VSCode integration with --header-opt-out

[matthewford]

Hi, I would like to use the opt-out argument to aid the adoption on a complex project.

However, it looks like the VSCode extension does not support additional arguments. Would it be possible to make --header-opt-out a default setting?

[reese]

I think I'd be okay with making the header-opt-out functionality a default (other code formatters also have something like this by default, e.g. // prettier-ignore), as long as we still had the reverse (like --no-header-opt-out or whatever). PRs would be welcome here!

[fables-tales]

@reese let's add a ~/.rubyfmtrc and $(pwd)/.rubyfmtrc that passes flags in? I think the VS code extension runs rubyfmt from $(pwd) so that should work just fine?

we should not pass the opt out by default

[fables-tales]

oh and XDG_CONFIG_HOME

[reese]

I think an rc file would be fine, I guess my main concern there is just adding more complexity (e.g. is there a world where we need to support ENV variables when reading the rc file, we probably now need to handle how precedence works if the system file and pwd file conflict, etc.)

[fables-tales]

I buy that, maybe the right thing to do is instead have folks build a wrapper shell script that passes the flags they want?

[fables-tales]

i.e. your/project/rubyfmt

#!/bin/bash
rubyfmt --flags --you --care --about $*

[reese]

Yeah, the wrapper script is the way we did it (and still do it) for the Stripe rollout. I don't think it's a perfect end state, but I think it's good enough for most cases at the moment, especially since we have pretty few flags at the moment.

[fables-tales]

@matthewford does the suggestion of using a wrapper script meet your needs here?

[reese]

Just as another option for this particular case, I don't know if we ever documented it (although we should), but there's some support for a .rubyfmtignore file where you could add globs/directories you haven't rolled out to yet, that could be a good option if you're using this for a rollout.

[fables-tales]

@reese oh, let's document that

[matthewford]

A wrapper works, rubyfmtignore would be really useful too