5.0 I18N specification

	Description	Milestone	Status (as of M12)
mail/news
charset for a new mail	Charset of a new mail always set by the default charset of pref (i.e. no inheritance from a current window).	M4	Done
charset for reply/forward mail	Charset of the original message (main body) is used.	M14	Done
charset for mailtourl	Charset of a new mail always set by the default charset of pref (i.e. no inheritance from a current window).	M15	Done
charset for a new mail by address book	Charset of a new mail always set by the default charset of pref.		Done
attachment/send	No charset label for attachments unless HTML with META charset specified. If the main body charset is ISO-2022-JP, HTML attachments are base64 encoded.	M10	Done
thread pane display	Display subject/address in multiple charset by using charsets encoded in MIME header.	M4	Done
thread pane sorting	Use the application locale, store it in a message db (at folder creation time) then use it for locale sensitive string comparison.	M5	Done
thread pane date display	Use locale sensitive date/time format interface. Always use application locale.	M5	Done
message body view	libmime to convert message to unicode before passing to the layout.	M5	Done
charset override		M15	5938
attachment/view	Display multiple charsets.  Decide the charset by following process.  1) Content-Type charset 2) Charset menu selection 3) Auto charset detection if available (e.g. Japanese).	M6	Done 1)&3)
attachment view by browser	Follow browser's charset/auto-detect setting		Done
message save as	text/plain -> charset convert to platform file charset text/html -> no charset conversion	M14	23418
folder pane view	Multi lingual display by unicode.	M14	7844
message search widget	Multi lingual display by unicode.	M15
local message search	Header search, apply MIME decode and charset conversion before comparison if necessary.  Body search, apply decoders (quoted printable, base64, html named entity and NCR), plus charset conversion if necessary.	M15	11659
IMAP search	Send UTF8 query, plus fall back mechanism by mail charset of the mail folder, finally asscii.  4.51 to support the fall back, 4.5 only supports ascii search. 5.0 should support utf-8 query.	M16	5933
IMAP folder name	When a folder name is stored locally, it should be UTF-8 or modified UTF-7 instead of system charset.	M14	Done
message filter	Same issue as local message search	M15
newsgoup search	Whatever supported by 4.5 (need investigation)
address book
sorting	Same as the message thread sorting.  Use the application locale, store it in a message db then use it for locale sensitive string comparison.	M11	Done
address book widgets	Multi lingual display by unicode.	M11	Done
type down	Multi lingual support.
name completion	Multi lingual support.	M15
address book search		M15	16354
LDAP search
Preference
intl.mailcharset.cyrillic	not needed
intl.mailcharset.override_1	not needed
mailnews.send_hankaku_kana		M14	Done
default mail send charset UI		M14	23540

View:

• Charset labeled attachments can be viewed correctly.
• Unlabeled attachments may be viewed if charset detection is available.
• Otherwise, unlabeled (and not iso-8859-1) attachments cannot be viewed inline (i.e. displayed as a link).
• There is a post beta 1 feature which enable the charset menu to view unlabeled attachments (charset override #5938).

Send:

• Use HTTP charset as a charset label if available (highest priority).
• Use META charset for HTML as a charset label if available (second priority).
• Otherwise, no charset label  is attached to the sending attachments.
• Apply Base 64 encoding for Japanese attachments only.

• Charset for the storage is UTF-8 (escape/unescape for 8 bit data may be done by database).
• Charset conversion of UCS2 and UTF-8 (both direction) is needed between RDF and address book storage.
• Importing ldif needs no conversion (base64 decoding only) since ldif charset is UTF-8.
• Importing 4.x address book needs conversion from pref specified charset to UTF-8.
• I18n to provide a mapping function from csid to charset name since csid is used in 4.x pref.

text/html - Apply charset conversion first. For characters not convertered from unicode, fallback to named entity then NCR.

text/plain and message header - Apply charset conversion first. For characters not convertered from unicode, try transliteration (EUR, (tm)) then fallback to '?' (question mark).

Header search

• MIME decode and charset conversion for headers - changed, search term is now unicode (used to be a folder charset), we should compare it against MIME decoded and unicode converted header strings (see nsMsgI18NDecodeMimePartIIStr).

• Charset conversion for the search term - changed, use mail default charset (used to be a folder charset), (see nsMsgI18NGetDefaultMailCharset).
• Optional international search - new, parse the Content-Type charset per message and apply charset conversion to the search term (not to the body for performance reason),
• UI (a checkbox) for international search.
• QP decode for body (plain and HTML)
• Entity (CER) and NCR decode for HTML body - new, also could be an option (international search)

UTF-8 - new
Fallback search charset (mail default charset) - changed, used to be a mail folder charset

UTF-8 search by server

todo