https://test.wikidata.org/wiki/Q22
When entering "1 January 23" as a date and setting precision to century or less formatting seems to be broken. "+00000000023-01-01T00:00:00Z" is shown instead.
Version: master
Severity: major
Whiteboard: u=dev c=backend p=0
That's actually a feature I recently introduced.
Formatting the year 123 as a century results in "1. century". That's a bit odd (you loose a lot of information) but kind of ok.
But what do you expect from the year 23? "0. century"? "0 million years ago"? I think this needs a fallback. My current fallback is to show the internal format. Other possible fallbacks:
Please decide. I suggest the last one.
(In reply to Thiemo Mättig from comment #1)
Formatting the year 123 as a century results in "1. century". That's a bit
odd (you loose a lot of information) but kind of ok.
Surely that's the second century, not first, unless you're talking about time deltas and not dates? (I don't know the context, seen this scroll by on IRC.)
(In reply to Bartosz Dziewoński from comment #2)
that's the second century
Correct. We just realized that's an other bug in the current implementation. The "2. century" can also be expressed as "1 century A.D.". And it depends on the language what you need for what precision. We usually say "1 million years" and not "2. age", but this may be different in other languages. So all these messages need two placeholders or what?
Not mentioning the 1990s. Is "199 decades A.D." the same as "1990s" and is that the same as "199. decade A.D."?
Oh wow, language support is fun.
Change 166622 had a related patch set uploaded by Thiemo Mättig (WMDE):
Display full year if precision is to high
Seems like it has been forgotten to set the ticket to "fixed" (at the end, it is just high/major anyway... ;) ). Since the change has been merged, I just verified its behaviour--no ISO string returned anymore.