Hi Wangsan,
currently, UDFs have very strict result type assumptions. This is
necessary to determine the serializers for the cluster. There were
multiple requests for more flexible handling of types in UDFs.
Please have a look at:
- [FLINK-7358] Add implicitly converts support for User-defined function
- [FLINK-9294] [table] Improve type inference for UDFs with composite
parameter and/or result type
- [FLINK-10958] [table] Add overload support for user defined function
I you think those issues do not represent what you need. You can open a
new issue with a little example of what feature you think is missing.
Regards,
Timo
Am 28.11.18 um 09:59 schrieb wangsan:
Hi all,
When using user-defined table function in Flink SQL, it seems that the result
type of a table function must be determinstic.
If I want a UDTF whose result type is determined by its input parameters, what
should I do?
What I want to do is like this:
```
SELECT input, f1, f2 length FROM MyTable, LATERAL TABLE(unnest_udtf(input, v1,
v2)) as T(f1, f2), LATERAL TABLE(unnest_udtf(input, v3, v4, v5)) as T(f3, f4,
f5)
```
I can surely register the same UDTF with different name and configuration, but
I guess that’s not a good idea :(.
If we can not make this in Flink SQL for now , may be we should consider this
feature in future?
Best,
wangsan
