[TINKERPOP-2643] Equality for NaN and BigDecimal

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 3.5.1
• Component/s: process
• Assignee: Unassigned
• Reporter: Stephen Mallette
• Created: 04/Nov/21 12:41
• Updated: 04/Nov/21 12:41

Description

JDK11 seems to produce a different error from JDK8 when it comes to BigDecimal comparisons that hit NaN and such. For JDK8 they seem to produce NumberFormatException but for JDK11 you get stuff like:

gremlin> g.V().has("key", Float.NaN)
Character N is neither a decimal digit number, decimal point, nor "e" notation exponential mark.

When Double/Float is stored, it always throws. With the proposed change, it wouldn't throw but because NaN is not equal to any numbers this returns empty result.

Equality around BigDecimal and special values which cannot be parsed as Integer such as{{NaN}}, INF should not produce exceptions and should filter.

gremlin> g.addV().property('key',Float.NaN)
==>v[0]
gremlin> g.addV().property('key',1.0f)
==>v[2]
gremlin> g.V().has('key',Float.NaN)
==>v[0]
gremlin> g.V().has('key',1.0f)
==>v[2]
gremlin> g.V().values("key").is(eq(1.0f)) // 3.5.x
==>1.0
gremlin> g.V().has('key',1.0) // 3.5.x - likely due to Groovy going to BigDecimal for "1.0"
java.lang.NumberFormatException
Type ':help' or ':h' for help.
Display stack trace? [yN]n
gremlin> g.V().values("key").is(eq(new BigDecimal(1.0f))) // 3.5.x
java.lang.NumberFormatException
Type ':help' or ':h' for help.
Display stack trace? [yN]
gremlin> g.V().has('key',1.0) // proposed
==>v[2]
gremlin> g.V().values("key").is(eq(1.0)) // proposed
==>1.0