As far as I know, PQ creates some internal presentations for calculating M scripts, reshuffling M queries and making something like tunnels/pipelines to data source from a number of steps, which are described in M scripts.
E.g. internally
s0 = #table..,
s1 = Table.AddColumn(s0,"2", each f(_)) ,
s2 = Table.RemoveColumns (s1,"1")
is the same as
s0 = #table.. ,
s1 = Table.TransformColumns(s0, {"1", each f(_)}) Defenitely, it would be great to virtually see the resulting "tunnels", just like SQL programmers can see query plans of SQL queries.
It would help M developers in many ways:
1. They could really understand what M constructions are more effective that others. Even if you are an experienced M developer, you cannot imaging the whole tunnel for a query of 10 Queries of 10 - 50 steps in each.
2. They could really understand when they should stop keep visual representing of step result, and start to hard-code 10-ths folding of M functions in one step in PQ UI.
Now, from my experience, in case of complex ETL, the time, which you gain due to using functional language & rich concept & functionality of M, will be loosed on boring analysis of the possible variants of tunneling and trying different variants without clues what is really more effective or not.
