214467 - HTML attachments in Hebrew Visual are displayed incorrectly by Mozilla Mail.

Status: NEW

(Core :: Layout: Text and Fonts, defect)

Description

[pavelp]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4) Gecko/20030624
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4) Gecko/20030624

The page is in Hebrew Visual. http://www.mof.gov.il/itc/branches.htm
When opened in browser, the page displays correctly.
When attached to mail, it is displayed *inline* as if it was Logical. Changing
"View->Character coding" does not affect the display in any way.
When the attachment is opened separately in browser, it displays correctly again.

Reproducible: Always

Steps to Reproduce:
1. Go to http://www.mof.gov.il/itc/branches.htm, Observe it is displayed correctly.
2. Mail it to yourself as an attachment, the mail will display inverted.
3. Double-click on an attached document, it will open the browser document
that's displayed correctly again.

Actual Results:  
Attached Visual Hebrew HTML is displayed as Logical Hebrew. Manually changing
Character Coding does not help.

Expected Results:  
The attached HTML should not be inverted.
Changing "Character Coding" should affect the display.

Comment 1

[Simon Montagu :smontagu]

Confirming.

This is probably soluble in theory, but my own feeling is that the cost in code
bloat would be greater than the benefit. We might want to add a release note.

Comment 2

[Katsuhiko Momoi]

I'm trying to revisit this issue. (Adding Jungshik also.)
As I recall, in MailNews, we handle each MIME part separately and this means we should be able to handle different charsets in MIME parts. (I think we made this change sometime ago -- certainly before this bug was filed). If it is working correctly, we should not have a problem like this. (Unless we have to make a special case of Visual Hebrew in Mail rendering.)

Looking at the sample page at: http://www.mof.gov.il/itc/branches.htm
the charset parameter specified is charset=Visual. I cannot access lxr right now and can't tell if this name is treated as a synonym of ISO-8859-8. If not, that may explain this bug. If it is treated as a synonym, then there might be another reason.

Comment 3

[Simon Montagu :smontagu]

visual is defined as an alias for ISO-8859-8:
http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/intl/uconv/src/charsetalias.properties&rev=1.41&mark=258#253

We correctly decode the characters, so the problem is not with the charset itself. The problem is in using different text types for separate sections of a single document. One way to display the attachment correctly (probably what I had in mind in comment 1) would be to insert an LRO character (U+202D) at the beginning of each line and a PDF character (U+202C) at the end.