lib/helpers: fix _command_exists()

[gaelicWizard]

This is a tiny PR since I'm having trouble rebasing my helpers PR (#1934). This change was in the middle and just much easier to pull it aside 😃

Description

The weird subshell is weird AF. Just do a normal if. Ditto _binary_exists(), _completion_exists(), and _is_function()! Alsö, make all_groups() literally 100x faster.

Motivation and Context

Simplicity, consistency, and #1696! And git rebase is hard sometimes.

How Has This Been Tested?

Locale and all tests pass.

Types of changes

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• My code follows the code style of this project.
• If my change requires a change to the documentation, I have updated the documentation accordingly.
• I have read the CONTRIBUTING document.
• If I have added a new file, I also added it to clean_files.txt and formatted it using lint_clean_files.sh.
• I have added tests to cover my changes, and all the new and existing tests pass.

[gaelicWizard]

@NoahGorny, this one is tiny tiny if you could review & approve ♥

[davidpfarrell]

Greetings! Got some feedback on this one, but generally happy to see it get some attention ...

[gaelicWizard]

This is ready for merge unless someone wants to OK removing the default message 👍

[davidpfarrell]

I'd rather remove the default message

Wondering what that would look like? i.e would we lose $2 altogether as well?

As these checks are often the core reason that components bail early and skip activation, I feel like its probably important for some form of logging to show up in the bash-it doctor logs.

I also feel like, in general, we are not passing $2 along when we invoke these (even in newly added code), nor are we adding our own logging statements.

So both removing the default message OR removing logging altogether feel like missed opportunities to give the user useful bash-it doctor messages.

[gaelicWizard]

the core reason that components bail early and skip activation

It seems like only a few plugins/components actually do this, and it feels like logging about why bailing early should be in the plugin. If bash-it doctor prints Command 'git' does not exist! without mentioning plugin/git, it isn't super helpful for a user who isn't already very familiar with Bash and/or Bash It. The log message should be much clearer, like plugin/git: skipping plugin activation because 'git' cannot be found or similar. It would need to be different for every component, and sometimes only part of a component is skipped, e.g. the alias for t, the alias for fuck, &c.

I definitely agree there should be some message when a plugin/completion is skipped, but it should be logged from the component script not the _command_exists function. I can do a PR to update everywhere that this is needed if you agree that's the way to go.

[gaelicWizard]

@davidpfarrell, here's an example I just made up for plugin/less-pretty-cat:

if ! _command_exists pygmentize; then
	_log_warning "${BASH_SOURCE[0]##*/}: unable to initialize without '$_' installed."
	return 1
fi

Since we're bailing early, we can't just return from a function; it needs to be in the file itself.

Compare to plugin/zoxide:

if _command_exists zoxide; then
	eval "$(zoxide init bash)"
else
	_log_error 'zoxide not found, please install it from https://github.com/ajeetdsouza/zoxide'
fi

EDIT: ok, well, it seems I've been nerd-sniped in to this being my next project. I do need at least one specific bit of assistance: should this be _log_warning or _log_error? My intuition is that ń̥o҉̙̻̖̲̮͓̩̀t̵̬̳͔ḩ̡̪͇͈i̼̦̮̙͝n̸̸̲̲͍̟g͏̭͖̠̩̭ is broken. It's just not there. An error would be if we didn't bail early and then tried to use the missing app and things went sideways, no?

[gaelicWizard]

Ok, I've got a branch I'm working on for early-bail from plugins that require other commands but it's way to big for this PR simply due to the number of files. I think this branch should be good to merge, and then I'll open a new PR when I've got all the plugins ready and we can revisit how the logging should flow then. 👍

[NoahGorny]

@davidpfarrell, here's an example I just made up for plugin/less-pretty-cat:

if ! _command_exists pygmentize; then
	_log_warning "${BASH_SOURCE[0]##*/}: unable to initialize without '$_' installed."
	return 1
fi

Since we're bailing early, we can't just return from a function; it needs to be in the file itself.

Compare to plugin/zoxide:

if _command_exists zoxide; then
	eval "$(zoxide init bash)"
else
	_log_error 'zoxide not found, please install it from https://github.com/ajeetdsouza/zoxide'
fi

EDIT: ok, well, it seems I've been nerd-sniped in to this being my next project. I do need at least one specific bit of assistance: should this be _log_warning or _log_error? My intuition is that ń̥o҉̙̻̖̲̮͓̩̀t̵̬̳͔ḩ̡̪͇͈i̼̦̮̙͝n̸̸̲̲͍̟g͏̭͖̠̩̭ is broken. It's just not there. An error would be if we didn't bail early and then tried to use the missing app and things went sideways, no?

I would say its a warning and not an error- error is something truly not working properly in Bash-it IMO

[davidpfarrell]

Approved !