CLI docs have incorrect default values for options using store_false action

[victorlin]

Scope

Any CLI option that uses the store_false action, which are these options on version 25.2.0:

• augur curate format-dates --no-mask-failure
• augur filter --no-probabilistic-sampling
• augur refine --no-covariance
• augur refine --greedy-resolve
• augur mask --no-cleanup

Issue

The CLI docs show a default value of True for these options. I noticed this because it was confusing on the CLI docs for augur filter, where --probabilistic-sampling and --no-probabilistic-sampling are supposed to be mutually exclusive:

augur filter --help text does not contradict, but there is no text provided for --no-probabilistic-sampling which still isn't great:

  --probabilistic-sampling
                        Allow probabilistic sampling during subsampling. This is useful when there are more groups than requested sequences. This option only applies when `--subsample-max-sequences` is
                        provided. (default: True)
  --no-probabilistic-sampling

and then I realized that the lack of help text may be intentional to avoid the same issue:

augur/augur/refine.py

Line 125 in 988380c

parser.add_argument('--no-covariance', dest='covariance', action='store_false') #If you set help here, it displays 'default: True' - which is confusing!

Possible solutions

1. Describe defaults in help text and avoid printing any defaults according to argparse. I think this is what Nextstrain CLI does for its arguments that use store_false such as nextstrain build --no-download.
2. Use store_true and flip the boolean internally. Something similar was done in Clarify default values for inference of ambiguous bases #613

[corneliusroemer]

I'm pretty sure I remember noticing this in the past - but I must have forgotten to make an issue. Thanks!

[joverlee521]

Ah, I thought I fixed this issue for augur curate format-dates --no-mask-failure with 4d02220.

Looks okay in the CLI

augur curate format-dates -h
...
  --no-mask-failure     Do not mask dates with 'XXXX-XX-XX' and return original date string if date formatting failed. (default:
                        False)

But I'm seeing that this can be confusing in the docs