This file contains Label Generation Rules (LGR) for the Latin script as would be appropriate for the Root zone. The format of this file follows [RFC 7940].
According to Section 5, "Repertoire" in [Proposal-Latin]. The repertoire contains the 193 code points needed to write hundreds of languages in the Latin script. An additional 7 code points are available as part of 21 explicitly defined combining sequences which form part of the repertoire. One additional in-script sequence is defined for use in variants. For a total of 215 repertoire elements.
The repertoire is based on [MSR-4], which is a subset of Unicode 6.3 [Unicode 6.3].
Each code point is tagged with the script or scripts that the code point is used with. For each repertoire element, one or more references document sufficient justification for inclusion in the repertoire; see "References" below. Comments identify the languages using it along with their [EGIDS] levels.
The Latin script is a major writing system of the world today, and the most widely used in terms of the number of languages and speakers, with circa 70% of the world’s readers and writers making use of this script. The list of languages taken into consideration contains relevant data for 455 languages using Latin script. More details in Section 3, "Background on Script and Principal Languages Using It" of the [Proposal-Latin].
According to Section 6, "Variants", in [Proposal-Latin], this LGR defines the principles and methodology for developing both in-script and cross-scripts variants. The variant sets are defined in Section 6.3, "Variant Sets"" and detailed analysis can be found in Appendix D of [Proposal-Latin]. See also Section 6.4, "Other Considerations for Variant Analysis".
Variant Disposition: With the exception discussed below, all variants are of type "blocked", making labels that differ only by these variants mutually exclusive: whichever label containing either of these variants is chosen earlier, the other one equivalent variant label should be blocked. There is no preference among these variants.
The specification of variants in the Root Zone LGR follows the guidelines in [RFC 8228].
In the case of U+00DF Latin Small Letter Sharp S used in German, the LGR includes the code point together with a variant relationship with the sequence of two letters "ss" (U+0073 U+0073), as described in Section 6.4.2, "IDNA2003 Compatibility". To reduce the number of allocatable labels in case the applied-for label contains multiple sharp s letters, the LGR defines special variant types and actions so that only two allocatable variants are allowed for each label: the label as applied-for (original label), and the label with only the sequence ss. Note that this restriction is independent of the restriction on variants of dotless i, see below, so that a total of up to four allocatable labels may exist.
Overlapped Variant Sequence: Both “ss” and “s” coexist in the repertoire and “s” has variant relationships on its own. These variants thus overlap: making the variant set well-behaved for index variant calculation requires that the sequence "ss" also be given variants to all permutations of variants for the letter s followed by itself, as well as all transitive variants due to other variants for U+00DF. See also Section 6, [TBD], "[TBD]" in [Proposal-Latin].
The LGR defines a variant between U+0131 Dotless i and U+0069 such that labels containing the any dotless i have allocatable variant labels that only have U+0069, as described in Section 6.4.2, "IDNA2003 Compatibility". To reduce the number of allocatable labels in case the applied-for label contains multiple instances of dotless i, the LGR defines special variant types and actions so that only two allocatable variants are allowed for each label: the label as applied-for (original label), and the label with only dotted letters i. Note that this restriction is independent of the restriction on variants of sharp s, see above, so that a total of up to four allocatable labels may exist.
U+00DF Sharp S has been given the reflexive variant type "r-eszett" and U+0131 Dotless i has been given the reflexive variant of type "r-dotless". (By convention, the prefix “r-“ marks a type used in a reflexive variant mapping, that is, it represents an instance of the original code point at that location in a variant label, see Section 5.3.4 in [RFC 7940].)
The variant mapping from U+00DF Sharp S is to "ss" is of type "eszett-to-ss", while the variant type for the mapping from "ss" to Sharp S is "blocked". The variant mapping from U+0131 Dotless i to "i" is of type "dotted", while the variant type for the mapping from "i" to Dotless i is "blocked". Special <action> elements defined for this LGR use these types to ensure the following restrictions. (See also "Latin-specific Actions" below.)
To limit the number of allocatable variants, a label is only allocatable when the variants for any U+00DF in the label are all "ss"; or the variants for any Dotless i are all "i"; or both; or when it is the original label as applied for. As a result of these actions, any label may result in at most four allocatable labels including the original label as applied-for, however, if it is applied for without Sharp S or Dotless i it will not result in any allocatable variant labels.
The Latin script is strongly related to the Greek and Cyrillic scripts, with a more distant relation to the Armenian script. These relationships give rise to shared forms that necessitate cross-script variants. A number of letter forms like "i", "o" and "c" are rather generic in appearance and thus give rise to cross-script variant relations with otherwise unrelated scripts. Because variants sets, including cross-script variant sets must be symmetric and transitive some cross-script variants, in particular Greek variants, additionally impose in-script variants for certain Latin letters with diacritics.
For more details, see Section 6.[TBD] "[TBD]" in [Proposal-Latin]).
This LGR inherits additional cross-script variants by integration; they may not be listed here unless they result in in-script variants. However, for consistency, the variant mappings among the related scripts Armenian, Cyrillic, Greek and Latin are listed in full. See the merged, Common LGR [RZ-LGR-5] for details of all applicable cross-script variants, including any not listed here; always use the Common LGR for determining cross-script collisions of labels.
Overlapped sequence: one sequence with a variant "ss" is overlapped with a code point "s" that has a variant of its own. To ensure that the sets of variant labels are well-behaved, additional variant mappings, including an out-of-repertoire sequence had to be defined: U+0455 U+0455.
The LGR defines no character classes.
The LGR includes the set of required default WLE rules and actions applicable to the Root Zone and defined in [MSR-4]. They are marked with ⍟.
This LGR does not define Latin-specific Whole Label Evaluation Rules.
The LGR includes the set of required default actions applicable to the Root Zone and defined in [MSR-4]. They are marked with ⍟.
The LGR contains additional Latin-specific actions as described in Section 6 of [Proposal-Latin]. These resolve the extended set of variant types into a disposition for variant labels of either "allocatable" or "blocked". Latin-specific actions that are triggered by the LGR-specific variant types described above limit the "allocatable" variant labels to those containing only "ss" or dotted "i" variants or both, while disallowing mixed use of "ss" and "ß" or Dotless i and "i" respectively, except as in the original applied-for label. To account for original labels, reflexive variant mappings with an "r-" prefix are used. (See [RFC 7940]).
Note that variant mapping types are not symmetric: they depend on which code point is considered the source or the target in a given mapping. As specified in [RFC 7940], mapping types are evaluated for each permutation of a label and its variants, with code points that are unchanged in a given label given the type of their "reflexive" mapping. The actions finally evaluate the collected set of mapping types and resolve them into one of two dispositions for the variant label. Per [RFC 7940] actions are always applied one after the other, and the evaluation stops at the first action that assigns a disposition to a given label.
For more information on how to assign a variant label disposition under this LGR, see Section [TBD], "[TBD]" in [Proposal-Latin]. The specification of variants in the LGR follows the guidelines in [RFC 8228].
The Root Zone LGR for the Latin script was developed by the Latin Generation Panel. For additional detail on methodology and contributors see Sections 4 and 8 in [Proposal-Latin].
Following general references are cited in this document:
For references consulted particularly in designing the repertoire for the Latin script for the Root Zone please see details in the Table of References below. References [0] and up refer to the Unicode Standard versions in which corresponding code points were initially encoded. References [101] and up correspond to a source given in [Proposal-Latin] for justifying the inclusion of the corresponding code points. Entries in the table may have multiple source reference values.
]]>