This simple example shows the issue I've run into, but I don't understand why...
I'm testing for the location of the first character that is either a lower or upper case letter, a single dash, or a period in a string parameter passed to me.
These two pattern matches appear to check the same thing, and yet run this code yourself and it will print a 0 then a 3:
PRINT PATINDEX ( '%[a-z,A-Z,-,.]%', '16-82')
PRINT PATINDEX ( '%[-,a-z,A-Z,.]%', '16-82')
I don't understand why it works only if the dash character is the first one we check for.
Is this a bug? Or working as designed and I missed something... I'm using SQL Server 2016, but I don't think that matters.
A dash within a character group may play either of the two roles:
[-abc]
[a-z]
.In your particular example, the character group [a-z,A-Z,-,.]
denotes the following:
a
to z
,
A
to Z
,
to ,
(i.e. just the comma again)..
In fact, you probably wanted to write [-a-zA-Z.]