Re: [util-linux] cal doesn't honor first_weekday value from locale

From: Christian Rose (menthos_at_menthos.com)
Date: 2001-11-26 16:25:43

Andries.Brouwer@cwi.nl wrote:
> > You could help by pointing out for which locales you think
> 
> Ha! Are you saying that you trust my opinion for all countries
> except Sweden? :-)

No, I don't trust you on anything like this. Your version of the truth
for the situation in Sweden should give enough reason to not do so. :(
But still, if you give a hint what locales you think are wrong, this
could help investigating these things further by asking people who
actually live in the country and know about what's used and what's not
and why, and ask them if the current value is wrong.


> In fact I think that in many countries the usage varies quite
> a lot, partly randomly, partly depending on complicated factors.

This is certainly true, some countries have a recommended standard, but
one or more different ways of doing things used in practice. In these
cases it is not at all entirely clear what the value should be, and one
can argue for either way. Personally, I think that for locales to be
useful, they have to represent the most common way of doing things in
that country, since that behavior is what most users would expect.
Nevertheless, if there are ways of doing things that are both common,
then the behavior that most closely matches the recommended standard in
the country is probably best. But in all these cases, the problem of
selecting the proper behavior as a setting in the locale information is
not trivial. However, that doesn't mean that refusing to use this
information is the right thing to do (see below).


> Too complicated to have a simple variable in a locale.

Here I most certainly do not agree with you. You are forgetting that the
above scenario does not apply to every country, in some countries there
is an exact match of the recommended standard and what's actually used,
and there is no problem at all. The only problem is if you choose not to
honor the locale information and choose another hardcoded value by
default that happen to be completely wrong in that country, both
standards-wise and with respect to what's used.
Also, in general this situation may also apply to countries where one or
more behaviors are correct -- in these cases, either one of the values
that are correct may not match your hardcoded default, while either one
of the values used in the country (provided by glibc locale information)
would have been a better and more correct choice.

In general, I think that refusing to use locale information because the
value *might* be wrong in some cases, and ignoring that the information
most probably is correct in most cases (and thus certainly is better to
use than a hard-coded default that by the very nature of hardcodedness
is bound to be wrong in many more cases), is a very strange thing to do.
Even if you do not agree that the information already is correct for
most locales, you have to realize that locale information can get
corrected and also *has* been corrected over time, providing a means for
fixing the behavior of many applications at once. Recompiling util-linux
for every single locale however, just to change the hardcoded default,
isn't realistic at all.


Christian

Arkiv genererat av hypermail 2.1.1.