collate

[Dag Henriksson]

I'm trying to compare strings written in Swedish using:

std::locale loc(""); // System locale is Swedish
loc(first, second);

But this does not give me the right return value.
for example,

loc(string("å"), string("ä"));

should yield true, since 'å' is less than 'ä' in Swedish, but it doesn't.

Is this a bug?

--
Dag Henriksson

[Bo Persson]

"Dag Henriksson" <da************@quidsoft.se> skrev i meddelandet
news:%2****************@TK2MSFTNGP11.phx.gbl...

I'm trying to compare strings written in Swedish using:

std::locale loc(""); // System locale is Swedish
loc(first, second);

But this does not give me the right return value.
for example,

loc(string("å"), string("ä"));

should yield true, since 'å' is less than 'ä' in Swedish, but it doesn't.

Is this a bug?

Well, not really, it just doesn't work.

The C++ standard doesn't say much what happens for letters outside of a-z. In Windows their positions even varies between different
character sets (try looking at them with the program
"Teckenuppsättning"). Most often 'ä' comes before 'å' (in an area
filled with varying sets of other accented glyphs that are not
"proper" letters).

Yes, I know that the C++ standard does not put any requirements on

any other facets than "C", but if an implementation can create a locale
named for example "Sweden_Swedish", shouldn't the collate facet in that locale then follow the rules of Swedish, or are you saying that this is impossible in Windows because the AscII of a letter depends of the
character set? Most lists I've seen of supported locales say what
charmap is used for each locale, for example ISO-8859-1 for Swedish, so this should not be a problem.
Ok, I change my statement to "it might work". I see that in the latest
library collate<char> indirectly calls the undocumented function
_Strcoll. Don't know what that means though.

If you say that it still doesn't work, I'm not the least surprised.
Bo Persson
bo**@telia.com


All book I have read suggests that it should be possible to compare
strings according to local rules using collate. See for example
http://www.research.att.com/~bs/3rd_loc.pdf
where Stroustrup writes:
"given the dk locale, we can construct a locale that reads and compares strings according to the rules of Danish (that give three extra vowels compared to English)"

But maybe I've misunderstood something...

--
Dag Henriksson