DMN 1.4 - scales

[StrayAlien]

Hey all, with the DMN 1.4 PRs here now, there are quite a number that deal with numeric scales.

Admittedly, my understanding of how we're supposed to handle negative scales is pretty patchy. Hence leaving off tests until I / we all have an understanding. TBH, it is not is much "what is a negative scale", that is okay. More like, "how to transition between scales when a negitve is involved". Debugging through the Java BigDecimal code is is not straight forward at all.

So, opening up a discussion here so we can all understand better.

First helpful comment is by @opatrascoiu here: #592 (comment). :-)

[opatrascoiu]

According to DMN spec, Section 10.3.2.3.1 number

FEEL Numbers are based on IEEE 754-2008 Decimal128 format, with 34 decimal digits of precision and rounding
toward the nearest neighbor with ties favoring the even neighbor. Numbers are a restriction of the XML Schema
type precisionDecimal, and are equivalent to Java BigDecimal with MathContext DECIMAL 128.

Looks obvious to me the tests should match the setScale() method for BigDecimal with MathContext DECIMAL 128.

I think the same behavior as java BigDecimal can be achieved in Python by using decimal.Decimal. I have identical negative scale tests running both on Java and Python.

[opatrascoiu]

According to this https://docs.oracle.com/javase/8/docs/api/java/math/BigDecimal.html the scale is a 32 bit signed integer.

Hence, the tests that cover overflow / underflow of scale should use the -2,147,483,648 and 2,147,483,647.

[opatrascoiu]

Following our discussion today, I created two DMN issues to cover the scale issues:
https://issues.omg.org/browse/DMN16-81
https://issues.omg.org/browse/DMN16-82