ArrayIndexOutOfBoundsException: 200 on numbers with more than 200 digits.

[larsp]

On a simple data structure such as

{  
   "values":{  
      "datatypes":{  
         "test":999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999
      }
   }
}

Where the field values is a Map<String, ?>

I'll get a ArrayIndexOutOfBoundsException: 200 when the number has more than 200 digits.

Stacktrace:

Caused by: java.lang.ArrayIndexOutOfBoundsException: 200
    at com.fasterxml.jackson.core.json.UTF8StreamJsonParser._parsePosNumber(UTF8StreamJsonParser.java:1216) ~[com.fasterxml.jackson.core.jackson-core_2.4.3.jar:2.4.3]
    at com.fasterxml.jackson.core.json.UTF8StreamJsonParser.nextToken(UTF8StreamJsonParser.java:726) ~[com.fasterxml.jackson.core.jackson-core_2.4.3.jar:2.4.3]
    at com.fasterxml.jackson.databind.deser.std.UntypedObjectDeserializer$Vanilla.deserialize(UntypedObjectDeserializer.java:429) ~[com.fasterxml.jackson.core.jackson-databind_2.4.3.jar:2.4.3]
    at com.fasterxml.jackson.databind.deser.std.MapDeserializer._readAndBindStringMap(MapDeserializer.java:451) ~[com.fasterxml.jackson.core.jackson-databind_2.4.3.jar:2.4.3]

[cowtowncoder]

Thanks, sounds like some bounds check is off.

[cowtowncoder]

Actually, sort of; when calculating max bound, was forgetting to account for one character already appended. So was OBO problem.

Also: was a regression between 2.3->2.4 (due to low-level optimizations in squeezing out some more speed); meaning that 2.3.x was NOT affected, just 2.4.

[larsp]

Thanks!
To cite my favorite computer science joke: "The two most common problems in computer science are: Off by one errors"

[cowtowncoder]

Indeed!

But I have realized that cost of trying out tons of low-level optimization with 2.4 was non-trivial. I over-estimated coverage of test suite; there have been almost half a dozen cases similar cases.
Could be worse however; I think all have been array-out-of-bounds, and none actual data corruption (skipping a digit or such).

Anyway, here's hoping that we are done with finding obscure regressions. :)