Question about CONCAT_NULL_OUTPUTS_STRING #1789
-
Why is The comment on Does this mean in this case, that NULL is treated as an empty string in postgres (basically being ignored), which is why this flag is true in postgres? I'm asking, because this breaks postgres.
However, it's converted into Which is why I am not sure, what problem this flag is trying to solve. Maybe it should be a generator setting instead of a parser setting? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
EDIT: fixed in 1b62c0a and deployed with v16.3.0 Hey @mpf82, Sorry for the inconvenience, I'll take a look and try to resolve this issue. The motivation behind this flag was to improve Thanks for the report. |
Beta Was this translation helpful? Give feedback.
EDIT: fixed in 1b62c0a and deployed with v16.3.0
Hey @mpf82,
Sorry for the inconvenience, I'll take a look and try to resolve this issue. The motivation behind this flag was to improve
the transpilation of
CONCAT
when targeting dialects that have strict requirements regarding the types of its arguments,e.g. Trino. I must've missed the case you're referring to in my implementation; I didn't expect
COALESCE
to break here(e.g. your example doesn't break in DuckDB and T-SQL, which also set the flag to
True
).Thanks for the report.