Re: [PATCH] cy_GB/en_GB: set am/pm in times

[Petr Baudis <pasky at ucw dot cz>]

  Hi!

  I think it is significantly preferable to base locale changes on
referenced material - ideally an official spec, though often these
are inaccessible due to high access costs, but at least multiple
government or other widely used websites (e.g. newspapers, popular
online portals).

  This should be usually easy to find in case the change is
non-controversial and has less sampling error than personal
experience.  It is not about personal distrust of anyone, but
as people have pointed out in this thread, personal background,
(era of) education etc. can influence this in less clear-cut cases.
Also, if someone wants to change a locale we are less familiar with
than en_GB, they will usually bring along those who agree with the
change, but users who prefer the original form are less likely to
notice.

On Thu, Apr 12, 2012 at 01:03:41AM -0400, Carlos O'Donell wrote:
> I've CC'd the authors of the locales for comment. I'm also including
> Petr Baudis whom I consider our locale expert.

  Oh. :-) I would just like to note that my availability is rather
unpredictable so please don't let anything block on me (at least not
for too long).

> > diff --git a/localedata/locales/cy_GB b/localedata/locales/cy_GB
> > index 9615c3f..0415e78 100644
> > --- a/localedata/locales/cy_GB
> > +++ b/localedata/locales/cy_GB
> > @@ -248,8 +248,11 @@ mon ? ? ? ? "<U0049><U006F><U006E><U0061><U0077><U0072>";/
> > ?d_t_fmt ? ? "<U0044><U0079><U0064><U0064><U0020><U0025><U0041><U0020><U0025><U0064><U0020><U006d><U0069><U0073><U0020><U0025><U0042><U0020><U0025><U0059><U0020><U0025><U0054><U0020><U0025><U005A>"
> > ?d_fmt ? ? ? "<U0025><U0064><U002E><U0025><U006D><U002E><U0025><U0079>"
> > ?t_fmt ? ? ? "<U0025><U0054>"
> > -am_pm ? ? ? "";""
> > -t_fmt_ampm ?""
> > +am_pm ? ? ? "<U0041><U004D>";"<U0050><U004D>"
> > +t_fmt_ampm ?"<U0025><U006C><U003A><U0025><U004D><U003A><U0025><U0053><U0020><U0025><U0050><U0020><U0025><U005A>"
> > +date_fmt ? ?"<U0025><U0061><U0020><U0025><U0062><U0020><U0025><U0065>/
> > +<U0020><U0025><U0048><U003A><U0025><U004D><U003A><U0025><U0053><U0020>/
> > +<U0025><U005A><U0020><U0025><U0059>"
> 
> "AM";"PM"
> "%l:%M:%S %P %Z"
> "%a %b %e/
> + %H:%M:%S /
> +%Z %Y"
> 
> The problem I see here is that both the UK Parliament and the UK
> Government website use `%A %e %B' (Wednesday 14 Apil) and never `%a %b
> %e' (Wed Apr 14). Even `The Sun' uses `%a %e %b', which is just a
> shortening of `%A %e %B'. In addition I see uses of `am' and 'pm'
> lowercase, and `%H.%M' not `%H:%M', on government documents (I'm
> willing to ignore these as government quirks).
> 
> How was this format decided?

  Thanks a lot for the research, Carlos! Do you still have the exact
URLs handy where you have seen the various usages?

  Let me sum up some of the references and discussion.

  am_pm "am;pm"

	In sync with all the online sources and confirmed by discussion
	with en_GB natives.

  t_fmt_ampm "%l:%M:%S %P %Z"

	So shall we use %l:%M:%S or %l.%M:S (i.e. hh:mm:ss or hh.mm:ss)?
	It seems that available online references suggest the latter
	but discussion suggests the former, though rather implicitly.
	I think more online sources would be good regarding this.

  date_fmt "%a %b %e %H:%M:%S %Z %Y"

	Both discussion and online sources seem to agree that %a %b %e
	is pretty much the only format _not_ used and that it does not
	matter too much. Online sources suggest "%a %e %b" and there is
	no agreement on short/long form. I would go with "%a %e %b"
	purely on the grounds that it is closest to the original patch,
	but if anyone has other suggestions...

-- 
				Petr "Pasky" Baudis
	Smart data structures and dumb code works a lot better
	than the other way around.  -- Eric S. Raymond