If you run or write code as part of your research, sooner or later something isn’t going to work as expected.

Debugging checklist

• Compare your code against the last working version (git diff). If it previously worked, then something you have changed will most likely have brought about the current error.
• Google the error message. It’s quite possible that someone else has encountered the same problem as you, and had it answered on stack overflow.
• Are you running the latest version of the software? You might have found a bug that has subsequently been fixed. Updating your software might resolve the problem.
• Check on the software’s issue tracker. Many software packages have a publicly accessible issue tracker (on GitHub or elsewhere) that you can search, or use to create a new issue for the problem you’re facing.

How to ask for help

When asking for help (e.g. from colleagues, on stack overflow), you should aim to make it easy for others to help you. This means

• Give some context - what were you trying to do? This is more useful than just asking what an error message means.
• Given them something they can run to reproduce your error. A minimum working example is the target here. A link to your code on GitHub would be great - maybe you want to create an issue too which explains everything so the conversation can stay with the code rather than getting lost on slack.
• What versions of which software are you using?
• Show the whole error message - the important clue might not be where you think it is
• What platform are you running on (laptop/CSF etc)?

Very often the process of verbally explaining / writing down the problem causes you to fix the problem on your own. This is known as rubber ducky debugging.