MySQL Bugs: #30263: soundex function returns incorrect values in some cases

Bug #30263	soundex function returns incorrect values in some cases
Submitted:	7 Aug 2007 2:20	Modified:	8 Aug 2007 5:03
Reporter:	Lindsay Patten	Email Updates:
Status:	Verified	Impact on me:	None
Category:	MySQL Server: General	Severity:	S4 (Feature request)
Version:	5.0.45-community-nt	OS:	Any
Assigned to:		CPU Architecture:	Any

Description:
The values returned by soundex(String) do not conform to the correct values for some strings.  For example, soundex('Mannon') returns 'M000' but should return 'M550'. The soundex algorithm ignores a character if it has the same code as the previous character and vowels do not output a code, but a character that follows a vowel is not treated the same as a character that is a repeat. The first 'n' in Mannon follows an 'a' so it should generate a code of 5. The second 'n' follows directly after the first so it is ignored, but the final 'n' follows an 'o' so it also should generate a code.
Mnnn would generate M000 but Mannon should generate M550.
You likely fail to reset the variable storing the last code when you process a vowel character, but I have not looked at your code.

How to repeat:
select soundex('Mannon');

Suggested fix:
Correct the soundex function.

Thank you for a bug report. Verified just as described.

P.S. On Oracle, for example, one will get:

SQL> select soundex('Mannon'), soundex('Mnnn') from dual;

SOUN SOUN
---- ----
M550 M000

http://dev.mysql.com/doc/refman/5.0/en/string-functions.html#function_soundex
"
  Note: This function implements the original Soundex algorithm,
  not the more popular enhanced version (also described by D. Knuth).
  The difference is that original version discards vowels first and
  duplicates second, whereas the enhanced version discards duplicates
  first and vowels second.
"

Please implement a second version of soundex that is compatible with 99% of the other implementations, not the obsolete and deeply flawed original version.

Sorry, if not a bug formally, this is a reasonable feature request anyway.