ARROW-9827: [C++][Dataset] Skip parsing RowGroup metadata statistics when there is no filter

[jorisvandenbossche]

No description provided.

[github-actions]

https://issues.apache.org/jira/browse/ARROW-9827

[jorisvandenbossche]

@bkietz ideally we would also skip this if the flilter only involves the partition_expression and not any actual columns of the file. What is the best way to check this? (simplify the filter first with partition_expression ?)

[bkietz]

Simplifying the filter by the partition expression would work, but it's not safe without loading the physical schema since we need to validate the filter against that. Otherwise a filter without implicit casts (for example) would result in an assertion failure. Let's leave that optimization for a follow up; the right way to do this is probably to rewrite Expression::Assume to return a Result (allowing it to fail non-catastrophically even without a schema against which to validate), but that's a much larger project

[bkietz]

Looks good to me, please add a comment explaining the purpose of the new code inline

[bkietz]

Simplifying the filter by the partition expression would work, but it's not safe without loading the physical schema since we need to validate the filter against that. Otherwise a filter without implicit casts (for example) would result in an assertion failure. Let's leave that optimization for a follow up; the right way to do this is probably to rewrite Expression::Assume to return a Result (allowing it to fail non-catastrophically even without a schema against which to validate), but that's a much larger project

[bkietz]

Suggested change

	} else {
	} else {
	// since we are not scanning this fragment with a filter, don't bother loading statistics

[jorisvandenbossche]

@bkietz with some delay, added the suggested comment

[bkietz]

LGTM

https://issues.apache.org/jira/browse/ARROW-9945 for follow up to add the more general case

[jorisvandenbossche]

Thanks for opening the follow-up issue!