nealrichardson commented on a change in pull request #11000:
URL: https://github.com/apache/arrow/pull/11000#discussion_r695959406
##########
File path: r/tools/nixlibs.R
##########
@@ -215,10 +215,8 @@ download_source <- function() {
# Given VERSION as x.y.z.p
p <- package_version(VERSION)[1, 4]
- if (is.na(p) || p < 1000) {
- # This is either just x.y.z or it has a small (R-only) patch version
- # Download from the official Apache release, dropping the p
- VERSION <- as.character(package_version(VERSION)[1, -4])
+ if (is.na(p)) {
+ # This is just x.y.z so download the official Apache release
Review comment:
The logic I was going for is:
```
if version == x.y.z:
download from apache
elif version == x.y.z.p and p != 9000:
download from s3 ("nightly")
else
don't download because this is our dev version sentinel (x.y.z.9000)
```
5.0.0.1 (and now I guess 5.0.0.2) has C++ source on our "nightly" S3 bucket.
What this changes is removes the logic of allowing 5.0.0.1 to use the 5.0.0
apache release, as in when we have to do R only things to massage a CRAN
release through. It would mean that if we had to do a 6.0.0.1 release because
of a new whitespace requirement from CRAN, we would have to also produce a
/libarrow/src/arrow-6.0.0.1.zip and host on the nightly repo. All of this is
moot if I/we start bundling the C++ source in the R package, as discussed on
that other jira issue.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
