72074 - Increased frequency of failure in Character Code East Asian AutoDetection of EUC-JP

Status: VERIFIED DUPLICATE of bug 57872

(Core :: Internationalization, defect)

Description

[hirata masakazu]

2001031408 and 0312 build.
Goto any site in ISO-8859-1 with Auto Detect set to Japanese.
Then jump to a site in Euc-JP (seen in some Shift-JIS sites, too).
Much more often than previously observable, Browser fails to switch character
coding.
URl is an example of Euc-JP site which fails frequently.
It has gotten worse in recent builds.

Comment 1

[nhottanscp]

Reassign to shanjian.

Comment 2

[Andreas Becker]

Changing QA Contact to [e-mail].

Comment 3

[Shanjian Li]

this problem might be related with the latest JS change.

Comment 4

[Yuying Long]

On 03-08 Mac Mtrunk build, it works fine with the page in url field.

Comment 5

[hirata masakazu]

I think this is invalid.
I forgot that I used a fresh profile for other testing and AutoDetect had been
turned off by that.
Sorry for the spam.

Comment 6

[Yuying Long]

Mark as verified.

Comment 7

[hirata masakazu]

I would like to reopen this, since setting AutoDetection to East Asian is
actually failing a lot at EUC-JP sites including the URL.
The steps to reproduce is the same as the original description.
Mozilla tends to switch to Simplified Chinese for those sites.

Comment 8

[Yuying Long]

Checked on 03-16 Mtrunk build, it will works fine with Auto-Detect Japanese but
does't work well with Auto-Detect East Asian and Auto-Detect All(Linux OK
though) on all platform.
Change platform to all.  Also I just filed another bug 72299 with EUC-JP site.

Comment 9

[hirata masakazu]

I see a seemingly related problem:
If you manually set the character coding to Shift-JIS for those EUC-JP sites, in
some cases, the window goes blank. The URL above is not one of such cases, though.

Comment 10

[Shanjian Li]

This problem should be resolved after my new charset detector lands. The 
problem  Hirata Masakazu mentioned in his 3/16 comment should have been 
resolved. If not, please file a new bug with test case.

Comment 11

[hirata masakazu]

This has been getting worse in recent builds.
Even the sites with metatag are now switched to GB.
try http://www.asahi.com/ with Autodetect=East Asian.

2001040911 Mac trunk.

Comment 12

[hirata masakazu]

Even when set to Japanese, this Shift-JIS site is switched to UTF-16;
http://www.h3.dion.ne.jp/~haneko-i/

2001041204 Mac trunk

Comment 13

[Jaime Rodriguez, Jr.]

Shanjian - Please assigned priority P3, and milestone M0.9.1.

Will your "my new charset detector" be both in moz and commercial?

Comment 14

[Jaime Rodriguez, Jr.]

Marking as nsbeta1, and intl. Adding Teruko.

Comment 15

[Shanjian Li]

East Asian charset detector will detect EUC-JP as GB2312, that is a known problem. 
My new charset detector will solve the problem, but it will not be landed on mozilla
any time soon. 

Other problems reported in this bug report are irrelavant to above one. I will file 
bug 78229 for asahi problem. I will mark this bug as future.

Comment 16

[Shanjian Li]

Euc-jp is detected as GB2312 was reported in bug 57872. Marked this one as 
duplicate of that one.

*** This bug has been marked as a duplicate of 57872 ***

Comment 17

[Yuying Long]

This Euc-jp is not work well when auto-detect Esat asian is a dup of 57872.
But even Auto-detect / All will detect this page as Big5. 
Shanjian, is there any bug talking about this auto-detect All issue? or maybe I 
need open another bug?
 

Comment 18

[Yuying Long]

Auto-Detect/All works fine on 05-15 trunk build.  Mark this Auto-detect East Asian issue as dup.