Yes Daniel - 2002 is Struct and 2003 is Array (I'm seeing both)

This is not a Drill problem but a squierrel problem, and an annoying one at
that, I've tried using the option to display unknown data types as strings,
now I just <Error> rather than the "Error Unknown Type (2003 or 2002)"



On Mon, Oct 12, 2015 at 3:37 PM, John Omernik <[email protected]> wrote:

> That seems to be what I am looking for, however, when I do
> convert_from(flatten(kvgen(allias.`column`)), 'json')  I get an error
> "Missing function implementation (convert_fromjson, "map required")  when I
> do "convert_to(flatten(kvgen()), 'json') then I get what appears to be hex
> encoded data... (Just long strings of hex, I wonder if Squirrel is doing
> that too)
>
>
> On Mon, Oct 12, 2015 at 2:57 PM, Jason Altekruse <[email protected]
> > wrote:
>
>> We don't implement casts on array or map, but we do have a convert
>> function
>> that will convert a complex structure to json.
>>
>> You can invoke it like this:     convert_from( map_or_list_column_name,
>> 'JSON')
>>
>> This will return the data serialized into JSON in a varchar column.
>>
>> On Mon, Oct 12, 2015 at 12:27 PM, John Omernik <[email protected]> wrote:
>>
>> > Is there an easy way in drill to convert from an ARRAY or MAP to a
>> string
>> > representation of the same? I know this is an odd question, but I
>> realized
>> > I just was spinning my wheels because I thought I was doing something
>> wrong
>> > with JSON and KVGEN and FLATTEN but in reality, it was working fine, I
>> just
>> > didn't see it because SQL Squirrel kept returning "<Unknown Type
>> (2,002)>
>> > instead of the results of the function.
>> >
>> > I guess, if I could take the output from FLATTEN(KVGEN()) and wrap it
>> in a
>> > "CAST(FLATTEN(KVGEN()))" then at least I could continue exploration.
>> >
>> > I know that it is a workaround for the limitations of Squirrel, but I'd
>> be
>> > interested in other's thoughts on that, is there a tweak to squirrel
>> that
>> > would allow that to work properly, or just casting it as needed. (Note I
>> > tried CAST as STRING  and CASTS as VARCHAR with not much success.
>> >
>>
>
>

Reply via email to