Formatter doesn't seem to have appropriate exit codes

[anicolao]

Environment information

CLI:
  Version:              11.0.0
  Color support:        true

Platform:
  CPU Architecture:     aarch64
  OS:                   macos

Environment:
  ROME_LOG_DIR:         unset
  NO_COLOR:             unset
  TERM:                 "xterm-256color"

Rome Configuration:
  Status:               loaded
  Formatter disabled:   false
  Linter disabled:      false

Workspace:
  Open Documents:       0

Discovering running Rome servers...

What happened?

1. npx rome format source complains about some source files
2. echo $? outputs 0

Expected result

I expect the exit code to indicate a problem so that scripting can react to an incorrectly formatted project.

Code of Conduct

• I agree to follow Rome's Code of Conduct

[ematipico]

That's intended. If you wish to take advantage of exit codes, rome ci is the command you want.

We might re evaluate exit codes in the future

[anicolao]

Is there a design doc you can point me to that can help me understand why I would want to use a command line tool that is like no other I've used in 40 years?

[ematipico]

Some issues you might want to follow up on:

• 🐛 rome format with errors exits with code 0 #3443
• rome format . does not log to stderr #3168

We plan to change the behaviour of the command. The current behaviour is intended and not a bug.

[anicolao]

Ah I am sorry I didn't see issue #3443 before, I should have found it before logging a duplicate. Very sloppy of me.

FWIW I think @JaySal's comment there is correct, that rome format should act as a pipeline command and print formatted output to stdout by default, and the verbose error message-style output belongs in some other command. But I don't care too much about this as rome format --write is more convenient than dealing with classic pipelines anyway, and not so far off the beaten path as to raise eyebrows.

As far as I'm concerned rome ci does too much and has no point in existing, when I could instead run rome lint followed by rome format or the other way around. As more tools get added to rome it is more and more likely that the users will want a subset of them, and not a single "does it all" ci command that then needs phases disabled with flags. Just my opinion.