GeezerVison

Looking inside _parse_known_args, I see some error paths doing raise ArgumentError and others calling self.error, without any apparent pattern. If they all did raise ArgumentError, that would fix this bug.

There are other calls to self.error elsewhere in the class that may be invoked from _parse_known_args, so this still leaves some paths where it may exit. In that sense, your original solution was more robust. I am honestly not sure which solution is better.

I found a duplicate with some more discussion: bpo-41255.

From the penultimate comment:

I'm not sure I like the idea of changing the exit or error methods since they have a a clear purpose and don't need to be repurposed to also include error handling. It seems to me that adding checks to self.exit_on_error in _parse_known_args to handle the missing required arguments and in parse_args to handle unknown arguments is probably a quick and clean solution.

Alternative to #27295

Backref to #85427 (@jacobtylerwalls please could you update the title to start with gh-85427?).

A

@JelleZijlstra Forgive the ping! I just I noticed a duplicate issue (#103498) and PR (#103519) rolled in for this. Do you have a reviewer you might recommend to take a look?

Where we left off was comparing 1.) guarding the calls to self.error() or 2.) altering self.error() itself.

I am honestly not sure which solution is better.

I was somewhat convinced by the comment quoted in #30832 (comment). Plus it's more backwards compatible to leave error() alone. Appreciate any advice, as you may be able!

@sobolevn and @barneygale expressed some interest in argparse issues recently. I don't feel strongly here but I'll try to find some time to read up again and form an opinion.

Also closes: #103498

So, we have at least two PRs that do opposite things.
This PR uses raise ArgumentError: #30832
And the alternative uses self.error() instead: #103519

Now, we need to decide what is better :)

Thank you for your contribution @jacobtylerwalls. I apologize for the fact that this PR was neglected for a long time. The problem has already been solved in a way similar to the way proposed in your PR.