Charset-determining heuristics faulty for sent mail

Bug #9893 reported by Mikko Rauhala
16
Affects Status Importance Assigned to Milestone
evolution (Ubuntu)
Fix Released
Medium
Sebastien Bacher

Bug Description

This is a bug I submitted to Ximian, but they said that Evolution as shipped by
them does not have this bug and that the deb patches are at fault. The URL above
references my original report, but I'll cut/paste it here for completeness:

Description of Problem:

Evolution is overly conservative in determining a charset to use for
sending mail in the case where the default charset as set in the settings
is insufficient.

In particular, it seems to

Steps to reproduce the problem:
1. Set default charset for composer to be, say, ISO-8859-1, in order to not
send in more complicated encodings if it's not necessary.
2. Write a mail containing, say, "䀤" and, to be sure, nothing else
(that's a with umlaut, the euro sign and the currency sign). This is
unrepresentable with any single charset of the ISO-8859 family.
3. Send the mail (eg. to yourself) and watch in woe as Evolution chooses to
send the mail as ISO-2022-KR, which, while apparently preserving the euro
and currency signs, loses the a with umlaut altogether.

Actual Results:
Message does not contain all the characters it is supposed to contain as
explained above.

Expected Results:
Evolution should be less enthusiastic in trying to cram the message into a
charset that can't represent it and fall back to something that can,
probably UTF-8.

How often does this happen?
Predictably. It also occurs on longer messages that contain multiple
"scandinavian" characters (åäö), although I recommended the use of the
simplest message to exhibit this behaviour above to confirm the problem.

Additional Information:
When sending a message with only the euro sign being a "weird" character,
Evolution quite correctly chooses to send in ISO-8859-15 instead of
ISO-8859-1. This is okay, since in this case I didn't use any characters
not representable in that charset.

Also, when I add more and more weird characters such as the U263A SMILING
FACE and some kanji, Evolution _does_ give in and just use UTF-8 as it is
supposed to.

http://bugzilla.ximian.com/show_bug.cgi?id=68990: http://bugzilla.ximian.com/show_bug.cgi?id=68990

Revision history for this message
Sebastien Bacher (seb128) wrote :

confirmed

Revision history for this message
Sebastien Bacher (seb128) wrote :

evolution (2.1.2-0ubuntu1) hoary; urgency=low
...
  * debian/patches/camel-charset-map.patch:
    - removed this bugged patch (Hoary: #3219).

Revision history for this message
Sebastien Bacher (seb128) wrote :

*** Bug 25892 has been marked as a duplicate of this bug. ***

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.