From 7b53dca6b55aeb2b935de1009d6d61df47b35be4 Mon Sep 17 00:00:00 2001 From: Alan Liu Date: Tue, 31 Jul 2001 18:28:23 +0000 Subject: [PATCH] ICU-1076 edit documentation X-SVN-Rev: 5400 --- icu4c/source/i18n/unicode/unifilt.h | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/icu4c/source/i18n/unicode/unifilt.h b/icu4c/source/i18n/unicode/unifilt.h index a18e1dd0e5..d4aee9c2d9 100644 --- a/icu4c/source/i18n/unicode/unifilt.h +++ b/icu4c/source/i18n/unicode/unifilt.h @@ -17,6 +17,21 @@ * Transliterator} to only process selected characters through a * transformation. * + *

Note: UnicodeFilter currently stubs out two pure virtual methods + * of its base class, UnicodeMatcher. These methods are toPattern() + * and matchesIndexValue(). This is done so that filter classes that + * are not actually used as matchers -- specifically, those in the + * UnicodeFilterLogic component, and those in tests -- can continue to + * work without defining these methods. As long as a filter is not + * used in an RBT during real transliteration, these methods will not + * be called. However, this breaks the UnicodeMatcher base class + * protocol, and it is not a correct solution. + * + *

In the future we may revisit the UnicodeMatcher / UnicodeFilter + * hierarchy and either redesign it, or simply remove the stubs in + * UnicodeFilter and force subclasses to implement the full + * UnicodeMatcher protocol. + * * @see UnicodeFilterLogic * @draft */