Using NVL for date

Question

I am creating a view and one of the columns is this:
(b.startdate || ' To ' || b.enddate) salesDateRange
however when I was trying to handle the null value of date, I found the following problems,
case 1
(nvl(b.startdate, 'NA') || ' To ' || b.enddate) salesDateRange
when i use nvl, I am able to create the view, but when I retrieve the data, I get this error msg.
ORA-01858: a non-numeric character was found where a numeric was expected
01858. 00000 - "a non-numeric character was found where a numeric was expected"
*Cause:    The input data to be converted using a date format model was
incorrect. The input data did not contain a number where a number was
required by the format model.
*Action:   Fix the input data or the date format model to make sure the
elements match in number and type. Then retry the operation.
is nvl only available for numerical value?

Answer 1

user8944947 wrote:
I am creating a view and one of the columns is this:
(b.startdate || ' To ' || b.enddate) salesDateRange
however when I was trying to handle the null value of date, I found the following problems,
case 1
(nvl(b.startdate, 'NA') || ' To ' || b.enddate) salesDateRange
when i use nvl, I am able to create the view, but when I retrieve the data, I get this error msg.
ORA-01858: a non-numeric character was found where a numeric was expected
01858. 00000 - "a non-numeric character was found where a numeric was expected"
*Cause:    The input data to be converted using a date format model was
incorrect. The input data did not contain a number where a number was
required by the format model.
*Action:   Fix the input data or the date format model to make sure the
elements match in number and type. Then retry the operation.
is nvl only available for numerical value?In addition to the answers already given, you have to consider that NVL is an overloaded function, designed to deal with different datatypes. It accepts 2 arguments and the version of NVL that is used depends on the arguments given (that's the nature of overloading - look it up on the internet if you're not sure).
So, there is defined e.g.
NVL({noformat}<number arg>, <number arg>{noformat})
NVL({noformat}<varchar2 arg>, <varchar2 arg>{noformat})
NVL({noformat}<date arg>, <date arg>{noformat})
etc.In your example, as the first argument you are passing in is a date, Oracle is opting to use:
NVL({noformat}<date arg>, <date arg>{noformat})But your second argument is not a date, it's a varchar2.
However, Oracle is clever and can perform "implicit" conversions between datatypes i.e. it will try and convert datatypes without you actually telling it to. (It's generally good coding practice to do it yourself though).
So, in you case Oracle is taking your varchar2 string "NA" and trying to implicitly convert this to a date, using the only way it knows how, which is to apply a date format mask as defined by the NLS_DATE_FORMAT parameter for the session. For example, internally it will be trying to do the equivalent of:
TO_DATE('NA','DD-MON-YYYY')(assuming your NLS_DATE_FORMAT parameter is set to DD-MON-YYYY)
If we try that ourselves, we get...
SQL> select to_date('NA','DD-MON-YYYY') from dual;
select to_date('NA','DD-MON-YYYY') from dual
ERROR at line 1:
ORA-01858: a non-numeric character was found where a numeric was expected... which is the same error you were getting.
So, in order for your NVL to work you would either need to supply a DATE as the second argument (some default date to use if the first argument is null), or some varchar2 string that represents a date in the correct format as per your NLS_DATE_FORMAT setting (dangerous because if the NLS_ setting changes then your code can suddently error), or you provide the first argument as a VARCHAR2 string instead of a date (using TO_CHAR), so the the function overloading causes Oracle to choose the varchar2 argument version of NVL, for which 'NA' is a perfectly valid second argument.