93017 - The auto-detect will keep the charset that previous page

Status: VERIFIED WORKSFORME

(Core :: Internationalization, defect)

Description

[Yuying Long]

Build: 07-27 0.9.2 branch build.

Steps to reproduce:
1. Launch browser, and go http://www.yahoo.com
   Note: The charset will be marked as ISO-8859-1 no matter the auto-detect set
to OFF, All or some CJK charset, which is correctly.
2. Doesn't matter clear cache here or not ( by going: Edit | Preferences |
Advanced | Cache ).
3. Manually set the charset to Shift-jis, GB2312 ... etc. And set auto-detect to
All, Japanese, Chinese ...etc.

Result:
The Charset will marked as the previous manually set charset ( x-jis, GB2312
...).  Even you clear cache, and reload again still get same result.
If you manually set charset to iso-8859-1, then doing auto-detect ALL, 
auto-detect Japanese or Chinese, will get charset as iso-8859-1.

It may related with bug 92856, but I don't know if they are same problem though.

Comment 1

[Yuying Long]

I first found this was on Japanese localized build - because the charset sets to
x-jis or EUC-jp depends on OS with auto-detect Japanese.
So, in those Japanese localized build, if I go these western page 
(yahoo, bugzilla ...) will always get the wrong charset by default.

Comment 2

[Roy Yokoyama]

assigning to shanjian.

Comment 3

[Shanjian Li]

I don't think this is a problem. A pure ascii webpage (as above) can be treated 
as almost any charset. Auto charset detection just choose to report nothing. If 
user manually set charset to be something, we assume (s)he did it with right 
reason and keep using that charset. In localized build, if user default charset 
is not 8859-1, that default will be used if there is no other clue. 

Comment 4

[Yuying Long]

Mark it as verified for now.