1--- 2title: Direct Modifications to CLDR Data 3--- 4 5# Direct Modifications to CLDR Data 6 7*See also: Bulk Import of XML Data.* 8 9### 1\. Verifying changes 10 11Please check that your changes don't cause problems. A minimal test is to run ConsoleCheckCLDR with the following parameter: 12 13\-f(en) 14 15This will run the checks on en: you can substitute other locales to check them also (It is a regular expression, so -f(en.\*|fr.\*) will do all English and French locales). 16 17I recommend also using the following options, to show opened files, and increase memory (some tests require that). 18 19\-Dfile.encoding=UTF-8 -DSHOW\_FILES -Xmx512M 20 21An example of where a DTD broke, the invalid XML: 22 23\<!ELEMENT commonlyUsed (true | false ) #IMPLIED \>. 24 25I changed to \<!ELEMENT commonlyUsed EMPTY \> to get it to function; other changes might be necessary. 26 27### 2\. Explicit defaults 28 29Don't use them, since they cause the XML to be fluffed up, and may interfere with the inheritance unless you make other modifications. 30 31\<!ATTLIST fields casing ( titlecase-words | titlecase-firstword | 32 33lowercase-words | mixed ) "mixed" \> 34 35\=> 36 37\<!ATTLIST fields casing ( titlecase-words | titlecase-firstword | 38 39lowercase-words | mixed ) #IMPLIED \> 40 41Instead, the default should be documented in the spec 42 43### 3\. Mixing meanings. 44 45Attribute and element names should be unique, unless they have the same meaning across containing elements, and same substructure. This is a hard-and-fast rule for elements. For attributes, it is better to have unique names (as we've found by bitter experience) where possible. It is \*required\* when the attribute is distinguishing for one element and not for another. 46 47So the following is ok, but would be better if one of the attribute values were changed. 48 49\<!ATTLIST standard casing (verbatim) #IMPLIED \> 50 51\<!ATTLIST fields casing ( titlecase-words | titlecase-firstword | lowercase-words | mixed ) #IMPLIED \> 52 53