tag:blogger.com,1999:blog-273593670040001243.post1564123885498545413..comments2008-05-25T09:51:36.250-07:00Daniel Ehrenberghttp://www.blogger.com/profile/00902922561603041049noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-273593670040001243.post-38023383359237482452008-05-25T03:57:00.000-07:002008-05-25T03:57:00.000-07:00I've used:<BR/> http://www.unicode.org/Public/5.1.0/ucd/NormalizationTest.txt<BR/>and<BR/> http://www.unicode.org/Public/5.1.0/ucd/NormalizationCorrections.txt<BR/><BR/>To test my normalization implementation. I'm pretty sure that when those tests succeed your implementation is up to par. For grapheme cluster support I defined tests similar to the normalization tests. For added confidence you can always test against another implementation.Manfredhttp://www.blogger.com/profile/11350159894668652569noreply@blogger.comtag:blogger.com,1999:blog-273593670040001243.post-3114084686545681802008-05-23T16:08:00.000-07:002008-05-23T16:08:00.000-07:00Worth knowing: Windows' Unicode collation support isn't even always self-consistent, in that a < b, b < c, and c < a for some values of a, b and c. Random data => bad for testing Unicode sorts.Barry Kellyhttp://www.blogger.com/profile/10559947643606684495noreply@blogger.comtag:blogger.com,1999:blog-273593670040001243.post-46944093052972802572008-05-23T13:47:00.000-07:002008-05-23T13:47:00.000-07:00Dan, the link to the PDF survey is broken.Slava Pestovhttp://www.blogger.com/profile/02768382790667979877noreply@blogger.com