Error handling function TRY_TO_TIMESTAMP not working as expected - snowflake-cloud-data-platform

Question

I have a column (for ex Col1) which holds string value 'Jan 14 2019 5:30PM' and would like to convert it into timestamp_ntz(6) format in the target table in Snowflake.
Using Logic like this
INSERT INTO <target_table>
SELECT try_to_timestamp(<Col1>,'MON DD YYYY HH12:MI PM')
FROM <source_table>
This is throwing an error:
TRY_CAST cannot be used with arguments of type TIMESTAMP_NTZ(9) and TIMESTAMP_NTZ(9).

Answer 1

SELECT try_to_timestamp(column1,'MON DD YYYY HH12:MI PM')
FROM VALUES
('Jan 14 2019 5:30PM');
works:
TRY_TO_TIMESTAMP(COLUMN1,'MON DD YYYY HH12:MI PM')
2019-01-14 17:30:00.000
BUT pre-parsing the string to an actually timestamp, and then trying to parse that fails
SELECT try_to_timestamp(column1,'MON DD YYYY HH12:MI PM')
FROM VALUES
(to_timestamp('Jan 14 2019 5:30PM','MON DD YYYY HH12:MI PM'));
gives the error:
Function TRY_CAST cannot be used with arguments of types TIMESTAMP_NTZ(9) and TIMESTAMP_NTZ(9)
original this type of problem would also happen if you passed a BOOLEAN to TRY_TO_BOOLEAN.
If you do not know if "all you input is timestamps or text (like if it's coming from a VARIANT data where auto magic conversion could be happening.
I would say
You can cast to ::TEXT and then parse again.
but that also errors.
is_timestamp_ntz(b)
Fails if passed non-variant data. So that's not helpful.
A really yucky work around:
setup:
create temp table data(json variant);
insert into data select parse_json(column1) as json
FROM VALUES
('{"date":"Jan 14 2019 5:30PM"}'),
('{"date":"2021-01-24 12:34:56"}');
sql:
SELECT
is_timestamp_ntz(json:date) as a
,try_to_timestamp_ntz(json:date::text,'MON DD YYYY HH12:MI PM') as b
,try_to_timestamp_ntz(json:date::text) as c
,try_to_timestamp_ntz(nvl(b,c)::text) as d
,nvl(try_to_timestamp_ntz(json:date::text,'MON DD YYYY HH12:MI PM'),try_to_timestamp_ntz(json:date::text)) as sigh
FROM data;
thus I suggest opening a support ticket, as it's gross the auto classification of JSON data, then exploding the parsing code, and in the mean time here is the work around, cast it to text, and parse if your expected format, and let the system default catch the things there where timestamp, that are turned to text, and then back again.
nvl(try_to_timestamp_ntz(json:date::text,'MON DD YYYY HH12:MI PM'),try_to_timestamp_ntz(json:date::text)) as sigh

Answer 2

try this:
select try_cast(to_varchar(column1) as timestamp) from table;