Should we error when generating `first_value(...) IGNORE NULLS` for dialects that don't support it?

[NickCrews]

Is your feature request related to a problem? Please describe.
I am implementing ibis-project/ibis#11311, which is trying to compile a FirstValue expression with a include_nulls: bool property.

This is the current behavior of sqlglot:

import sqlglot as sg

duckdb_sql = """SELECT FIRST_VALUE(t1.c1) IGNORE NULLS OVER (PARTITION BY t1.c2 ORDER BY t1.c3) FROM t1;"""
pg_sql = sg.transpile(duckdb_sql, read="duckdb", write="postgres")[0]
print(pg_sql) # SELECT FIRST_VALUE(t1.c1) IGNORE NULLS OVER (PARTITION BY t1.c2 ORDER BY t1.c3) FROM t1
# eg the sql is emitted unchanged

But, if you look at the postgres docs, at the bottom you will find

Note
The SQL standard defines a RESPECT NULLS or IGNORE NULLS option for lead, lag, first_value, last_value, and nth_value. This is not implemented in PostgreSQL: the behavior is always the same as the standard's default, namely RESPECT NULLS. Likewise, the standard's FROM FIRST or FROM LAST option for nth_value is not implemented: only the default FROM FIRST behavior is supported. (You can achieve the result of FROM LAST by reversing the ORDER BY ordering.)

So, the generated SQL above errors at runtime if you actually try to execute it.

Describe the solution you'd like
My question is: Should sqlglot error during compilation? I'm not sure the precedent in other parts of sqlglot.

Describe alternatives you've considered

• throw a warning?
• keep the existing behavior of doing nothing

Additional context
NA

[NickCrews]

Ahh, I just found in the sqlite compiler:

        def ignorenulls_sql(self, expression: exp.IgnoreNulls) -> str:
            self.unsupported("SQLite does not support IGNORE NULLS.")
            return self.sql(expression.this)

so it looks like there is a precendent for how we should handle this.

[georgesittas]

Yeah, seems like it. Feel free to take a stab at this if you want @NickCrews.