Ok i got it , thank a lot Danny
Le mar. 26 nov. 2019 à 01:53, Danny Chan <yuzhao....@gmail.com> a écrit : > Thanks, Steve ~ > > Calcite did support LIST type syntax, such as “INT ARRAY” or “INT MULTISET > ARRAY”, but we do not support MAP syntax yet, because it is not a SQL > standard, so the MAP type would always parsed to other type. > > You can reference a ARRAY column item with syntax “column_name[_index]”, > and MAP column value with syntax “column_name[‘key_name’]”, these syntax > also works with “OTHER”/“ANY” data type. > > Best, > Danny Chan > 在 2019年11月25日 +0800 PM8:22,Steve Robert <contact.steverob...@gmail.com > >,写道: > > Hi , > > > > I saw an interesting exchange about the CSVAdapter > > > > I am also in a similar case where I want to be able to perform predicate > > filters from table > > after conversion from JSON to JsonScannableTable . > > > > > > the first dimension are correctly converted into the right type VARCHAR > ect > > but other dimension like LIST or MAP are converted to "OTHER"-type > > it is therefore impossible to execute CONTAINS / IN OR ANY on column > > transformed to OTHER-TYPE > > > > > > I would like to know where it is the code best place to convert LIST MAP > > to a SQL type ARRAY_LIST and no OTHER > > > > I am new in calcite I have a little trouble understanding the type > > definition workflow. I looked at the side of JSONEnumerator but it seems > > opaque. > > > > thank you in advance > > > > Steve >