jonkeane commented on a change in pull request #9896:
URL: https://github.com/apache/arrow/pull/9896#discussion_r607335099
##########
File path: r/R/arrow-package.R
##########
@@ -47,12 +47,27 @@
}
# Create these once, at package build time
- dplyr_functions$dataset <- build_function_list(build_dataset_expression)
- dplyr_functions$array <- build_function_list(build_array_expression)
-
+ if (arrow_available()) {
+ dplyr_functions$dataset <- build_function_list(build_dataset_expression)
+ dplyr_functions$array <- build_function_list(build_array_expression)
+ }
invisible()
}
+.onAttach <- function(libname, pkgname) {
+ if (!arrow_available()) {
+ msg <- paste(
+ 'The Arrow C++ library is not available. To retry installation with
debug output, run:',
+
+ ' install_arrow(verbose = TRUE)',
+
+ 'See `vignette("install", package = "arrow")` for more guidance and
troubleshooting.',
Review comment:
I think this is a good general message. A few thoughts:
* Should we refer to the arrow c++ as a dependency? I think that's a word
that people with less experience might look for.
* Should we mention that in typical installations the arrow c++ library is
either bundled with the installation (e.g. our binaries/CRAN) or built as part
of the package installation process. I can't think of a short and pithy way to
say this, but what I think we _don't_ want is someone to think that this means
they need to now go and `apt install arrow` or the like (that should work for
releases as long as the versions match up, but there are lots of ways that
could be wrong).
* Is there any prohibition on offering the url
https://arrow.apache.org/docs/r/articles/install.html it's the same content as
the vignette, but I suspect more people are familiar with packagdown sites than
they are with vignettes these days.
This is separate from this message, but I wonder if we should have another
message onload (ideally onload, only the first time like many of the tidyverse
packages do) that tests for `codec_is_available("snappy")` (or all of the ones
not in the minimal build) that gently nudges that in the case of linux you'll
want to set libarrow_binary=true or libarrow_minimal=false to get a more fully
featured build? I know that a few times when I was managing rstudio server on
linux I installed arrow, thought everything was fine and then later someone had
to bug me that snappy wasn't enabled anymore so I had to go back and install
it. It's not a huge deal but getting it out of the way when I'm in the
installing mindset would have been nice compared to finding out later and
having to go back to it.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
