659309 - Implement RFC 5891 (updated IDNA rfc)

Status: RESOLVED DUPLICATE of bug 479520

(Core :: Networking, defect)

Description

[tosh]

User-Agent:       Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_6_6; de-de) AppleWebKit/533.21.1 (KHTML, like Gecko) Version/5.0.5 Safari/533.21.1
Build Identifier: 

"Latin Small Letter Sharp S" in the new standard (IDNAbis), from 4. August 2010 does not work in the Morzilla Browser. 

My Web Adress is: www.kochstraße.de (www.xn--kochstrae-n1a.de)  is automatically delivered to
www.kochstrasse.de




Reproducible: Always

Steps to Reproduce:
1.Type in:  "www.kochstraße.de" with the German Letter "ß"
2.
3.

Actual Results:  
www.kochstrasse.de (wrong adress)

Expected Results:  
Forward to the correct location: www.kochstraße.de
 (www.xn--kochstrae-n1a.de)

Comment 1

[AndreiD[QA]]

Reproducible on:
Mozilla/5.0 (Windows NT 6.1; rv:6.0a1) Gecko/20110523 Firefox/6.0a1

*Note: It's not clear if this is a bug or not since it has the same behavior on other browsers as well, and moreover when I click-right on the website link (in this bug) and paste it in the location bar I get the address with double "s", and no "ß" letter.
Remaining Unconfirmed.

Comment 2

[Henrik Skupin [:whimboo][⌚️UTC+1]]

I'm not sure either. CC'ing some networking guys to get an explanation.

Comment 3

[Boris Zbarsky [:bzbarsky]]

The IDN normalization code very purposefully converts eszet to "ss" during nameprep.  

In particular, RFC 3491 defines nameprep as using the B.1 and B.2 tables from RFC 3454.  RFC 3454 table B.2 says that the codepoint U+00DF is mapped to the two-codepoint sequence U+0073 U+0073 (so lowercase eszet maps to "ss").

RFC 3490, which is what we implement, requires that IDNA input strings be processed with nameprep.  See RFC 3490 section 4.1 step 2.

RFC 5891 appears to be the RFC that's referred to in comment 0.  The steps in RFC 5891 section 5 appear to not use nameprep at all, as far as I can tell.

Simon, do you know anything about this stuff?  In particular, how web-compatible RFC 5891 would be (given that browsers all seem to implement RFC 3490 and the processing models seem to be very different)?

Note that RFC 5891 seems to also call for case-sensitive compares on U-labels, btw.

Comment 4

[Jo Hermans]

This is a duplicate of bug 479520.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Indeed.