fix: TRY_CAST returns NULL for timestamp/date overflow

[fengys1996]

Which issue does this PR close?

• Closes TRY_CAST returns an error instead of NULL for timestamp/date overflow #22896.

Rationale for this change

TRY_CAST should return NULL on cast failure, but overflowing date/timestamp casts returned errors.

What changes are included in this PR?

• Make scalar temporal overflow checks respect CastOptions.safe.
• Skip DataFusion’s array pre-check for safe casts so Arrow can return NULLs.
• Add regression tests.

Are these changes tested?

Yes:

cargo test -p datafusion-common timestamp_overflow_returns
cargo test -p datafusion-expr-common timestamp_array_to_timestamp_overflow
cargo test --test sqllogictests -- datetime/timestamps.slt

Are there any user-facing changes?

Yes. TRY_CAST for overflowing date/timestamp casts now returns NULL; regular CAST still errors.

Known Limitation

This PR does not add Date array-path coverage yet. For example:

SELECT TRY_CAST(d AS TIMESTAMP(9))
FROM (VALUES (DATE '3000-01-01')) t(d);

This depends on the upstream Arrow fix in apache/arrow-rs#9825. Once DataFusion updates to an Arrow version containing that fix, we can add this regression test.

[Jefffrey]

This PR does not add Date array-path coverage yet. For example:

SELECT TRY_CAST(d AS TIMESTAMP(9))
FROM (VALUES (DATE '3000-01-01')) t(d);

This depends on the upstream Arrow fix in apache/arrow-rs#9825. Once DataFusion updates to an Arrow version containing that fix, we can add this regression test.

perhaps we can add this right now, but with expected error; that way when the upstream fix is merged and we upgrade to the arrow version we'll automatically know about the fix

[Jefffrey]

Suggested change

	if cast_options.safe {
	if multiplier > 1 && value.checked_mul(multiplier).is_none() {
	return ScalarValue::try_new_null(target_type);
	}
	} else {
	ensure_timestamp_in_bounds(value, multiplier, &source_type, target_type)?;
	}
	match ensure_timestamp_in_bounds(value, multiplier, &source_type, target_type)
	{
	Ok(()) => {}
	Err(_) if cast_options.safe => {
	return ScalarValue::try_new_null(target_type);
	}
	Err(e) => return Err(e),
	}

perhaps like this to reuse the validation logic from ensure_timestamp_in_bounds?

[fengys1996]

Thanks for your review. @Jefffrey

perhaps we can add this right now, but with expected error; that way when the upstream fix is merged and we upgrade to the arrow version we'll automatically know about the fix

I tried adding the Date array-path SLT coverage by catching the panic as an expected error in 3f4ad71.

But the behavior is profile-dependent: debug panics, while release wraps and succeeds. It is not stable for CI, so I reverted it.

perhaps like this to reuse the validation logic from ensure_timestamp_in_bounds?

Good suggestion, Done in 970d1ae.