Github user paul-rogers commented on a diff in the pull request:
https://github.com/apache/drill/pull/863#discussion_r124953472
--- Diff: exec/java-exec/src/main/codegen/templates/CastFunctions.java ---
@@ -54,14 +56,14 @@ public void setup() {}
public void eval() {
<#if (type.from.startsWith("Float") && type.to.endsWith("Int"))>
boolean sign = (in.value < 0);
- in.value = java.lang.Math.abs(in.value);
- ${type.native} fractional = in.value % 1;
+ ${type.native} value = java.lang.Math.abs(in.value);
+ ${type.native} fractional = value % 1;
--- End diff --
`out.value = Math.round(in.value)` ?
`Math.round()` rounds a double to the closest long; handling overly large
and small values, which our code does not do. The Java method is likely to be
much more efficient than this code.
Since the particular case is cast from `double` to `int`:
`out.value = Math.round((float) in.value)`
Returns an int.
I realize that this is existing code. But, it is broken. So, we might as
well fix it the right way...
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
