Should matches use perl = TRUE?

[hadley]

I doubt this will affect any existing code, but does make the pattern syntax a bit closer to stringr.

[ant-durrant]

I just came here to check whether \p{} not being supported was possibly a bug and saw that perl = FALSE is the culprit.

Put that as at least one vote for yes!

From the docs:

For matches() this is a regular expression, and can be a stringr pattern.

I am aware that using grep() sub() I have to think about a perl flag, but I am not aware of ever thinking about it when using tidyverse stuff.

Just checked, and yep - at least stringr::str_detect() does not have perl as an argument, so if it is to be included, having the default as TRUE makes sense for me.