Corrigendum #2: Yod with Hiriq Normalization
Corrigendum |
Effective Date |
Applicable Versions |
Fixed Version |
Result Documented In: |
Corrigendum #2: Yod with Hiriq Normalization |
2001-Jan-31 [86-M20] |
3.0.0 and 3.0.1 |
3.1.0 2001-March |
CompositionExclusions.txt |
In the production of the normalization tables for Unicode 3.0, the
character U FB1D HEBREW LETTER YOD WITH HIRIQ was mistakenly omitted
from
Composition Exclusions. During the public review period, this
mistake was reported, but the report was misinterpreted and thus
overlooked. This corrigendum corrects that omission.
Add the following entry to CompositionExclusions.txt in the Scripts Specifics section of that data file:
FB1D # HEBREW LETTER YOD WITH HIRIQ
This corrigendum does affect backwards compatibility of normalization
forms NFKC and NFC for strings containing this character. Text containing this character that is in normalization form NFKC or NFC as defined in Unicode 3.0 is no longer in that normalization form after the application of this corrigendum. It is recommended that all implementations of those normalization forms upgrade to the Unicode 3.1 data tables (or later), to ensure interoperability with later versions of the standard.
Background
The reasons for issuing this corrigendum are enumerated below.
- The omission had been reported during the public review period for
Unicode 3.0.
- There were no normative references to Unicode 3.0 Normalization
from our liaison organizations (particularly IETF and W3C), although
normative references are expected soon after Unicode 3.1.
- YOD WITH HIRIQ is one of a class of characters ("marked" Hebrew
presentation forms within the range U FB1D .. U FB4E) that were to be
handled all in the same way, during all review and discussion of
Normalization in the UTC. The other characters in this class were
uniformly included in Composition Exclusions.
- YOD WITH HIRIQ is a very rare character. The amount of existing
data containing it is infinitesimal as a proportion of all
computerized text. Even if it takes some time for implementations to
upgrade, this change should pose no significant
backwards-compatibility issue in practice.