string partial comparison

Question

Possible Duplicate:
Why does string.Compare seem to handle accented characters inconsistently?

I have the following code

var s1 = "ABzzzzz2";
var s2 = "äbzzzzz1";

var cmp = StringComparison.InvariantCultureIgnoreCase;

Console.WriteLine(string.Compare(s1, 0, s2, 0, 7, cmp)); //prints -1
Console.WriteLine(string.Compare(s1, 0, s2, 0, 8, cmp)); //prints 1

How could it be that part of the first string is less than part of the second, while the whole first string is greater than the whole second one?
I've tested it on x64, .net 2.0, 3.5, 4.0

It seems to me that the numbers at the end take precedence in defining the order. — John Dvorak, Nov 08 '12 at 14:02
You can simplify your sample: `var s1 = "a2"; var s2 = "ä1";` Btw, the Umlaut matters. — Tim Schmelter, Nov 08 '12 at 14:05
Perhaps the amount of ignorance is reduced if the strings would be equal otherwise? — John Dvorak, Nov 08 '12 at 14:06
I assume that you can find the answer here: http://stackoverflow.com/q/1371813/284240 — Tim Schmelter, Nov 08 '12 at 14:08
@TimSchmelter Good find! Key sentence from the accepted answer there: *Accent differences are typically ignored, **if there are any differences in the base letters**.* — anton.burger, Nov 08 '12 at 14:25
@Tim Schmelter: can you please write an answer so I could accept it? — alpha-mouse, Nov 08 '12 at 14:27
It's also worth noting that in unicode, there is a difference between any letter with an accent on it, and that same letter with same accent at some other code point. — MarioDS, Nov 08 '12 at 15:24

score 1 · Answer 1 · edited May 23 '17 at 10:24

My theory is that the algorithm first normalizes strings and then does the comparison. According to this "äbzzzzz1" normalized as "abzzzzz1". First comparison in normalized form results equality but returning 0 would be inccorent since the actual strings are not equal. So it reverts to ordinal comparison and results -1.

In the second case, after normalization, it is clear that "abzzzzz2" is greater than "abzzzzz1" so the result is 1.

This approach also explains cases mentioned in this question For normalization details check MSDN page

string partial comparison

1 Answers1