L2/07-335

Date: Wed, 3 Oct 2007
From: Mark Davis
Subject: Linebreak Test Files

==========

For most boundary specifications that we have, we have a chart and a test
file, both generated semi-automatically. That is, from a set of sample
strings the values are generated. For example, see

http://unicode.org/Public/UNIDATA/auxiliary/WordBreakTest.html
http://unicode.org/Public/UNIDATA/auxiliary/WordBreakTest.txt

The test file, in particular, allows people to verify that a stock
implementation (one with no tailoring) conforms to the specifications that
we have.

With LineBreak, we are missing these files. We've found it quite useful in
ICU to be able to use the test files, and so I suggest that we generate them
for LineBreak as well, putting them in the auxiliary directory. Attached are
two sample documents for the registry. (These were built with a previous
version of the data and UAX (5.0.1), but serve as illustrations.)

-- 
Mark