GNOME Bugzilla – Bug 599312
Autodetect ICQ text-encoding
Last modified: 2012-08-29 13:53:15 UTC
This is a user-experience "bug". Problem: When chatting with a person that have a non-standard text encoding (e.g. a European person) special characters are displayed wrongly and an error message is added in the chat-window. Error message: "There was an error receiving this message. Either you and <ICQ-no> have different encodings selected, or <ICQ-no> has a buggy client." System: Running Empathy 2.28.0 with Karmic (9.10) on an amd64 machine. Temporary solution: Setting the text-encoding to ISO-8859-15 solves the problem (for me), this though is _very_ difficult for a novice user since he/she have to manually write the "ISO-8859-15" text in the preferences. Prefered solution: My proposal is that a drop-down list of possible text-encodings is added to preferences for text encoding to make it easier to set-up this. The best though would be if it is made automatic. For example Pidgin does not have this problem at all (what I know at least). For more information see Launchpad-bug: https://bugs.launchpad.net/ubuntu/+source/empathy/+bug/253452 Regards, Johan Andersson, Sweden
updated the lP bug with french characters like é è and ô guess this is already covered by your report here as you're speaking of European chars. regarding your prefered solution, why doesn't it simply auto fit into the system wide settings, which were able to handle e.g. ü, ä or ö
Due to the ICQ protocol and the nature character encodings, it's not possible to automatically detect which encoding is being used. Also, we can't just use the system encoding because it might be different to those being used by your contacts. Given these facts, the only potential way to address this is to have the user specify which encoding they need. We are not planning to add user interface to the 2.28 branch, and besides, I suspect that making UI changes at this point would be counter to the Ubuntu freeze rules.
Hi Dafydd, I know that karmic freeze is over but as empathy replaced pidgin as default chat tool we may better find a way to at least hide this message which makes it nearly impossible to track a chat with other users. It wouldn't be a very big problem if ü or ä would be spelled as ue and ae, but it is a problem if every average user will get this message on a default installation if he's writing in german. I understand your position that features in this direction may not be planned for the latest version, but at the other site I#m sure there is something possible. I've never recieved an error like this whith pidgin which was default for the last few years, so I'm sure there is a way to auto fix this. As you seem to have more knowledge in this I got an simple idea: pidgin is open Source too, can't you take a look in there and see how they fixed it ?
I think it is common to fallback to system local (ISO-8859-15 for French) if the string is not valid UTF-8. That should fix most of the problems, but is of course not error-prone. What charset does pidgin use by default? I see telepathy-haze has a charset parameter for ICQ so surely that's configurable in pidgin. There is no perfect solution here...
- I don't have pidgin anymore, but if someone would tell me where to look for it I could double check it at my brothers PC who's still running an older version of ubuntu. You talked about fallback to system default, and this may be causing my problem. I'm from germany, and use a german keyboard, but I've installed the system to come up with English UI, I thought it would use utf8 as default too, but if you know a way to double check it, just let me know.
xavier Claessens, I didn't see where to change text encoding manually in previous time. I've changed to (ISO-8859-15) and the issue is not there anymore for the following mixture of signs which incldude german AND french ones: ß öüä èéê ç -> guess pidgin does the same, so wouldn't it be possible that someone patch's this fallback you mentioned earlier ? + where can I see system wide encoding settings ? - I'm not sure whether mine is using UTf8 as I'm using the english interface.
Hi Dafydd ! you wrote: Due to the ICQ protocol and the nature character encodings, it's not possible to automatically detect which encoding is being used... there might be no information on the used encoding... but all (good) libraries for en-/decoders i know support an "automatic" en-/decoder guessing the encoding by the received characters - would be great to guess the encoding on a per-message-basis, per-buddy-basis or per-chat-basis... Since sending of text may also be affected, it would be great to use the guessed encoding also in sending. I agree most to copy good working code from pidgin! I think the per-user-fix to chenge his encoding on a per-account-base is as good as it will work on any original icq-client? But simply the default is strange ... but i see no good workaround for that. I would find it best to use auto-encoding as described above, and perhaps set and save it as buddie details locally?
Shouldn't this be a bug to be filed in telepathy rather than empathy? Otherwise, sounds like a "please provide a patch" kind of bug.
*** Bug 611468 has been marked as a duplicate of this bug. ***
Yeah, if anything this should be done in Haze.