dandi download enums

[CodyCBakerPhD]

While reviewing some PRs I decided to download a full dataset to test on

I used the syntax

dandi download DANDI:001350

but recieved

Usage: dandi download [OPTIONS] [URL]...
Try 'dandi download --help' for help.

Error: Invalid value for '-e' / '--existing': 'error' is not one of 'ERROR', 'SKIP', 'OVERWRITE', 'OVERWRITE_DIFFERENT', 'REFRESH'.

and so on for various flags. To get it to work as expected I had to do instead

dandi download DANDI:001350 -e ERROR -f PYOUT --path-type EXACT

which worked as I expected

Is this behavior intended? Can others reproduce it? If not assign me and I believe I know where and how it goes wrong

[yarikoptic]

that is odd, are you sure you do not have some alias for dandi somewhere in your shell config? here is what I see

❯ dandi --version
0.68.0+2.ga7743558

[CodyCBakerPhD]

dandi --version

> 0.68.0

In a fresh Conda environment (dandi was an unrecognized entrypoint prior to fresh install from PyPI)

(Using Windows 11)

[stephprince]

I have not looked into it in detail yet, but this error also popped up in the NWBInspector daily workflow logs

[CodyCBakerPhD]

Thanks for convincing me that I have not (yet) gone completely insane @stephprince

The logs are very helpful in that they indicate the issue is on every Python version above 3.9

I had forgotten to indicate that I was using Python 3.13 before

I just tried it again with 3.9 and verified the issue does not occur there

The new mystery is why this suddenly changed a mere 2 days ago when no DANDI related package has cut a new release in that span...

[jwodder]

Hint: pallets/click#2911

[CodyCBakerPhD]

Thanks for figuring it out @jwodder

I noticed the DANDI CLI testing suite has actually been catching this: https://github.com/dandi/dandi-cli/actions/runs/14964999717/job/42033240845#step:8:1034

But it was very hard to notice amongst all the obo lib failures

[yarikoptic]

But it was very hard to notice amongst all the obo lib failures

yeah... we should mask those out somehow. I also started to try "CI insights" which should provide summary across runs but yet to get back to it now for Mergifyio/pytest-mergify#118 and see if it would have helped us to observe it more rapidly

[yarikoptic]

For now just let's forbid that version of click in #1628, hopefully next one addresses this regression. @CodyCBakerPhD please confirm that going back to 8.1.8 resolves it for you.

[CodyCBakerPhD]

Working now 👍