Reconsider use of internal ref() function

[jemus42]

In some cases like `r ref("resample()")`, the resulting output is functionally identical to the output that downlit already generates when `resample()` is encountered in the text.
I'm wondering if the overhead is worth it, given that we're working around/on top of downlit for questionable benefit.

The ref() helper provides more functionality, yes, but in the simplest case I'd argue it's not worth it.

[mllg]

I haven't checked what's happening with the current version of quarto/downlit, but in the past the inline code chunks (using a single backtick) were not processed/linked. Did that change?

[jemus42]

As far as I can tell, yes. Tried it in a local make serve and the main difference is that downlit'd links aren't blue and underlined, but they're still links.

Not sure when you checked the last time, but quarto integrates downlit but RMarkdown/bookdown didn't (out of the box). Maybe the appropriate option wasn't set before?

[mllg]

AFAICT we would have two different styles of links and R6 objects are not linked (r-lib/downlit#140). Maybe we just stick to ref() and wait for downlit to improve?

[jemus42]

Ah, fair point. I was hoping we could spare ourselves some homebrewery, but I agree.

[pat-s]

I tried that during the move to quarto already and I think we should still with our own ref(). Thanks for double-checking though!