Replace doing funny things

• Simon Sabin

  SSCrazy Eights

  Points: 8142

  More actions

  November 6, 2003 at 7:06 am

  #84728

  Can anyone explain the different behaviour of replace,

  create database a collate SQL_Latin1_General_CP1_CI_AS

  go

  use a

  go

  declare @C binary(100)

  set @C =cast( 'simon' + replicate (95, 0x00) as varbinary(100))

  select @C

  select replace(@c ,0x00,0x20)

  select cast(@c as varchar(100))

  go

  use master

  go

  drop database a

  go

  create database a collate Latin1_General_CI_AS

  go

  use a

  go

  declare @C binary(100)

  set @C =cast( 'simon' + replicate (95, 0x00) as varbinary(100))

  select @C

  select replace(@c ,0x00,0x20)

  select cast(@c as varchar(100))

  go

  use master

  go

  drop database a

  
  Simon Sabin
  SQL Server MVP
  
  http://sqlblogcasts.com/blogs/simons

• fhanlon

  SSChampion

  Points: 13131

  More actions

  November 10, 2003 at 9:09 am

  #481449

  The different behavior has to do with the different collation sequences. To see what the collations you defined are run:

  SELECT * FROM ::fn_helpcollations ()

  WHERE name = 'Latin1_General_CI_AS' OR name = 'SQL_Latin1_General_CP1_CI_AS'

  According to the SQL Server Books Online, "All Unicode data uses the same Unicode code page. Collations do not control the code page used for Unicode columns, only attributes such as comparison rules and case sensitivity. " The comparison rules are different so REPLACE works differently, even if the collations you chose are similar like 'Latin1_General_CI_AS' and 'SQL_Latin1_General_CP1_CI_AS'

  Francis

• Simon Sabin

  SSCrazy Eights

  Points: 8142

  More actions

  November 12, 2003 at 7:08 am

  #481693

  That doesn't explain why it has replaced everything with nothing for the Latin1 collation. It must be a bug.

  Simon Sabin

  Co-author of SQL Server 2000 XML Distilled

  http://www.amazon.co.uk/exec/obidos/ASIN/1904347088

  
  Simon Sabin
  SQL Server MVP
  
  http://sqlblogcasts.com/blogs/simons

• NPeeters

  SSChampion

  Points: 12535

  More actions

  November 12, 2003 at 8:21 am

  #481712

  The difference in Collation might explain the difference.

  The SQL_Latin1_... collation is a UNICODE collation. The Latin1_General_... is not.

  Under the hood, SQL Server is converting everything internally to a double byte string representation, which means the second byte is always 0x00. Replacing that, yields a very strange unicode representation of a string. That probably breaks some stuff...

  (Still thinking about this, like you say, it is not completely logical what is happening...)