Equality comparer returned by CollationInfo.GetCollationInfo("Latin1_General_CI_AS") doesn't work as expected #22

Steps to reproduce:

1. Create a database with collation Latin1_General_CI_AS
2. Execute the following scripts in this database

create table unicode_test ([u_name1] nvarchar(10), [name1] varchar(10), [u_name2] nvarchar(10), [name2] varchar(10))
insert into unicode_test ([u_name1], [name1], [u_name2], [name2]) values (N'xxßmm', 'xxßmm', N'xxssmm', 'xxssmm');

select 'non_unicode_comparison' from [unicode_test] WHERE [name1] = [name2];
select 'unicode_comparison' from [unicode_test] WHERE [u_name1] = [u_name2]

Each select statement returns one row.
3) Create a sample application and add a reference to Microsoft.SqlServer.Management.SqlParser package (I tested v172.0.1). Then use this code to test the comparison:

var collation = CollationInfo.GetCollationInfo("Latin1_General_CI_AS");
var equalityComparer = collation.EqualityComparer;
            
var t1 = equalityComparer.Equals("ß", "SS");
var t2 = equalityComparer.Equals("ß", "ss");

Expected result: variables t1 and t2 are true.
Actual result: variables t1 and t2 are false.

The reason behind is that newer versions of .NET (>=5) switched to ICU-based comparison (where "ß" != "ss") and you didn't implement the workaround with CompareOptions.IgnoreNonSpace flag.

I totally understand that if this flag is used then the equality comparer will be exactly same as for Latin1_General_CI_AI collation, which is weird. But maybe someone can provide a detailed explanation why exactly you made a decision not to fix this bug and what is the recommended way of handling it?

Source issue