Hello, team. Is there any way to solve this case for custom ETL source. For example, i have DataFlow with DataJob, which insert data from 10 datasets to another 10 datasets

insert into a select * from k;
insert into b select * from l;
insert into c select * from m;
insert into d select * from n;
insert into e select * from o;
insert into f select * from p;
insert into g select * from q;
insert into h select * from r;
insert into e select * from s;
insert into j select * from t;

if i set to this DataJob outputs = a,b,c,d,e,f,g,h,e,j and inputs = k,l,m,n,o,p,q,r,s,t i get mixed (and wrong) lineage for datasets. Now i generate fake 10 DataJobs (outputs = a/inputs = k; outputs = b/inputs = l and etc.) to solve this. Is there any other better solution?

I think technically your data job consumes all those datasets and produces the others and it is correct you see 10 inputs and 10 outputs. What you also want to see is the dependencies between the dataset. That I think should be emitted separately. This you can get on Redshift, Snowflake, Bigquery if you enable lineage generation.