55286 - Composer: links to named anchor's with "umlaute" like ä ö ü won't work

Status: VERIFIED FIXED

(Core :: DOM: Editor, defect, P3)

Description

[staublis]

reproduce: start composer insert an named anchor with a name wich contains
"umlaute" like "ä" "ö" or "ü".  now insert a link to that anchor. view page with
browser. ==> link to the anchor does not work.

the problem seems to be in the link, caus it replaces "umlaute"
egg: "ö" gets replaces with "Ã?¶" and the browser doesn't recognice this.


mfg
staubli

(sorry for all my silly composer bugs, but at the moment im working with
composer to creat a page for my boss, he wanted frontpage, but i didnt agree :)

Comment 1

[Asa Dotzler [:asa]]

over to editor where Composer bugs live.

Comment 2

[Simon Fraser [no longer active]]

-> brade. More entity encoding fun.

Comment 3

[Kathleen :Brade]

What character encoding are you using?  Can you attach a sample file with this 
problem?

p.s.  Please keep filing bugs as you see issues that we need to address!

Comment 4

[Kathleen :Brade]

This appears to be some kind of weirdness in the Link dialog.  Unfortunately, I 
don't think we'll be able to come up with a safe, easy fix before rtm'ing.  
Setting to Future, reassign to cmanske since he wrote the Link dialog.

Workarounds include: don't open link dialog, reselect the anchor after opening 
link dialog, click cancel to link dialog, correct the link / anchor in the HTML 
Source view.

Comment 5

[Simon Fraser [no longer active]]

Remving rtm from keywords

Comment 6

[rubydoo123]

rtm - this one and leave at future

Comment 7

[Charles Manske]

The foreign characters seem ok (HTML source shows proper escape codes) in the
document, in the attribute value text, such as in the <a name="...
I could create a link to a named anchor with an umlauted "a", save the file, and
the link did jump to the anchor when viewed in the Browser, although in debug
build, it asserts: NS_ERROR("not a UTF8 string").
Back in Composer, viewing the link in the Link Properties dialog will mangle
the umlauted "a".
So my first question is: Are escaped characters allowed at all in URLs?
If they are, then this bug is in some URL code and probably not the editor.

Comment 8

[Charles Manske]

Attachment: Test of a named anchor and link containg a foreign character

Comment 9

[Charles Manske]

When viewing the test file, reduce window size so the anchor at the bottom
is not visible, but the link at the top is. When you click on the link, it will
scroll the window so the named anchor is visible.
Load this into Composer and the bug is revealed by simply clicking on the link
and then on the Link toolbar button to view in Link Properties dialog.

Comment 10

[Charles Manske]

Naoki: I was told that you have been dealing with character conversion issues in
named anchors etc., so I added you to this bug.
Well I just found the basic problem: If we get the "href" string from the <a>
tag like this: globalElement.href, then the character is not converted
correctly, but if we use: globalElement.getAttribute("href"), the conversion is
correct!

Comment 11

[Charles Manske]

Attachment: Fix for bug.

Comment 12

[nhottanscp]

The bug I fixed was bug 58819. Non Latin1 anchors were not recognized even if
they matched with a document charset.
I fixed to treat them as UTF-8 first then fallback to a document charset.
Here is spec.
http://www.w3.org/TR/html4/appendix/notes.html#h-B.2.1.

Comment 13

[Kristi Fredrickson]

Replacing kristif with robinf. Robin Foster-Clark is now the docs owner for 
things Composer.

Comment 14

[Charles Manske]

checked in

Comment 15

[Charles Manske]

checked in

Comment 16

[sujay]

we need to mark this RESOLVED-FIXEd....the bug has no resolution...thanks!

Comment 17

[Charles Manske]

Marking fixed

Comment 18

[sujay]

verified in 2/7 build.


- created a test html file (see attachement) with mozilla composer {build
id: 2001020720} on a winnt pc.
- tested the file with mozilla: correct.
- ie 5.0 handled the file allso correct.