What does Positron's 'code:5' crash mean?

[marcdotson]

I've really appreciated the fact that the kernel crashes separately from Positron, but I finally broke it. Twice now I've been running a model only to have the Positron equivalent of the RStudio bomb appear:

What does 'code:5' mean? I can't do much to diagnose the problem. A PyMC model was running overnight both times this has happened. Any insight would be appreciated!

[marcdotson]

To add some more context:

• To avoid losing the model run, I ran the model and the code to save output together
• The Positron crash happened before the model appeared to finish sampling
• And yet the saved model object was written almost 1.5 hours after the crash

So I'm even more confused than I was previously. It looks like the kernel didn't crash even though Positron did? Which is cool and consistent with the two being separate -- but what is really happening?

[wesm]

This seems to be a type of crash that has been reported upstream in VSCode: microsoft/vscode#143101. There are some instructions about producing a crash dump file that could help with debugging the problem here: https://github.com/microsoft/vscode/wiki/Native-Crash-Issues#creating-a-crash-report. If you're able to produce a crash, you can open an issue and we can have you e-mail it to us to see if we can diagnose the problem.

[marcdotson]

Thanks, @wesm. I've replicated the code:5 crash and produced a crash dump file. Where should I email it?

[marcdotson]

I opened an issue as requested, #8368.