92856 - With auto-detector turned on CJK, Netcenter France page is marked as window-1252 page

Status: VERIFIED DUPLICATE of bug 78229

(Core :: Internationalization, defect)

Description

[ji]

Build: 07/27 branch build (win32), 07/26 branch build (linux, mac)

With the latest branch build, when the auto-detector is turned on any
double-byte languages, like Japanese, Chinese or Korean, Netcenter France page
is not recognized as ISO-8859-1 page, instead windows-1252 is marked on the
charset menu.

Steps to reproduce
1. Launch browser with a new profile or clear cache wiht a old profile.
2. Turn on auto-detector on Japanese/Chinese/Korean
3. Go to http://home.netscape.com/fr
4. Select View | Character Coding, you'll see Western (Windows-1252) has dot
marked on the left. 
5. View page source on this page, the charset set meta tag is set to iso-8859-1.

So far above URL is the only page that has this problem, I don't see this on
Netcenter Germany page.

Comment 1

[ji]

It doesn't happen on 07/24, 07/25 branch build.
It seems a regression.

Comment 2

[ji]

Please note that the fix for bug 90288 has been checked-in on  07/26 branch build.

Comment 3

[Yuying Long]

On my win2k-cn, when I set Auto-detect to All, the charset marked as
windows-1252 as well.

I checked it on 07-25 win32 branch build on same machine, got same result as
build 07-27 branch build.

Comment 4

[ji]

On 07/24, 07/25 branch build, I don't see this problem with the first try, but 
on the second try (clear cache, go to the page again), the problem appears.

Comment 5

[Roy Yokoyama]

assigning to shanjian.  he knows the code best.

Comment 6

[ji]

Added Kat on the cc list.

Comment 7

[Yuying Long]

I saw same problem on WinME-Ja / 07-25 branch build, but WinME doesn't has
auto-detect All problem.

For Win2k-cn, if I using a new profile, then auto-detect All won't get problem,
but if I use a old profile and clear the cache, then auto-detect All and CJK,
Esat Asian will have this problem.  This is same case on 07-25 and 07-27 branch
build.   I guess it might not related with the check-in of bug 90288.

Comment 8

[ji]

Attachment: Saved Netcenter France page and attached to the report.

Comment 9

[ji]

On linux, with a 07/26 build + Japanese lang pack, I saw this page recognized as 
a EUC-JP page, so accented chars are displayed as question marks.

Comment 10

[Jon Rubin]

cc jbetak

Comment 11

[Shanjian Li]

In netscape french page, meta charset is specified in the middle of the 
document. Meta charset sniffing does not work for this page, and CJK charset 
detector return win1252 as detecting result, which is perfectly fine. The 
problem is, when meta charset is met later, reload/update does not happen  
correctly in certain situation. I trace the problem several times, and some time 
I can see the problem. However, I still need more time to find out what is the 
problem. 

Comment 12

[Shanjian Li]

After I simplified the testcase, I found out that this is a dup of 78229, which 
has a fix already. Marked as so. 

*** This bug has been marked as a duplicate of 78229 ***

Comment 13

[Shanjian Li]

I need to reopen this bug. It doesn't seem to be the case.

Comment 14

[Shanjian Li]

After some investigation, I need mark this one as dup of 78229 once again. 
Unfortunately, the proposed fix in 78229 does not work in this case, so a right 
patch need to be worked out. 

*** This bug has been marked as a duplicate of 78229 ***

Comment 15

[Yuying Long]

Mark it as verified per Shanjian's comments.
After bug 78229 fixed checked in, if still see same problem, then re-open it.