switzerland
I think it we should do that, alexpott also agreed, the wrapped exception is almost useless.
- 🇦🇺Australia acbramley
I think it we should do that
Are you able to expand on exactly what we should do?
- 🇦🇺Australia acbramley
So maybe we could just add $e = $e->getPrevious() ?: $e, at least until we have a better default handling of the backtrace including previous exception(s).
From #72
I'm guessing this is it? I've pushed that. Having issues getting gitlab to rerun the failed tests (assuming a random) but this should be ready for a re-review.
- 🇦🇺Australia acbramley
I tested the exception backtrace with and without the getPrevious() and IMO (for the issue described in the IS at least with missing a title) I find the backtrace without the getPrevious easier to parse.
- 🇬🇧United Kingdom jonathanshaw Stroud, UK
I can see the argument that the previous is more precise. I don't think it matters either way.
- 🇨🇭Switzerland berdir Switzerland
> I find the backtrace without the getPrevious easier to parse.
Yes, but it also has less information. It's kind of minor for the title case, but try it for when for example a hook_node_update() implementation that throws an exception (which is what the test is doing) because it tries to access a field that doesn't exist. Having that information in the backtrace is a lot easier to finding where it's coming from.
