Re: ls default time style

[Markus Kuhn]

Bruno Haible wrote on 2001-12-11 12:19 UTC:
> > -rw-r--r--    1 bin     bin      2188323 12-03 00:00 fileutils-4.1.3.tar.gz
> > -rw-r--r--    1 bin     bin      1812537 2001-04-29  fileutils-4.1.tar.gz
> I was talking about the "12-03" line which means 3rd of December, not
> 12th of March. I repeat that it has never been seen in Germany.

Admittedly, the "12-03" does not even conform strictly to ISO 8601:2000,
which would require --12-03 if you omit the year (§5.2.1.3 d), as 12-03
stands for March of the year 2012 (or whatever the implied century is).
But I find that it is a rather obvious notation, as in German usage the
hyphen-separator implies already that you refer to a bigendian date,
whereas the dot-separator implies the litteendian version.

Could the date field to occupy 14 characters? In that case, I would
advocate the format "YY-MM-DD HH:MM" to be used everywhere. In practice,
POSIX implementations can't represent dates outside 1970-2038, therefore
the century is really redundant (use "stat" instead of "ls" in case of
doubt), and I find a single notation of constant precision to be both
easier to read for humans as well as easier to parse.

> > It also leads the way towards reducing the differences between
> > locales
> 
> Why should this be a goal? There are differences between the ways
> Germans and Americans write the dates, the currencies, the numbers.
> Internationalization means to make it possible for programs to follow
> the locale conventions, not the contrary.

Internationalization merely means to enable programs to be useful in
many countries. Localization is one way of achieving that, namely to
modify the bahaviour of programs to follow historically grown national
conventions. Standardization is an alternative, namely to find
conventions that are acceptable in many if not all countries. I advocate
the school of thought that standardization is in any respect far
superiour to localization. For some locale data groups, we have now
excellent solutions for the standardization approach:

LC_CTYPE:       ISO 10646/UTF-8
LC_TIME:        ISO 8601
LC_MEASUREMENT: ISO 1000
LC_NUMERIC:     ISO 31-0 with decimal dot

Also very good further candidates (except for a small number of
locales):

LC_PAPER:       ISO 216
LC_COLLATE:     ISO 14651

which leaves really only LC_MESSAGES and LC_MONETARY where there is an
actual need to localize instead of to standardize.

Markus

-- 
Markus G. Kuhn, Computer Laboratory, University of Cambridge, UK
Email: mkuhn at acm.org,  WWW: <http://www.cl.cam.ac.uk/~mgk25/>