[ARROW-8624] [Website] Install page should mention arrow-dataset packages

• Type: Improvement
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 0.17.0
• Fix Version/s: 1.0.0
• Component/s: Packaging
• External issue URL: https://github.com/apache/arrow/issues/24786
• Assignee: Kouhei Sutou
• Reporter: Neal Richardson
• Created: 29/Apr/20 15:10
• Updated: 11/Jan/23 08:01
• Resolved: 26/May/20 21:29

Description

I've seen a few reports like https://github.com/apache/arrow/issues/7055, where the user reports that they've installed the arrow system packages, we can see that they exist, but pkg-config reports that it doesn't have them. I think this is because -larrow_dataset isn't found. As the output on that issue shows, while arrow core headers and libraries are there, arrow_dataset is not.

Searching through the packaging scripts (such as https://github.com/apache/arrow/blob/master/dev/tasks/linux-packages/apache-arrow/yum/arrow.spec.in), while there is some metadata about a dataset package, I see that ARROW_DATASET=ON is not set anywhere, so I don't think we're building it. So apparently we are building it, but we aren't documenting how to get it.