R 3.4.4 from the Ubuntu repositories:
> "µV"
[1] "\302\265V"
Same computer, R 3.4.4 (and also 3.2.0, and also 3.5.1) compiled from the sources obtained from CRAN:
> "µV"
[1] "µV"
I'd rather prefer the second behaviour. Where does the difference come from?
Encoding("µV")
returns "unknown"
in the first case and "UTF-8"
in the second case, but setting the encoding of a string variable manually doesn't seem to improve its representation.
For some reason, Sys.getencoding()
was different in those two builds. Running Sys.setlocale("LC_COLLATE", "en_US.UTF-8")
on the first build seems to have fixed the issue.