From: Kenneth Whistler (kenw@sybase.com)
Date: Mon Feb 08 2010 - 16:42:14 CST
Denis asked:
> PS: I just discovered parts 2 and 3! So, 3 more questions:
>
> -7- Is part2 "Canonical Order Test" organised the same way
(source;NFC;NFD;NFKC;NFKD)?
Yes.
> -8- Is is supposed to be used differently?
No.
> Else, why is it apart?
Because those test cases were generated separately, to address
particular edge cases for canonical ordering.
> (anyway, proper ordering is supposed to be ensured in all cases, no?)
Yes.
> -9- What is part3 "PRI #29 Test"?
Specific tests to address the edge cases discussed in:
http://www.unicode.org/review/pr-29.html
>
> What about a table of contents, when useful, and a few
> words about intended use of each part?
You'll have to address that to the author(s) of the test
cases incorporated in the file.
Basically, however, NormalizationTest.txt is designed as
a machine-readable file, for a test engine testing an
implementation of Unicode Normalization. It isn't supposed
to be something that one reads through manually, trying
to figure out what each test case was created for.
An implementation should pass *all* of the test cases,
and if it doesn't, it has a bug (or bugs) in it.
Of course, if your implementation of Unicode Normalization
has bugs in it, it might be helpful in figuring out those
bugs to have more information about the details of the
edge cases represented by any failing test case. But the
data for the test cases was not assembled with the
goal of it being a tutorial about normalization testing.
Rather, it just is fairly exhaustive test data checking on
the kinds of things that could fail in an implementation.
--Ken
>
> Denis
> ________________________________
>
> la vita e estrany
>
> http://spir.wikidot.com/
>
>
This archive was generated by hypermail 2.1.5 : Mon Feb 08 2010 - 16:47:08 CST