Alternative to slow SRF in SELECT?

[Markur Sens <markursens@xxxxxxxxx>]

I have the following  case

select 
	my_srf_func(otherfunc(h))
from (values (‘…’::mytype),
             ‘…’::mytype),
             (‘…’::mytype),
             ‘…’::mytype),
             (‘…’::mytype),)
         as temp(h);


I think it’s obvious what I’m trying to do.

My_srf_func is similar to unnest(array) function.

The problem now is that the result size of this could be around 400-500 elements for every function call of otherfunc(h) so when my_srf_func unnests these I should expect 2K+ items even for this simple case.

It’s also not advised to have SRFs in the SELECT of the query, becuase I do get some unexpected/unintuitive results,
but most importantly it’s way too slow.

Any alternatives on how I could rework the query or the my_srf_func to speed things up, without having too many subqueries? 

I guess there should be some kind of pattern there.

If there’s any help, both my_srf_func and otherfunc are coded in PL/Python

Thanks.