[SPARK-9277] [MLLIB] SparseVector constructor must throw an error when declared number of elements less than array length

[srowen]

Check that SparseVector size is at least as big as the number of indices/values provided. And add tests for constructor checks.

CC @MechCoder @jkbradley -- I am not sure if a change needs to also happen in the Python API? I didn't see it had any similar checks to begin with, but I don't know it well.

[mengxr]

LGTM. I think it is useful to add the same check to Python. @MechCoder could you add it after #7746 ? Thanks!

[SparkQA]

Test build #39046 has finished for PR 7794 at commit 6ffe34a.

• This patch fails Scala style tests.
• This patch merges cleanly.
• This patch adds no public classes.

[mengxr]

space after ,

[srowen]

Oh duh, fix coming. Every time I think I can't possibly need to run scalastyle as well as the test ...

[SparkQA]

Test build #39061 has finished for PR 7794 at commit e8dc31e.

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

[jkbradley]

@MechCoder Actually, it looks like the Python SparseVector constructor is really inefficient, requiring either (a) a full pass over the indices or (b) sorting the indices, depending on what is passed to the constructor. Would you mind making a JIRA for making the Python constructor equivalent to the Scala one? Thank you!

[MechCoder]

Thanks you for the ping! I'll have a deeper look now.

[MechCoder]

please check #7854