I think the problem comes for EpgSelection.py enigma2 coding:

if cur[1] is None or cur[1].getServiceName() == "":         	
                    	if self.key_green_choice != self.EMPTY:             	
                            	self["key_green"].setText("")               	
                            	self.key_green_choice = self.EMPTY

If the serviceName returns as empty string it will not show Add timer ... but how can this happen ?????

Simply edit the file EpgSelection.py

if cur[1] is None or cur[1].getServiceName() == "":

should be this:

if cur[1] is None:

Then Add timer button is back and it seems to work .... but DMM will not like this chain saw approach ...

Well please try editing it and confirm that this helps - then probably DMM can look into it how it can happen that the ServiceName is not properly returned at this place of the code.

But then it is not my/epg loaders fault I think ?

But DMM changed a lot of things internally in enigma2 C++ code to make all thse nice new features work.

We are now simply entering new land to run this code parts with external epg data instead of internal one - so it really could be that we simply found a bug here, and if not we will have to further investigate.

At least you are lucky that I like this kind of puzzles

schau mit ls -alh /etc/enigma2/epg.db nach wie gross deine epg.db ist.

> 100MB kosten bereits ca. 20 Sekunden shutdown und etwa 10 Sekunden boot Zeit

And yes, if it is strange enigma2 behaviour my possibility to help is limited - disabling the check in enigma2 pythin part is only a workaround for C++ part not returning correct name.

Problem with language from xml ist that this is currently not 100% properly done - means I would nee to change Plugin further - which I tried to avoid as it is not mine and the coding partly is ... strange. Never seen so many try: except to hide poor coding And believe me, I really KNOW how to do BAD code.

OK, I have added also the language extraction of the xm files to the Plugin, but I had to write a manual routine to convert from 2 char to 3 char iso639 language codes:

see xmltvconverter.py for the get_xml_language routine:

# ugly, I know ...                         	
                            	if lang == 'en':                           	
                                    	r = 'eng'                          	
                            	elif lang == 'de':                         	
                                    	r = 'deu'                           	
                            	elif lang == 'fr':                         	
                                    	r = 'fre'                          	
                            	elif lang == 'fr':                         	
                                    	r = 'fre'                           	
                            	elif lang == 'es':                          	
                                    	r = 'spa'	
                            	elif lang == 'it':                          	
                                    	r = 'ita'                           	
                            	elif lang == 'nl':                          	
                                    	r = 'dut'	
...........................


        	#                                                    	
                   	# continue list here ...            	
                   	#                                                    	
                   	else:                                 	
                           	print "[XMLTVConverter] unmapped language:", lang
                           	r = lang

Please check for wrong/missing languages there if you have problems with exotic languages !!!!

Mapping was copied from here:

http://en.wikipedia.org/wiki/List_of_ISO_639-1_codes

Please install r20 and report if situation has improved if you experienced strange characters in loaded epg.

Ciao
gutemine

I don't think so because the matching happens also during the extract, remember I re-wrote the channelFilter to simply check with the service refenrece if a ChannelName is given which is more or less the same code, why should it return a Name there and then not later in the EPGList ?.

Furrther more than all channels on 28.2/5 should be affected, not some yes some no.

But you could try it out - edit satellites.xml and test it.

EDIT: die EpgSelection.py verwendet direkt eServiceReference.getName
Mein Code verwendet die ServiceReference.py mit channel = ServiceReference(str(ref)).getServiceName() und da wird in python noch zusätzlich gechecked:

def getServiceName(self):
info = self.info()
return info and info.getName(self.ref) or ""

Wahrscheinlich kommt da der Unterschied her - aber wie schon gesagt das sollte sich DMM anschauen, mir hat schon gereicht ins Plugin den Language support einzubauen, weil das waren 2h Nachtarbeit.

ABER dein Input könnte der Grund sein warum ich kaum EPG bei BBC habe wenn ich die Standardkanallisten lade um auch BBC als Channels zu haben. Könnte mir jemand der überall auf den UK sendern EPG hat seine kompletten Settings hochladen um das zu checken ?

Well, while the early versions deleted content from most of the T_* Tables when loading epg the current versions only delete from T_Event the ones that are actually coming from xmltv source_id.

This means the desscriptions still keep the old timestamp and could be dropped by the housekeeping. So we either have the choice - go back to erase everything or start updating timestamps selectively - I prefere the first one as this is much simplier logic wise and already has prooven to work and it is also relatively performant.

But it means that after the xmltv load an EPG refresh would be also needed like on the old days (where also epg.dat was replaces and hence everything from before was more or less lost) to get the non-loaded epg back.

At least the language handling seems to be working so the late night shift was not completely wasted

EDIT: Please use r21 which now deletes all tables on load again to prevent further inconsistencies - but this means also manual deleting of epg.db should NOT be needed anymore.
And as we now fill language too the only field of the epg.db where I'm not sure is this damned dvb_event_id which gors into T_Event - I simply use a counter starting from 0 to make it unique for every event, but I have no glue if this is a valid approach - maybe DMM can enlight us here a little bit

me too, but the first focus was to get the data into the db, but now it has to be and stay also consistent when you repeat the exercise on a daily basis.