Aaargghh, well without a surprise kannel created a a DCS=8 (UCS-2) message from your advice, so not what I wanted but what I expected. But, it turned out, that one of my developers didn't do what I told him to do (use the kannel API I have written) but did what everyone should expect what a developer does when an admin tells him to not touch an API: he touched the api and added an "mclass"-switch, what explains, why kannel sets a "wrong" DCS and charset. After deleting the "optimised" API and installing the original without mclass everything works fine.
Solution: never trust a developer! So sorry for wasting your time, I'm now going to to massacre a developer. If the police doesn't understand this, you won't hear from me for a couple of years, because I don't regret anything, but at least I don't have to raise sheep after all. ;-) Thanks Falko Alexander Malysh schrieb: > Hi Falko, > > it should work... > > Please provide your kannel version and if it's not CVS head try it first. > If cvs head doesn't work please provide your smsc group config and > full debug when you send message. > > Thanks, > Alex > > P.S. You should at least see error like this: Failed to convert > msgdata from charset... > > Falko Ziemann schrieb: >> I currently can't test it. But I bet 20 euro that this doesn't make >> kannel encode in latin1. >> If it does I would really love to pay you the money, but then I think >> kannel should be written completly new and I should get a new job... >> may be I should grow sheeps or sth. >> Regards >> Falko >> >> 2008/11/12, [EMAIL PROTECTED] <[EMAIL PROTECTED]>: >>> Try put >>> >>> a)charset=UCS-2 >>> b)coding=2 >>> >>> made sure the text is url encoded. >>> >>> >>> >>> -----Original Message----- >>> From: Falko Ziemann [mailto:[EMAIL PROTECTED] >>> Sent: Wednesday, November 12, 2008 2:50 PM >>> To: seikath >>> Cc: Kannel User >>> Subject: Re: Problem with character encoding >>> >>> >>> Ah, thanks, but äöü and ÄÖÜ are in the Latin1 characterset. äöü is even >>> in the most IA5 charsets (there are localised versions around, but >>> kannel uses the german localised version, good for me). >>> See, my problem is not finding the correct charset, but that kannel >>> doesn't use the charset I told it to use... >>> >>> Again the problem in short: kannel ignores the switch "alt-charset = >>> ISO-8859-1", encodes the messages with IA5 and sets the DCS to 0xf1 >>> also >>> no message_class is set. >>> >>> Regards >>> Falko >>> >>> seikath schrieb: >>>> :) >>>> >>>> its not kannel related issue here, >>>> yes, UCS2 is not primary character set, but its used to deliver the >>>> non >>>> Latin1 characters. >>>> well, by my knowledge, >>>> there is no other way to encode äöü etc accents characters sent >>>> to mobile devices >>>> >>>> try to get the encoding of incoming from your smsc characters >>>> similar to >>>> äöü .. >>>> >>>> hope someone else will help you more >>>> >>>> >>>> >>>> >>>> Falko Ziemann wrote: >>>> >>>>> Yes, there are similar issues on the list, but I couldn't find a real >>>>> solution for it. >>>>> There were some chats about hardcoding the DCS or asking the >>>>> operator to >>>>> alter the charset, but there must be some possibility to make kannel >>>>> talk ISO ... or not?! >>>>> >>>>> But I didn't get this UCS2-thing... >>>>> The systems "feeding" kannel use ISO-8859-15, my service-provider is >>>>> expecting ISO-8859-1, kannel produces IA5. So where is the benefit to >>>>> make ISO-8859-15 to UCS2 before I post it to kannel? By the way: I >>>>> have >>>>> set up 4 SMSC from diffrent vendors (and I'm talking about real >>>>> SS7 SMSC >>>>> in MNO enviroments, ok all in western europe) but I have never seen a >>>>> SMSC using UCS2 or UTF-16 as primary charset, that would be a 2 byte >>>>> encoding for a 7 bit charset. And I absolutly don't want kannel to >>>>> talk >>>>> UCS2 for that would cut down the payload per message to 70 >>>>> characters. >>>>> >>>>> Regards >>>>> Falko >>>>> >>>>> seikath schrieb: >>>>> >>>>>> Search the list for identical issues, they are common >>>>>> the smsc uses UCS2 anyway >>>>>> use iconv to convert from your german encofing to UCS2 >>>>>> >>>>>> example of perl implementation of Bulgarian Cyrilic cp1251 >>>>>> before xml post to be injected >>>>>> if ($data->{cyr} == 1) { >>>>>> $encoding="ucs-2be"; >>>>>> $converter = Text::Iconv->new("cp1251", $encoding); >>>>>> $data->{sms}= $converter->convert($data->{sms}); >>>>>> $data->{sms}= hex_encode($data->{sms}, '%'); >>>>>> $coding=2; >>>>>> $alt_dcs=-1; >>>>>> } >>>>>> >>>>>> php is much convenient >>>>>> >>>>>> Falko Ziemann wrote: >>>>>> >>>>>> >>>>>>> Hi, >>>>>>> >>>>>>> I try to send some special character (german umlauts äöü) over a >>>>>>> smpp >>>>>>> connection to a smsc running iso-8859-1 charset. My config says: >>>>>>> alt-charset = ISO-8859-1 >>>>>>> >>>>>>> But I see in the connection log: >>>>>>> [...] >>>>>>> 2008-11-11 16:10:35 [4332] [12] DEBUG: data: 78 20 7c 7b 7e >>>>>>> 24 00 >>>>>>> 20 31 36 3a 31 30 x |{~$. 16:10 >>>>>>> [...] >>>>>>> which are umlauts in IA5 (7c 7b 7e). >>>>>>> >>>>>>> gw_config.h says: #define HAVE_ICONV_H 1 >>>>>>> iconv -l contains ISO-8859-1 >>>>>>> >>>>>>> Can I anyone tell how get kannel to accept the alt-charset? The >>> provider >>>>>>> only supports 8859-1 (yeah, that's sick, I know...) >>>>>>> >>>>>>> Thanks >>>>>>> Falko >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>> >> >> > >
