Repeat keys in cldr-json 39
General
Trac Info
General
Trac Info
Description
is duplicated by
relates to
Activity
Show:
Steven R. Loomis October 26, 2021 at 6:51 PM
note that the datetimeSkeleton work was done in https://unicode-org.atlassian.net/browse/CLDR-15113, this ticket is for the general fix.
Shane Carr October 20, 2021 at 4:54 PM
This is blocking for ICU4X, for reasons listed in CLDR-51113
Steven R. Loomis October 20, 2021 at 4:32 PM
See: CLDR-15113
Steven R. Loomis October 6, 2021 at 6:21 PMEdited
~running out of time in v40 to fix.~
Steven R. Loomis May 10, 2021 at 6:31 PM
@Konstantin Haase Yeah, that’s really broken, thanks for the report!
Need to run a linter on the output data. Can you recommend one? I suppose a round-trip might be sufficient, which could use GSon/Guava in Java.
Need to add this to the known issues on the release page.
Probably need to rewrite the output processing from a “stream” format (Attempt to calculate the JSON file in place!) for JSON to an “object” format (create a tree in memory and then serialize).
The new cldr-json export for CLDR v39 now has repeated keys. Example: https://github.com/unicode-org/cldr-json/blob/39.0.0/cldr-json/cldr-core/supplemental/grammaticalFeatures.json#L362-L391
While this is technically valid JSON, I am not aware of any JSON implementation that allows easy access to all these entries. Most notably, ECMAScript will only keep the last entry.
This may also impact other files, grammaticalFeatures.json was the one causing an exception for me.