removal of that insert optimization that probably doesn't do anything, turns out it was important about cartodb

[knownasilya]

maybe leave a comment in the code why it is 👯

[calvinmetcalf]

honestly I put it in because I thought maybe it would be cool, never tested or anything

[knownasilya]

Would love to know why it broke things though, not really familiar with all the things this lib does.

[calvinmetcalf]

so the way we parameterize stuff is by going like this for most stuff

BEGIN;
prepare _named_expression(numeric, text, boolean) as
select foo from bar where col = $1 and col2 = $2 and col3 = $3;
execute _named_expression(numeric $$5$$, $cartodb$string$cartodb$, true);
commit;

but for inserts we did

BEGIN;
prepare _named_expression(numeric,text, boolean) as
insert into table (col1, col2, col3) values ($1, $2, $3);
execute _named_expression(numeric $$1$$, $cartodb$string1$cartodb$, true);
execute _named_expression(numeric $$2$$, $cartodb$string2$cartodb$, false);
execute _named_expression(numeric $$3$$, $cartodb$string3$cartodb$, true);
commit;

but i had no proof that this was actually a speedup and assumed it was just me being clever so when I did the stuff to make it so you could quote columns and tables i changed it to be

BEGIN;
prepare _named_expression(numeric,text, boolean) as
insert into table (col1, col2, col3) values ($1, $2, $3), ($4, $5, $6),($7, $8, $9);
execute _named_expression(numeric $$1$$, $cartodb$string1$cartodb$, true, numeric $$2$$, $cartodb$string2$cartodb$, false, numeric $$3$$, $cartodb$string3$cartodb$, true);
commit;

which seems to be much less performant