Set default UTF-8 locale

[glin]

The Ubuntu images currently set a default UTF-8 locale (LANG=C.UTF-8), but the CentOS and openSUSE images don't (defaults to LANG=C). This sets a default UTF-8 locale in each base image and adds an R test to confirm that.

Note that the default locales that come with Ubuntu, openSUSE, and CentOS images are all different. Ubuntu and openSUSE come with C.UTF-8, while CentOS only has en_US.UTF-8 as the closest thing.

I also slightly changed how the locale was being set on Ubuntu. It was being set through the Renviron.site file before, which makes it hard to override. For example, if you wanted to change the locale in a Docker image:

FROM rstudio/r-base:3.5-xenial

ENV LANG pt_BR.UTF-8

The locale wouldn't change in R since the Renviron.site file takes precedence. I've set it through an environment variable in the base image instead, which is consistent with what other Docker images do.

References

• Locale discussion in the rocker images: Default locale is C rocker-org/rocker#19
• How locales are set in some of the official Docker images:
  • https://github.com/docker-library/python/blob/4e284ff8ad5458518743188afa3e64d83ea9986c/3.8-rc/stretch/Dockerfile#L14
  • https://github.com/docker-library/docs/blob/3e578ee6c54631939dd66570c1415ff0e26dc1a0/ubuntu/content.md#locales
• C.UTF-8 vs en_US.UTF-8: https://community.hpe.com/t5/General/Difference-between-C-utf8-and-en-us-utf8-points/td-p/4418194
• CentOS doesn't have a C.UTF-8 locale by default: https://bugzilla.redhat.com/show_bug.cgi?id=902094

[jonyoder]

👍 Thanks for doing the research to make this happen.