Closed
Bug 107869
Opened 23 years ago
Closed 23 years ago
Reply mail doesn't inherit the original charset
Categories
(MailNews Core :: Internationalization, defect)
Tracking
(Not tracked)
VERIFIED
WORKSFORME
People
(Reporter: ji, Assigned: nhottanscp)
References
Details
(Keywords: intl, regression)
Attachments
(2 files, 1 obsolete file)
(deleted),
text/plain
|
Details | |
(deleted),
patch
|
sspitzer
:
superreview+
|
Details | Diff | Splinter Review |
build: 10/31 trunk build. Reply mail doesn't inherit the charset of the original mail, instead it's using the folder charset. so as long as the folder charset is not consistent with the charset in the MIME header, the quoted mail will be garbled on reply mail compose window. And this only happens to single part mail. Steps to reproduce: 1. Have a single part ISO-2022-JP mail in you mail box. Make sure the foler charset is not ISO-2022-JP. 2. Select the mail and click on Reply, you can see the quoted mail is garbled. 3. Change the folder charset to ISO-2022-JP, and select the mail and click on Reply again. This time the quoted mail is displayed alright. Since the original mail has MIME charset info, reply mail should inherit that charset, it shouldn't use folder charset.
Keywords: intl,
regression
Assignee | ||
Updated•23 years ago
|
Status: NEW → ASSIGNED
Target Milestone: --- → mozilla0.9.6
Assignee | ||
Comment 2•23 years ago
|
||
I think this is related to the change of mimemsg.cpp rev=1.36. http://bonsai.mozilla.org/cvsview2.cgi?diff_mode=context&whitespace_mode=show&file=mimemsg.cpp&root=/cvsroot&subdir=mozilla/mailnews/mime/src&command=DIFF_FRAMESET&rev1=1.35&rev2=1.36 That is the place to set the original message's charset for a single part message. Reassign to shanjian, please take a look at this.
Assignee: nhotta → shanjian
Status: ASSIGNED → NEW
Assignee | ||
Comment 3•23 years ago
|
||
It is working with 2001-10-30-03-trunk, so it should be something else. Reassign back, sorry about that.
Assignee: shanjian → nhotta
Assignee | ||
Comment 4•23 years ago
|
||
Shanjian, I have to reassign again. I backed out your check in of yesterday (mimetext.cpp, mimetext.h) and that fixed the problem. I think you want to fix bug 107084 differently. E.g., you can check 'obj->options->override_charset' at MimeInlineText_rotate_convert_and_parse_line that is safer than your last change.
Assignee: nhotta → shanjian
Comment 5•23 years ago
|
||
It looks like a regression caused by the fix for bug 107084. Adding shanjian to the CC: list.
Comment 6•23 years ago
|
||
Oh, sorry. Seems you already know that...
Comment 7•23 years ago
|
||
shirley, can you try with a build before last tuesday? I want to know if you specified a folder charset, and checked "apply default charset to all message...", will the reply message use override charset or header specified charset? Thanks.
Status: NEW → ASSIGNED
Shanjian, the problem reported in this report doesn't exist on 10/29 trunk build. And with 10/29 build, if I force the folder charsert override, reply mail still inherit the charset of the original mail, it doesn't use the folder charset, which is the correct behaviour.
Comment 9•23 years ago
|
||
*** Bug 107983 has been marked as a duplicate of this bug. ***
Comment 10•23 years ago
|
||
changed platform to all and upgraded to critical... workaround is to set the charset on the message _before_ replying to it.
Severity: normal → critical
OS: Windows 2000 → All
Hardware: PC → All
Comment 11•23 years ago
|
||
I found what is the problem, I am working on a fix.
Comment 12•23 years ago
|
||
Comment 13•23 years ago
|
||
Some explaination to the problem and my solution: Before my patch in 107084, the text->charset is tried to resolve in "MimeInlineText_initialize", at which time obj->option hasn't been passed in yet. But header is available and text->charset can take over its value if it is available. When I fixing 107084, I delay text->charset initialization until everything is available, which happens after we begin to parse. However, in MimeMessage_close_headers we need to set charset for message window, and it relies on text->charset to be available and it will try to resolve charset from obj->option itself. Solution: We don't want to resolve charset in two different places. The approach used before patch 107084 also has some problems. If the charset is from autodetect or html metachar, message window's charset will not be updated. User will eventual see different charset used in viewing mail and replying mail. That's undesired. So I let MimeMessage instruct MimeInlineText either message window's charset need to be updated or not, and let MimeInlineText do the work.
Updated•23 years ago
|
Whiteboard: need r/sr
Assignee | ||
Comment 14•23 years ago
|
||
I think most of the messages has charset specified. Delay charset initialization is for special cases like override or auto-detection. I am not sure why can't we initialize it as before and reset the value when needed later (e.g by detected charset). It seems to me complicated, having three flags (inputAutodetect, initializeCharset, needUpdateMsgWinCharset).
Comment 15•23 years ago
|
||
You can set it as before, but how can you update it without setting a flag? In case of a multi-part message, only the first mime part should update the character set for message window. We can avoid this flag if we don't care about autodetection, but I am afraid that is not a acceptable solution.
Assignee | ||
Comment 16•23 years ago
|
||
One of the example is when the main body does not have a charset and auto-detect is on, right? If no other way to check the body has no charset then you need a flag, I agree. Can 'initializeCharset' be used if the body has a charset or not?
Comment 17•23 years ago
|
||
was the fixed for this checked in? I am not seeing this problem with this mornings smoketesting.
Comment 18•23 years ago
|
||
No, the fix hasn't been checked in yet.
Comment 19•23 years ago
|
||
Updated•23 years ago
|
Attachment #56109 -
Attachment is obsolete: true
Comment 20•23 years ago
|
||
I have done smoketesting and several other testing as suggested by naoki. No problem found. Ducarroz, can you review my patch?
Comment 21•23 years ago
|
||
Comment on attachment 56267 [details] [diff] [review] updated patch Looks good. R=ducarroz
Comment 22•23 years ago
|
||
Comment on attachment 56267 [details] [diff] [review] updated patch sr=sspitzer
Attachment #56267 -
Flags: superreview+
Comment 23•23 years ago
|
||
fix checked in.
Status: ASSIGNED → RESOLVED
Closed: 23 years ago
Resolution: --- → FIXED
Reporter | ||
Comment 24•23 years ago
|
||
Verified with 11/19 trunk builds. It's fixed.
Status: RESOLVED → VERIFIED
Comment 25•23 years ago
|
||
this has reappeared on mac build 2001-11-20-04-trunk
Status: VERIFIED → REOPENED
OS: All → Mac System 9.x
Hardware: All → Macintosh
Resolution: FIXED → ---
Comment 26•23 years ago
|
||
Naoki, could you take a look on mac?
Assignee: shanjian → nhotta
Status: REOPENED → NEW
Whiteboard: need r/sr
Assignee | ||
Updated•23 years ago
|
Status: NEW → ASSIGNED
Target Milestone: mozilla0.9.6 → ---
Reporter | ||
Comment 27•23 years ago
|
||
I don't see this, reply mail inherits its original charset, it doesn't take the folder charset. But I do see the attachment charset is overwritten by the mail body (bug 111055) and that is on all the platforms.
Assignee | ||
Comment 28•23 years ago
|
||
I tried mac build 2001-11-20-04-trunk and it's working except the problem reported as bug 111055.
Status: ASSIGNED → RESOLVED
Closed: 23 years ago → 23 years ago
Resolution: --- → WORKSFORME
Comment 29•23 years ago
|
||
verified on mac commercial build 2001-11-26-04-trunk
Status: RESOLVED → VERIFIED
Updated•20 years ago
|
Product: MailNews → Core
Updated•16 years ago
|
Product: Core → MailNews Core
You need to log in
before you can comment on or make changes to this bug.
Description
•