Mailing List Archive

On Fri, 21 Feb 2003, Dan Koehl wrote:

> Im sorry, the letter å is to common in our language that it makes sence to
> replace it with code å, at least for new wikipedians, it has to be
> autmatically done, like any modern HTML editor would do.
>
> Othervise were back to making editing difficult, when it should be easy??
>
> Since ä and ö which is present in the German langauges is working OK, while
> å is the unique Swedish letter (not present in any other language?) I stille
> belive there is something with the script, and how it translates letters and
> code. Im pretty sure that somewehere in the code theres lines which makes ä
> and ö working, while å is lacking. Am I wrong?

Yes, you are wrong. The script is not doing anything, as I said before, it is
the browsers (and/or more specifically the editors on them) that are at fault.
As far as I understand it, these editors do not have å in their character set,
while they do have ä and ö. Note that a similar thing happened in Tomasz
Wegrzanowski's email. And the email surely won't go through Wikipedia's PHP
script, will it?

Another way to see that it is not caused by the PHP script, is that it happens
for some people but not all - the PHP script just sees the text that is given
as input, so if it causes problems with some users, those users must be
inputting a different text.

Andre

On Fri, 21 Feb 2003, Dan Koehl wrote:
> Since ä and ö which is present in the German langauges is working OK, while
> å is the unique Swedish letter (not present in any other language?) I stille
> belive there is something with the script, and how it translates letters and
> code. Im pretty sure that somewehere in the code theres lines which makes ä
> and ö working, while å is lacking. Am I wrong?

No, the wiki doesn't make any attempt to convert characters of input text
to html references. It does to a limited degree convert incoming URLs, so
URLs can come in either UTF-8 or the native character set. And outgoing
language links have named references turned into latin-1 bytes, and
numeric references turned into utf-8 bytes in the generated URL. But
except on the Esperanto wiki where we do some replacement of /[cghjsu]x/
into UTF-8 and vice-versa due to the limited availability of appropriate
keyboard layout drivers, text is not touched, but saved as given to us
by the submitting browser.

If you can point out a specific example, I can check the web server logs
and see what web browser is being used that might be doing this.

-- brion vibber (brion @ pobox.com)