Question

I’m comparing text values in the same table and same column in SQL Server 2008, but the results are sometimes incorrect. Possibly relevant column info:

Data type: text  
Full Text: True  
Collation: SQL_Latin1_General_CP1_CI_AS

Out of 100 actual matches I only get 62. I have compared field values that are not picked up by LIKE in some comparison tools and they are 100% same, no additional spaces, line breaks etc.

Any idea why text LIKE doesn’t always work?

SQL:

SELECT a.text as one, b.text as two
  FROM [Db].[dbo].[Table] a join
  [Db].[dbo].[Table] b on
  a.text LIKE b.text

Answer 1

The following two examples show LIKE working exactly as expected. I suspect the reason you say “LIKE doesn’t always work” is because you don’t know how LIKE works.

Here’s an example where one match should occur:

CREATE TABLE #tmpTable
(
  [Text] TEXT
);

INSERT INTO #tmpTable VALUES ('Hello');

SELECT *
  FROM #tmpTable tmpTable1
       INNER JOIN #tmpTable tmpTable2
         ON tmpTable1.[Text] LIKE tmpTable2.[Text];

DROP TABLE #tmpTable;

“Hello is like Hello,” you say.

Here’s a match where no matches should occur:

CREATE TABLE #tmpTable
(
  [Text] TEXT
);

INSERT INTO #tmpTable VALUES ('[h]');

SELECT *
  FROM #tmpTable tmpTable1
       INNER JOIN #tmpTable tmpTable2
         ON tmpTable1.[Text] LIKE tmpTable2.[Text];

DROP TABLE #tmpTable;

“But [h] is like [h],” you say.

LIKE is not the equality operator, it has a special syntax for searching for single characters, ranges of characters and variable numbers of characters. If you happen to feed it text in a column that uses that special syntax (intentionally or not), it will treat the special syntax as character matching instructions, not as literal text.