Incorrect PrintHelpMessage return code

[pgreze]

Facing a similar issue than #41 I tried to display the help message + return an error code, as it's specified in PrintHelpMessage javadoc:

clikt/clikt/src/commonMain/kotlin/com/github/ajalt/clikt/core/exceptions.kt

Line 25 in 8078c48

* Execution should be immediately halted with an error.

Sadly the current behavior is returning the return code 0:

clikt/clikt/src/commonMain/kotlin/com/github/ajalt/clikt/core/CliktCommand.kt

Line 278 in b9c1891

exitProcessMpp(0)

Considering help message is the expected output when using -h or --help which is already implemented by clikt, user usage of PrintHelpMessage would probably always means that help message is the consequence of an error.
If you agree with previous sentence, I would encourage enforcing javadoc behavior by fixing the code but still provide a way to show help message without failing (like Abort logic?).

In any case, thanks for your this nice project, always useful when trying to create Kotlin CLI 👍

[ajalt]

$ for i in git man ls cat bash grep awk cloc rg; do $i --help >/dev/null; echo $i : $?; done
git : 0
man : 0
ls : 0
cat : 0
bash : 0
grep : 0
awk : 0
cloc : 0
rg : 0

Every command line tool I looked at exits with status code 0 when they print their help message, so I think the current behavior is correct, and this is a bug in the documentation.

[pgreze]

@ajalt I agree, but you always tried with "--help", which of course should return 0.

But if you're trying:

$ git
usage: git [--version] [--help] [-C <path>] [-c <name>=<value>]
           [--exec-path[=<path>]] [--html-path] [--man-path] [--info-path]
           [-p | --paginate | -P | --no-pager] [--no-replace-objects] [--bare]
           [--git-dir=<path>] [--work-tree=<path>] [--namespace=<name>]
           <command> [<args>]
...
$ echo $?
1

Even if output is similar to --help, this is still an error. That's what I was talking about, and that's why it's not obvious for me that show usage output should always finish without error.

[ajalt]

I see. That's different than the --help case mentioned in the initial issue. I agree it makes sense to use an error code when the printHelpOnEmptyArgs or invokeWithoutSubcommand cases are triggered.

I'll add an error : Boolean parameter to PrintHelpMessage and PrintMessage and use it for those cases. The --help case will continue to exit with 0.