[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#633837: libidn11: German ß (LATIN SMALL LETTER SHARP S) should no longer be converted to ss



On Thu, 14 Jul 2011 10:24:46 +0200
Simon Josefsson <simon@josefsson.org> wrote:

> Christian Hammers <chammers@netcologne.de> writes:
> 
> > Package: libidn11
> > Version: 1.15-2
> > Severity: normal
> >
> >
> > Quoting http://www.denic.de/en/domains/internationalized-domain-names/sharp-s.html
> > "Since 4 August 2010, the IDNAbis standard allows the Latin small letter sharp s – 
> > also known as "Eszett" or "sharp s" ("ß") – to be used as part of a domain name"
> >
> > The IDN library (1.15 as well as 1.22 from unstable) still converts it to "ss":
> >
> >     $ idn --quiet --idna-to-ascii baß.de
> >     bass.de
> >
> > Changing the behaviour will of course break backwards compatibility but
> > as the compatibility was broken by the IDNA standard itself, the library
> > should continue to follow the standard (at least in methods that have
> > "idna" in their name).
> 
> Hi Christian.  libidn.so implements the old IDNA standard, retroactively
> called IDNA2003.
> 
> The page above talks about IDNAbis, or usually called IDNA2008, which
> libidn.so and idn don't support.  IDNA2008 and IDNA2003 are not
> compatible.
> 
> The GNU Libidn project contains another library and tool, libidn2 and
> idn2, which implements the IDNA2008 algorithm.  It works like this:
> 
> jas@latte:~$ idn2  baß.de
> xn--ba-hia.de
> jas@latte:~$ 
> 
> So I believe your request should be re-categorised as 1) a request to
> package libidn2, and 2) modify any applications you are concerned with
> to support it.  If you just want to do the conversions on the command
> line, 1) will suffice.
> 
> I'll see if I can get some packaging up and running...

I would rather have libidn2 treated just as an update to libidn11 i.e.
package "idn (2.0)", "libidn-dev (2.0)" and "libidn20 (2.0)" that
replace the current /usr/bin/idn tool in the next Debian release.
(or libidn2 if the linker figures correctly that it is newer than 
libidn11)

bye,

-christian-




Reply to: