18105 - Wrong MIME header when the Japanese subject is very long

Status: VERIFIED FIXED

(MailNews Core :: Internationalization, defect, P3)

Description

[ji]

Build: linux 1999110508
OS: RH6.0

When the subject contains Japanese chars and is very long,
MIME header is wrong and the subject is displayed  blank in the thread pane.

Steps of reproduce:
1. Open the mail compose window.
2. Enter a long Japanese subject in the subject field (like 25 Ja chars)
3. Send the mail to the account itself

You'll see the long Japanese subject is displayed blank.
If you open the mail folder using vi, you'll see the MIME header is wrong.

Comment 1

[nhottanscp]

Please check if this is a linux only problem or reproducable in Windows.
Also, paste the encoded subject in the bug report.

Comment 2

[ji]

The MIME header viewed on 4.7 doesn't start with =?ISO-2022-JP?
It simply starts with Ja characters.

With today's linux build, the Ja characters in the long subject are displaying
as different chracters instead of blank on 110508 build.

Comment 3

[nhottanscp]

*** Bug 18262 has been marked as a duplicate of this bug. ***

Comment 4

[nhottanscp]

The problem happening here is that MIME decoder encountered some kind of error
condition so it aborted and the original string was used instead without MIME
encode.
The MIME encoder logic used for 5.0 is basically the same as 4.x. One difference
I found between 4.x and 5.0 is the line wrapping. 4.x does not do the correct
line wrapping (e.g. the data of 18262 does not wrap and encoded in one line
which is not desirable). 5.0 fixed this problem but it may have exposed a
potential problem which was not invoked in 4.x.
Need more investigation.

Comment 5

[nhottanscp]

Correction:
>The problem happening here is that MIME decoder encountered some kind of error
MIME encoder instead of decoder

Comment 6

[nhottanscp]

The problem was cause by my change to the encoder (substitutes unmapped
character to '?') a couple of weeks ago.
Fix was checked in to M12.

Comment 7

[Katsuhiko Momoi]

Verification should be done on M12 builds as follows and
report back here:

ji --> check JPN/UTF-8 for Linux
momoi --> check JPN/UTF-8 on Windows
marina --> Latin 1 -- easily reproducible when the subject contains
                      Latin 1 strings and long

Comment 8

[ji]

Verified on Linux M12 1999111208 build. It is fixed.

Comment 9

[Katsuhiko Momoi]

Looks good on 11/12/99 Win32 build (1999111208).
Long headers in JPN are now encoded properly.

Comment 10

[ji]

Verified with Linux 121008-M12 build. it is fixed.