Make style = TRUE the default?

[krlmlr]

This might make the life of package maintainers a tad easier. Two things to remember:

• styler throws an unhelpful error message if the code can't be parsed, we might want to do better to help users.
• Only scope = "line_breaks" guarantees that the input is identical to the output as far as the parser is concerned. This will keep = for assignment and won't add () after function calls in pipes but is the safer alternative.

CC @lorenzwalthert.

[lorenzwalthert]

There's been a debate about whether to use style = TRUE as a default in #108. Personally, I think it would make sense, but there is also an easy way to make reprex calls behaving as if style = TRUE was the default, which I use: Just use the global option to change the default of the argument:

options(reprex.style = TRUE)

The thing is probably that most users don't do that or know how to do it.

[batpigandme]

FWIW, I think (depending on how different an individual's coding style is from styler) this could be kind of confusing to the unsuspecting user.

I'm thinking more about Q&A than GitHub issues, I put my code in, and now it looks all different!

[krlmlr]

Ah, that's what the opt() does ;-)

reprex could talk about it, like:

Code formatted by styler. To turn off, set options(reprex.style) = FALSE.

Keeping this open, because the original discussion is now a year old and styler has matured a lot since then.

[jennybc]

I'm certainly open to this but won't do anything for the small patch release I want to make in the very near future. It would be nice if reprex offered more support, via an add-in or helper function, for people who want to customize package-level options. We could be more bold with defaults if we had that.

[jennybc]

Maybe style should default to NA, with default logic of: set style = TRUE if styler is installed.

This would need to be implemented and is also connected to centralized documentation for options: #259

[lorenzwalthert]

I know this is for interactive use primarily but isn't it generally considered bad practice to have code behaving differently depending on what other packages are installed? Do you plan to have similar behavior for other options?

[jennybc]

Well, there is precedent for this sort of thing for a worklow / interactive package. For example, httr and everything which depends on it uses a local web server for auth (vs out-of-bound auth) depending on whether httpuv is installed.

I don't think it's a big deal either way.

But we can also leave style = FALSE as the default.