QLocale::NarrowFormat is a format for cases when space is very limited. So the docs read. However, for some locales, it includes seconds. As other locales evade them, I assume the seconds are non-essential and should be omitted everywhere when the format is used.

Namely, the following languages return seconds in the Narrow format:

• C
• akk (Akkadian)
• ase (American Sign Language)
• egy (Ancient Egyptian)
• grc (Ancient Greek)
• arc (Aramaic)
• av (Avaric)
• ae (Avestan)
• ay (Aymara)
• ban (Balinese)
• bax (Bamun)
• bbc (Batak Toba)
• bi (Bislama)
• bug (Buginese)
• ch (Chamorro)
• cr (Cree)
• fj (Fijian)
• got (Gothic)
• hz (Herero)
• ho (Hiri Motu)
• inh (Ingush)
• ik (Inupiaq)
• kr (Kanuri)
• kv (Komi)
• kg (Kongo)
• kfo (Koro)
• kj (Kuanyama)
• lez (Lezghian)
• li (Limburgish)
• lzh (Literary Chinese)
• man (Mandingo)
• mh (Marshallese)
• men (Mende)
• na (Nauru)
• ng (Ndonga)
• new (Newari)
• oj (Ojibwa)
• sga (Old Irish)
• non (Old Norse)
• peo (Old Persian)
• pal (Pahlavi)
• pau (Palauan)
• pi (Pali)
• phn (Phoenician)
• sm (Samoan)
• saz (Saurashtra)
• ty (Tahitian)
• tkl (Tokelau)
• tvl (Tuvalu)
• uga (Ugaritic)
• ik (Inupiaq)
• kj (Kuanyama)

The behavior is neither consistent nor optimal for a format named Narrow.

As for the Documentation part, I'd like the docs to read what parts of the time representation a user will see in their locale. Not how, but what.

Currently, the docs read the following:

• LongFormat means a longer format,
• ShortFormat means a shorter format.

This explains next to nothing. Will there be TZ info? Seconds? Milliseconds? No clue. I have to test all locales to see.

Should I create a separate issue for this?