L2/04-151

Source: Mark Davis
Subject: Unicode Property Invariants
Date: Thu, 6 May 2004

After we found the Decimal problem (http://www.unicode.org/errata/,
2004-April-22) I extended my internal consistency tools to be more flexible.
They are now data-driven, by the file found at

http://oss.software.ibm.com/cvs/icu4j/~checkout~/unicodetools/com/ibm/text/UCD/UnicodeInvariants.txt

In particular, that Decimal error is now found by the following line:

$General_Category:Decimal_Number = $Numeric_Type:Decimal

I have added a few other cases, like:

a. The following two check that Unicode identifiers are always backwards
compatible:

$ID_Start ⊇ $×ID_Start
$ID_Continue ⊇ $×ID_Continue

b. The following checks that Whitespace always includes the general categories
of Space, Paragraph, and Line Separator:

$Whitespace ⊃ [$GC:Zs $GC:Zp $GC:Zl]

c. And I added a number that test the Derivations, such as that the Math
characters are the union of the Sm characters and the Other_Math characters:

$Math = [$GC:Sm $Other_Math]
...

Now that it is data driven, it is very easy to add other invariants, so I'd
request that other people consider which other invariants make sense to test
for.