177270 - Incorrect character encoding used for bookmarked pages

Status: RESOLVED DUPLICATE of bug 174197

(Core :: Internationalization, defect)

Description

[muhdv98]

A bug that has been keeping to reappear in a few versions of Mozilla now is that
swedish characters ('å','ä','ö') are sometimes substituted by '?' signs in
web-pages. If a swedish char appears in a word of a mail subject, the whole word
is substituted by a '?' sign. This is the same for Linux and Windows versions of
Mozilla.
I hope that this bug hasn't been reported yet, eventhough I think it is strange
if  it hasn't. 

/Mattias

Comment 1

[R.K.Aa.]

Please include an URL where the bug can be seen
Also: Always include the mozilla build ID in bug-reports.

Comment 2

[muhdv98]

Since I'm currently not able to find a web-page that shows this bug, it is
perhaps better to just concentrate on the mail subject issue.

I am currently using Mozilla Build ID 2002101612.

Comment 3

[Niklas Dougherty]

едц (and other non-ASCII characters) might be displayed erroneously on web pages
if 1) the document or server produces no character encoding but your default is
set to something like UTF-8, or 2) if the document or server produces no
character encoding and you have universal auto-detection on, or 3) if your
profile/cache is corrupt, which will make Mozilla use UTF-8 when it shouldn't. I
have yet to see something to that effect myself, though.

Mail headers are usually corrupted by other software, such as servers and weird
mail clients.

What's your default encoding? Do you have universal auto-detect on?

Comment 4

[Niklas Dougherty]

A curiosity: when opening this very page with universal auto-detect on, Mozilla
thinks this is Windows-1251, displaying едц as Cyrillic glyphs.

Comment 5

[muhdv98]

Ok, this is perhaps not a bug after all, since I can't reproduce the web issue
anymore, and the mail subject issue was fixed by checking the "Apply default
character coding to all messages in the folder" checkbox for my Inbox.
Previously I had it unchecked and Western (ISO-8859-1) as Default Character Coding.
I assumed it was a bug since I have never changed any language or character
coding settings, and the mentioned issues did not exist in previous versions of
mozilla (somewhere before v 1.0). (Or perhaps something has changed on the
mailserver I use since then, if you think that might be the case.)

Hope I haven't caused any unnecessary work and/or confusion.

Comment 6

[Christopher Hoess (gone)]

per reporter's comments, WORKSFORME.

Comment 7

[Yuying Long]

Mark as verified.

Comment 8

[José Jeria]

This still happends to me, with latest build on Windows XP. The weird thing is
that it only happends sometimes. Will try to figure out why and hopefully attach
a testcase.

Comment 9

[José Jeria]

Attachment: Testcase

When I see this file locally it shows swedish וצה as ???, but if I make a copy
of this file and rename it it shows the swedish וצה.

Hopefully the ??? will show now when uploaded...

Comment 10

[José Jeria]

Now when viewed in bugzilla I see the цед, this is really driving me crazy.

Comment 11

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

When you see the problem on the local disk, what do you see if you enter
javascript:void(alert(document.characterSet)) in the URL bar?

Comment 12

[José Jeria]

It alerts UTF-8.
The (working) copy alerts ISO-8859-I, hm, same html, different files names.

Comment 13

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

What filename works and what filename doesn't work?

Comment 14

[José Jeria]

The non-working is named menu.html, if I make a copy and rename to whatever name
the וצה will work.

Should I mail you the file?

Comment 15

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Does "whatever name" have a .html extension?  There's no need to mail the file
if it's attached to the bug.  I'm just trying to get a full decription of the
problem recorded in the bug so that it can be analyzed.

Comment 16

[José Jeria]

Yes, all the new copies also have .html extension.

Comment 17

[José Jeria]

Now I know how to reproduce this bug every time.

1. Load testcase.
2. Add the URL to your Personal Toolbar by dragging it.
3. Now activate the bookmark (click on it), you will notice that the page will
reload but that the characters are replaced by "?".

If you do this locally, after the characters are transformed to "?", it doenst
matter how you open the file. The file will never show correct characters again.

Comment 18

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Confirming (on my Linux opt build from yesterday), changing component to
Internationalization, and raising severity to major.

Steps to reproduce:
 1. Load http://bugzilla.mozilla.org/attachment.cgi?id=105326&action=view
 2. Drag the icon in the URL bar to the personal toolbar.
 3. Hit the back button.
 4. Click on the icon in the personal toolbar just created.

Expected results:
 B O-umlaut R S E A-ring O-umlaut A-umlaut

Actual results:
 B ?-diamond R S E ?-diamond ?-diamond ?-diamond

Comment 19

[Simon Montagu :smontagu]

*** This bug has been marked as a duplicate of 174197 ***