From: Paul Lutus <nospam@...>
Date: 2006-11-28T04:10:17+09:00
Subject: Re: [OT] Re: Timestamp

Pit Capitain wrote:

> Paul Lutus schrieb:
>> dblack@wobblini.net wrote:
>>> Actually the whole thing (44,12 or 23,555 or whatever) is one float,
>>> using a comma separator instead of a dot.  The comma is the
>>> "preferred" separator in this context in the ISO 8601 standard, so I
>>> imagine that's being followed here.
>> 
>> Not that I have any doubt about your claim, but I find it hard to believe
>> that an international standard would specify a locale-specific radix
>> character.
> 
> Paul, what would you think to be a better separator, and why?

Anything but a separator that depends on locale. A separate field, for
example, one separated by a tab or any other character. Something like
that.

The database designers have obviously solved this problem. Their plain-text
databases are internationally acceptable because they choose field
separators they avoid locale-specific properties.

The remainder of the information provided doesn't suffer from
local-specificity (hh:mm:ss). A date might pose a problem, but many
databases have a default of yyyy-mm-dd (for plain-text database formats)
which seems to be acceptable to everyone.

When I first saw the data sample I assumed the comma was a field separator,
which would have made the three digits to the right a milliseconds value.
But ... no such luck.

To a European, "hh:mm:ss,nnn" might mean seconds and fractions thereof. To
an American, the ",nnn" part appears to be milliseconds. This is an easily
avoidable confusion.

-- 
Paul Lutus
http://www.arachnoid.com