http://bugzilla.opensuse.org/show_bug.cgi?id=1205236 http://bugzilla.opensuse.org/show_bug.cgi?id=1205236#c2 --- Comment #2 from Jan Engelhardt <jengelh@inai.de> --- ����� looks familiar. " * When the bytes of a UTF-8 encoded file are interpreted as indexes into the * cp1252 table or as indexes into the Unicode table, and the resulting * characters be stored as UTF-8 again, mojibake sequences like �����<U+009D> can * appear. " ���https://inai.de/2010/03/07 linking decode_mojibake_1252.c So it would appear ICU now emits a U+202F when being asked to generate the text representation of a timestamp. With my typography hat on, that seems a legit thing to want to do. origin https://github.com/unicode-org/icu commit 64b35481263ac4df37a28a9c549553ecc9710db2 -> origin https://github.com/unicode-org/cldr commit a83026ab8c8fa6ed88f1047c4d0c6089f88b7e5d -> https://github.com/unicode-org/cldr/pull/2001 The nodejs tests are.. flawed in the sense there is no one correct answer to how localized timestamp strings should look like. -- You are receiving this mail because: You are on the CC list for the bug.