[SPARK-14081][SQL] - Preserve DataFrame column types when filling nulls.

[traviscrawford]

What changes were proposed in this pull request?

This change resolves an issue where DataFrameNaFunctions.fill changes a FloatType column to a DoubleType. We also clarify the contract that replacement values will be cast to the column data type, which may change the replacement value when casting to a lower precision type.

How was this patch tested?

This patch has associated unit tests.

[traviscrawford]

The data type pattern match has been removed because I believe its unnecessary. I'm happy to put this back, along with a comment describing what special-case it covers, if someone can clue me into why its needed.

[rxin]

I think this is used to convert NaN values into null. Otherwise we won't fill NaN values.

[JoshRosen]

Jenkins, this is ok to test.

[SparkQA]

Test build #54216 has finished for PR 11967 at commit 834ee69.

• This patch fails Spark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[rxin]

@traviscrawford can you remove the part about this is your first contribution from the pr description? The pr description will actually become part of the commit.

[rxin]

Looks like there is a test failure.

[SparkQA]

Test build #54523 has finished for PR 11967 at commit 9dfe2dd.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[rxin]

@traviscrawford can you update the pr title to have a full sentence? github is cutting it off.

[rxin]

LGTM - we should merge this as soon as you update the title.

[traviscrawford]

Title updated.

[rxin]

Thanks - I've merged this in master.