And of course if you want to repair the situation in the source, something like update table set <column> = ' ' where <column> is null; would do the trick. (Where there is a single space between the apostrophes, but of course this does change your data. You would have to be sure it was a useful change.) Unless there are already single space values in this table that mean something different from the meaning of NULL, there is no change in meaning. Of course that does not mean it won't make some poor program that currently works break. mwf From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Mark W. Farnham Sent: Tuesday, April 15, 2014 2:04 PM To: oratune@xxxxxxxxx; howard.latham@xxxxxxxxx; giantpanda@xxxxxxx; 'ORACLE-L' Subject: RE: Null data in not null column Likely David is correct. If in the sqlplus session reporting the data you preface the query with set null ~ then the columns containing actual nulls will be definitively displayed visually. also, your query has a column header of length(ANZEIGE_NAME), but that is NOT the query you ran. Please clear up this fishiness. mwf From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of David Fitzjarrell Sent: Tuesday, April 15, 2014 1:28 PM To: howard.latham@xxxxxxxxx; giantpanda@xxxxxxx; ORACLE-L Subject: Re: Null data in not null column SQL> insert into emp(empno) values (""); ERROR: ORA-01741: illegal zero-length identifier SQL> SQL> SQL> SQL> SQL> insert into emp(empno) values (''); insert into emp(empno) values ('') * ERROR at line 1: ORA-01400: cannot insert NULL into ("GRIBNAUT"."EMP"."EMPNO") SQL> TJ is correct, the NOT NULL constraint was added after the NULL values were entered; it was created NOVALIDATE so it wouldn't fail on creation due to the existing NULL values. It's possible to disable the constraint on the source table before the export then enable it novalidate in the destination database after the table has been imported. Of course you would also need to enable the constraint novalidate in the source database after the export has completed. David Fitzjarrell Primary author, "Oracle Exadata Survival Guide" On Tuesday, April 15, 2014 11:13 AM, "howard.latham@xxxxxxxxx" <howard.latham@xxxxxxxxx> wrote: Set it to "" ----- Reply message ----- From: "Ingrid Voigt" <giantpanda@xxxxxxx> To: "oracle-l@xxxxxxxxxxxxx" <oracle-l@xxxxxxxxxxxxx> Subject: Null data in not null column Date: Tue, Apr 15, 2014 4:53 PM Hi, one of our customers' databases has a table with a not null column containing empty data. Do you know how this can be set up? I need to copy the table (via exp/imp) and hit either ORA-01400 (when defining not null before insert) or ORA-02296 (when I insert first and later modify the column). More details: Windows 2008 R2 Oracle 11.2.0.3 + Patch 27 (January 2014) Table definition (more columns left out): CREATE TABLE USR_USER ( OID NUMBER(15) NOT NULL, ANZEIGE_NAME VARCHAR2(40 BYTE) DEFAULT '' NOT NULL, ) TABLESPACE USERS; And the data: > select oid, anzeige_name, '.'||anzeige_name||'.' from usr_user; OID ANZEIGE_NAME LENGTH(ANZEIGE_NAME) ---------- ---------------------------------------- -------------------- 223 224 (more rows left out) These are not strings consisting of spaces, but really empty ones, i.e. nulls. Insert into the table without specifying anzeige_name also hits ORA-01400, the condition is active. The corresponding check constraint is enabled and validated. I'd appreciate help in reproducing this. Thank you. Ingrid Voigt -- //www.freelists.org/webpage/oracle-l