Re: [icu-support] Collation rules error

[Sarah B. <sbe...@ta...>]

Thank you Markus!!  This totally worked!!

From: Markus Scherer [mailto:mar...@gm...]
Sent: Wednesday, January 27, 2016 2:28 PM
To: ICU support mailing list <icu...@li...>
Subject: Re: [icu-support] Collation rules error

On Wed, Jan 27, 2016 at 1:39 PM, Sarah Berrier <sbe...@ta...<mailto:sbe...@ta...>> wrote:
Thanks for all of your help so far!!  I was pretty excited about the suggestion below Markus!

Unfortunately &\u0000=*0-9 doesn’t seem to succeed in making the digits ignorable.

If I try either of these rule strings:
"&\\u0000='0'='1'='2'='3'='4'='5'='6'='7'='8'='9'"
"&\\u0000=*0-9"
And run:
collator.Compare("a1bc", "abc")
I get 1 as a return value.

The collation rule parser does not unescape rules. In Java, don't double the backslash. In C++, find a way to get an actual character U+0000 in there. Modern C++ does support \u0000 in UTF-8 or UTF-16 string literals.

If I try this rule strings
"&'0'='1'='2'='3'='4'='5'='6'='7'='8'='9'=\\u0000"
And run:
collator.Compare("a1bc", "abc")
I get -1 as a return value.

That would be the opposite, making an igorable character non-ignorable, and equal to digit 0. Except that you are actually making the sequence \ u 0 0 0 0 sort like one digit 0.

markus