Replace t1 col with t2 col when [condition] + different datatype - sql

Question

I have this situation: deceduti(sigla,denominazione_provincia) , province(sigla_provincia,denominazione_provincia)
where sigla is a NUMBER and sigla_provincia is a CHAR(2).
I want to set deceduti.sigla=province.sigla_provincia where deceduti.denominazione_provincia=province.denominazione_provincia.
I have tried different ways but no one worked. I'd appreciate to understand how can i do this with join operator. There are my tries:
MERGE INTO DECEDUTI -- error with the datatypes
USING PROVINCE
ON ( DECEDUTI.DENOMINAZIONE_PROVINCIA = PROVINCE.DENOMINAZIONE_PROVINCIA )
WHEN MATCHED THEN
UPDATE SET DECEDUTI.Sigla = PROVINCE.sigla_provincia;
/
UPDATE -- oracle doesnt have the on update+policy react
(SELECT DECEDUTI.Sigla as OLD, to_number(PROVINCE.Sigla_provincia) as NEW
FROM DECEDUTI
INNER JOIN PROVINCE
ON DECEDUTI.DENOMINAZIONE_PROVINCIA = PROVINCE. DENOMINAZIONE_PROVINCIA
) t
SET t.OLD = t.NEW
/
UPDATE deceduti T
SET T.sigla =
(SELECT distinct sigla_provincia
FROM PROVINCE A
WHERE A.denominazione_provincia = T.denominazione_provincia);

Answer 1

This is what you have:
create table deceduti
(sigla number,
denominazione_provincia unknown_datatype
);
create table province
(sigla_provincia char(2),
denominazione_provincia unknown_datatype
);
Data stored within looks like this (according to datatypes you mentioned):
DECEDUTI:
sigla denominazione_provincia
----- -----------------------
25 1000
null 2000
PROVINCE:
sigla_provincia denominazione_provincia
--------------- -----------------------
AB 1000
X1 2000
As tables are joined on denominazione_provincia, you'd like to - for its value = 1000 - put AB into the deceduti.sigla column. How do you plan to put AB into a NUMBER datatype column? That won't work.
The simplest option is to
alter table deceduti modify sigla (char(2));
but only if sigla column is empty; shouldn't be problem to set it to NULL as it is supposed to accept a new value anyway. If you want to keep some "old" values, then it gets more complicated (creating a new column, store current sigla value in there, empty sigla, change its datatype, move data back, drop new column).
Other options are more complicated, e.g.
make sure that all sigla_provincia values are numeric so that to_number(sigla_provincia) returns a number
update deceduti.sigla values only for denominazione_provincias whose pair in province contains number in sigla_provincia