[
https://issues.apache.org/jira/browse/ARROW-17662?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Pierre Gramme updated ARROW-17662:
----------------------------------
Description:
*Use case:* I'm trying to install arrow's R package on a Linux (RHEL) server
without internet access. The server only has access to CRAN and a few other
repo's via a mirror on intranet (Artifactory instance). In particular, no
access to public RStudio Package Manager or to apache.jfrog.io or github.
Currently (v9.0.0), installation script does not really foresee the possibility
to install from binaries in such a case. By default it will build from source
without any external dependencies, which is slow and less feature-rich.
*Suggestion:* if the user has the possibility to manually download the right
zipfile for libarrow, he can set the (new) environment variable
`ARROW_DOWNLOADED_BINARIES` to the local path, and the installation script will
use that pre-downloaded file.
In addition, in case of failed download, the URL will be printed. With this
extra info, it will be easier for the user to identify the right URL for
libarrow zipfile and then download it manually.
I just proposed a PR...
was:
*Use case:* I'm trying to install arrow's R package on a Linux (RHEL) server
without internet access. The server only has access to CRAN and a few other
repo's via a mirror on intranet (Artifactory instance). In particular, no
access to public RStudio Package Manager or to apache.jfrog.io or github.
Currently (v9.0.0), installation script does not really foresee the possibility
to install from binaries in such a case. By default it will build from source
without any external dependencies, which is slow and less feature-rich.
*Suggestion:* if the user has the possibility to manually download the right
zipfile for libarrow, he can set the (new) environment variable
`ARROW_DOWNLOADED_BINARIES` to the local path, and the installation script will
use that pre-downloaded file.
In addition, in case of failed download, the URL will be printed. With this
extra info, it will be easier for the user to identify the right URL for
libarrow zipfile and then download it manually.
I will soon propose a PR...
> [R] Facilitate offline installation from binaries
> -------------------------------------------------
>
> Key: ARROW-17662
> URL: https://issues.apache.org/jira/browse/ARROW-17662
> Project: Apache Arrow
> Issue Type: Improvement
> Environment: linux
> Reporter: Pierre Gramme
> Priority: Major
> Labels: R, install, pull-request-available
> Time Spent: 10m
> Remaining Estimate: 0h
>
> *Use case:* I'm trying to install arrow's R package on a Linux (RHEL) server
> without internet access. The server only has access to CRAN and a few other
> repo's via a mirror on intranet (Artifactory instance). In particular, no
> access to public RStudio Package Manager or to apache.jfrog.io or github.
> Currently (v9.0.0), installation script does not really foresee the
> possibility to install from binaries in such a case. By default it will build
> from source without any external dependencies, which is slow and less
> feature-rich.
> *Suggestion:* if the user has the possibility to manually download the right
> zipfile for libarrow, he can set the (new) environment variable
> `ARROW_DOWNLOADED_BINARIES` to the local path, and the installation script
> will use that pre-downloaded file.
> In addition, in case of failed download, the URL will be printed. With this
> extra info, it will be easier for the user to identify the right URL for
> libarrow zipfile and then download it manually.
>
> I just proposed a PR...
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
