Can you tell us more about your use case? Is the JSON deeply nested or mostly flat? It wouldn't be difficult to create some built-in functions to extract JSON from a VARCHAR field, but performance wouldn't be ideal.
My gut tells me that JSON would be handled higher up in the stack, through translating JSON to a table structure on the way in and then back to JSON from a result sets on the way out. Some ideas I have on JSON would be to: 1) Implement the concept of a STRUCT 2) Map the JSON to a STRUCT 3) Support functional indexes so that parts of a STRUCT could be indexed. Unfortunately, none of these are implemented yet. Would be great for someone to step in and own this area. Thanks, James On Wed, May 14, 2014 at 5:12 PM, Stephen Sprague <[email protected]> wrote: > thought i'd ping again on this topic. Any json extract type functions on > the horizon? > > Cheers, > Stephen. > > > On Thu, May 8, 2014 at 5:14 PM, Stephen Sprague <[email protected]>wrote: > >> hey good people, >> >> what is the likelihood of a "json_extract" type function being available >> in phoenix any time soon? >> >> i was thinking the json would be stored in a vanilla varchar but here on >> the roadmap (http://phoenix.incubator.apache.org/roadmap.html) i see it >> discussed as a separate datatype. >> >> and speaking of which if the first sentence on that page is still correct >> by my calculations everything preceding the json line item (#7) i think is >> done. >> >> could that mean that json is around the corner? :) >> >> thanks, >> Stephen. >> > >
