Hi Yaroslav, I did a quick dive into FLIP-37 [1] which was on the rework of the type system. In the FLIP it's noted:
> Decimal considerations: We use Hive’s maximums: https://github.com/apache/hive/blob/master/storage-api/src/java/org/apache/hadoop/hive/common/type/HiveDecimal.java So that's the reason why it was set to 38 :) Best regards, Martijn Visser https://twitter.com/MartijnVisser82 https://github.com/MartijnVisser [1] https://cwiki.apache.org/confluence/display/FLINK/FLIP-37%3A+Rework+of+the+Table+API+Type+System On Tue, 19 Apr 2022 at 02:42, Yaroslav Tkachenko <yaros...@goldsky.io> wrote: > Hey everyone, > > I'm curious if anyone knows the reason behind choosing 38 as > a MAX_PRECISION for DecimalType in the Table API? > > I needed to process decimal types larger than that and I ended up > monkey-patching DecimalType to use a higher precision. I understand it adds > a bit of overhead, but I haven't noticed any serious problems. > > Maybe this number can be configured. I noticed it's hardcoded in a few > places, but it should be a fairly straightforward change. > > Thanks! >
