IRCloggy #hts 2013-10-04

Provider of IRC logs since 2005.

WARNING: As Freenode became unjoinable and lost all warnings in topics, we cannot log channels on Freenode anymore.

	2013-10-04
	← zombor_ left	00:06
	← IvanG1 left	00:14
	→ [1]JPeterson joined	00:24
	← JPeterson left	00:26
	→ zombor joined	00:38
	← zombor left	00:42
	← [1]JPeterson left	00:42
	→ _julian_ joined	00:53
	← _julian left	00:56
	→ zombor joined	01:09
	← zombor left	01:12
	→ JPeterson joined	01:30
	← JPeterson left	01:37
	→ zombor joined	01:45
	← zombor left	01:47
	→ zombor joined	02:04
	→ JPeterson joined	02:38
	← zombor left	03:10
	→ zombor joined	03:20
	← zombor left	04:02
	→ joobz joined	04:20
	← manio left	05:18
	→ manio joined	05:19
	→ NisseDILLIGAF joined	05:30
	→ serafean_ joined	06:10
bow	yay friday	06:12
	← serafean_ left	06:14
	→ serafean_ joined	06:14
eindano	finally!!!	06:17
manio	adamsutton: so there is a problem with capmt and dvbrewrite, right?	06:26
	adamsutton: you disabled one CAID which is passed as PMT and it worked, then?	06:27
bow	not sure about that... all ca pids should be passed, up to the receiving end to figure out which to use, regardless of order	06:28
manio	bow: exactly, it's up to oscam to decide on the caids	06:30
bow	if something needs to be forced there are ignore/prio lists there that can be configured	06:31
	← serafean_ left	06:32
	← eindano left	06:56
	→ IvanG joined	07:42
	→ serafean_ joined	07:46
	← linuxstb left	07:46
	→ hadifarnoud joined	07:50
	→ hadifarn_ joined	07:53
	← hadifarnoud left	07:54
	← serafean_ left	07:57
	→ Rudde joined	08:17
	→ linuxstb joined	08:24
adamsutton	lja: can you give gdb output for the crash	08:27
	lja: and you're seeing slow zaps? odd, I've not been seeing that at all, certainly not that order of magnitude, and mpmc seemed to think it was quicker (though I doubt that)	08:27
	bow manio: I think you misunderstood my point about the hacking out the CAID, that was just to test a hunch about what capmt code might be doing wrong. ANd the hunch was right.	08:28
	that's not a proposed solution though, just testing a theory	08:28
	but anyway, I'm considering re-writing that code, I see lots of problems in there related to connection managment that ought to be fixed, that and the fact that although I've tried to move some stuff to common code there is still a considerable overlap with cwc code.	08:29
bow	but I think you can assume there will always only be one capmt connection	08:30
adamsutton	that's not what I see from reading oscam	08:30
bow	if you mean the actual camd.socket	08:30
adamsutton	no	08:30
bow	what other connection is there? :)	08:31
adamsutton	oscam creates a single UNIX socket for clients to use, but they create a per adapter UDP socket (9000+adapter number) for oscam to connect back on	08:31
	presumably that's the fake ca interface?	08:31
bow	ah yes that part is outside capmt	08:32
	its manios oscam mode	08:32
adamsutton	ok, to "me" it's all the same :p	08:32
bow	previously handled by the wrapper	08:32
	but yeah faking ca devices is a separate headache	08:33
adamsutton	I've not read that wrapper code, did it trap ioctls or something? and present a "fake" ca0 device?	08:33
bow	yes	08:33
	exactly	08:33
adamsutton	and manio wrote a variant that just used UDP?	08:33
	so he added that code to oscam as well?	08:33
bow	yes	08:33
adamsutton	so he's one of the 100+ that I need to slap about?	08:33
	;)	08:33
bow	indeed	08:33
	vdr and other things do the same	08:34
	similar wrappers	08:34
	adamsutton slaps manio with a big trout stuff with spaces and tabs (whichever floats your boat)	08:34
adamsutton	*stuffed	08:34
	ok, well, it's that multi socket (oscam mode) code that's the issue, it used several patterns that only make sense when you have a single socket etc...	08:35
bow	it could be one	08:36
	and the adapter nr included	08:36
adamsutton	yeah, but you'd need one per client	08:36
bow	but that would change oscam too	08:36
	the boxtype pc modr	08:36
	which is shared with vdr	08:36
adamsutton	s'ok, I'm not against the multi socket approach (it wouldn't be the way I'd do it, but that's irrelevant)	08:37
	just the way the multiple sockets are managed	08:37
	← Tuomaz_ left	08:40
	← joobz left	08:44
	→ Tuomaz_ joined	08:47
lja	morning	08:58
	adamsutton: unfortunately I wasn't running gds	08:59
	gdb	08:59
adamsutton	just make sure you have coredumps enabled	08:59
	I very rarely run under gdb	08:59
lja	ok, I will	09:09
adamsutton	not happy it's still crashing, though not entirely surprised, I was never convinced I had explained all the crashes that had been reported.	09:10
lja	I remember it crashed while messing with UI stuff	09:10
adamsutton	hmm, ok, that might be "less" worrying in some respects	09:10
	probably implies a very stupid (and probably easier) issue, rather than some complex bit of lock/recursive calls problems	09:11
	but maybe I'm just hoping ;)	09:11
clandmeter	adamsutton: how is your campt spagetti?	09:19
adamsutton	I'm rewriting it	09:19
clandmeter	bit for bit?	09:19
adamsutton	3 reasons, 1) it has bugs, 2) it'll help me understand it better and 3) I have a habit I have to feed!	09:19
lja	talking about captm... how much work would it be to include a cccam client?	09:20
adamsutton	lja: don't go there ;)	09:20
	or at least, ask amet	09:20
	anyone but me	09:20
lja	I know close to nothing related to how CA works	09:21
	:(	09:21
adamsutton	I'm learning as I go	09:21
	pretty much sums up me and TVH :p	09:22
	I'm still padawan to andoma ;)	09:22
clandmeter	lja: whats wrong with oscam+dvbapi	09:23
bow	you can use oscam as a cccam client	09:24
lja	clandmeter: it forces you to have oscam in tvh box	09:24
clandmeter	lja: there are good reason to have it ;-)	09:24
lja	bow: I know I use it	09:25
bow	well the logic is you'd be forced to have oscam there in the tvh box anyway, you'd just be moving it into the tvh source tree hehe	09:25
	bit by bit	09:25
	but oscam as a dvbapi client is now zeroconf! or almost	09:25
	could add so that it automatically sets boxtype = pc	09:26
clandmeter	when your box is powerfull enough to run tvh, oscam doesnt matter.	09:26
lja	using the newcamd client from tvh you don't need oscam	09:26
bow	right, its just the hassle that needs to be adressed	09:26
	right but that only works in straightforward situations, the most basic use cases	09:27
	and even if thats enough for you atm sooner or later the provider will change something in the stream or switch ca systems and there is suddenly need for some or all of the oscam dvbapi functionality to figure it out	09:27
clandmeter	we could ask adamsutton to build oscam inside tvh	09:28
bow	gah no :P	09:28
lja	lol	09:28
clandmeter	i think ill write an issue	09:28
adamsutton	NO NO NO	09:28
clandmeter	haha	09:28
adamsutton	and a few more	09:28
	NO NO NO	09:29
	NO ^ 100	09:29
bow	the solution is just to make capmt + oscam as easy to setup as cwc, in the tvh web	09:29
adamsutton	it is isn't it?	09:29
bow	years since I tried, but it was a bit misleading then	09:29
adamsutton	why?	09:29
bow	wasn't clear whether you had to specify the camd.socket path or the portnr	09:29
	adamsutton goes to check what it looks like	09:30
bow	and there could be checks to determine whether oscam is actually running (camd.sock exists)	09:30
adamsutton	no I'd say its pretty obvious ;)	09:30
lja	I remember reading something about supporting mgcamd in cwc, will it happen?	09:30
bow	and a hint telling you to start it	09:30
clandmeter	bow: it has	09:30
adamsutton	and I'm pretty sure the debug does that tooo	09:30
clandmeter	green is ok red is trouble	09:30
adamsutton	plus that	09:31
bow	yeah but will the user understand what to do hehe	09:31
clandmeter	green is ok and red is: ask adamsutton	09:31
bow	or that they have to obtain oscam themselves	09:31
adamsutton	bow: it's not like joe bloggs actually does any of this :p	09:31
bow	in the stbs thats exactly who does it :)	09:31
adamsutton	this is NOT a STB	09:31
lja	-> I remember reading something about supporting mgcamd protocol in cwc, will it happen? <-	09:32
adamsutton	lja: ask amet	09:32
clandmeter	adamsutton: what about an ajax irc client in webui? query directly to you for support?	09:33
bow	multiple caids over one newcamd connection was added in some form yes	09:33
	not quite full extended newcamd but probably enough now that the worst bugs have been fixed	09:33
adamsutton	clandmeter: good idea, but I think I'll redirect all queries to random people in this channel ;)	09:33
clandmeter	:)	09:34
adamsutton	picking someone at random is as likely to get you a valid answer :p	09:34
clandmeter	amet said he cant code, but he can talk	09:35
adamsutton	well, that's definitely true	09:35
	← Rudde left	09:46
	← lja left	09:56
	→ lja joined	09:59
amet	hey, leave me alone	10:09
	linuxstb loves windows - it has a "timeout" command-line program to wait for N seconds, but N is arbitrarily limited to 99999. So to wait until Sunday to do something, I need two of them...	10:11
mpmc	linuxstb: Glad to hear your love for Windows ;)	10:13
linuxstb	This is probably the most ludicrous stackoverflow answer I've read - http://stackoverflow.com/questions/1672338/how-to-sleep-for-5-seconds-in-windowss-command-prompt-or-dos	10:13
	Answer: use ping	10:13
jwh	linuxstb: timeout #1 triggers timeout #2 at the end of the 99999 seconds :P	10:13
linuxstb	Obviously 99999 seconds is enough for anyone.	10:14
jwh	heh	10:14
	you know	10:14
	you could write one	10:14
	:P	10:14
linuxstb	Isn't that Microsoft's job?	10:15
jwh	no	10:15
	they already did it	10:15
linuxstb	So where do I file a bug report? ;)	10:15
jwh	(if you don't use powershell you're doing it wrong :P)	10:15
linuxstb	Oh yes, I'm sure I'm doing it wrong ;)	10:15
	"crontab -e" didn't work though...	10:16
jwh	PS C:\Users\jwh> Start-Sleep -s 25000	10:16
	etc	10:16
	NAME	10:17
	Start-Sleep	10:17
	SYNOPSIS	10:17
	Suspends the activity in a script or session for the specified period of time.	10:17
	SYNTAX	10:17
	Start-Sleep -Milliseconds <int> [<CommonParameters>]	10:17
	Start-Sleep [-Seconds] <int> [<CommonParameters>]	10:17
	mmhmm	10:17
linuxstb	Guess that's what I wanted then. But anyway, I just wanted to moan about the arbitrary limit, it wasn't hard to just use it twice.	10:18
jwh	'timeout' is old (like most remnants of dos commands)	10:19
	the modern way is powershell	10:19
	coz PS is pretty awesome actually	10:19
	adamsutton hasn't even used powershell, yet it still makes me chuckle!	10:19
adamsutton	windows, powershell, oxymoron?	10:19
linuxstb	Start-Sleep is a powershell command? I always thought powershell was some kind of third-part replacement command prompt/shell	10:19
jwh	kinda	10:20
	except it's not :P	10:20
	mpeg4ip.h:126:58: error: new declaration .char* strcasestr(const char, const char).	10:20
	/usr/include/string.h:369:28: error: ambiguates old declaration .const char* strcasestr(const char, const char).	10:20
	nice	10:20
linuxstb	jwh: I think I've hit that error. Can't remember how I fixed it...	10:21
jwh	well it seems obvious, ish	10:21
adamsutton	just delete the declaration that's no longer needed....	10:22
jwh	yes	10:22
	exactly what I've already done	10:22
	evil code	10:22
	if gcc complains, the code must be really broken	10:22
linuxstb	I think there were a few things needed to make that compile, nothing hard though I think.	10:23
jwh	tempted to use opencore-amr/libfdk_aac anyway	10:23
lja	someone give me a nice question for C and C++ to make to a guy that claims to be an expert in that	10:36
	short and simple would be even great	10:37
	(its for a job interview)	10:37
bow	ask if he's a tvh contributor	10:38
lja	well, I can ask if he is an oscam contributor...	10:38
	and fail him	10:38
	!	10:38
	or send him to adamsutton for a few slaps	10:39
adamsutton	I slap anyone that writes code like that!	10:39
	if(X) {	10:40
	if (Y) {	10:40
	if (Z) {	10:40
	do_stuff();	10:40
	}	10:40
	do_more_stuff();	10:40
	}	10:40
	i_am_lost();	10:40
	}	10:40
lja	haha	10:40
adamsutton	whoever wrote that code should be forced to rewrite it in python!	10:40
jwh	so many nested ifs	10:41
	I have a bad habit of doing that though	10:41
adamsutton	jwh: no, nested if's with ZERO indentatin :p	10:41
jwh	o	10:41
	yes	10:41
	bad	10:41
	READABILITY	10:41
adamsutton	when you get 50 lines into crap like that you have NO IDEA where you are!	10:41
	I was confused by that little bit I just typed :p	10:41
jwh	yes	10:41
	also people who don't label their #endifs	10:42
adamsutton	lja: do you really care if they're an expert in C/C++?	10:42
	yeah, I tend to do that, except if the endif is 2 lines below the if	10:42
jwh	yes	10:42
	but nested	10:42
	anoying	10:42
	annoying	10:42
adamsutton	ah, indeed, try to avoid that and keep such things to isolated bits of wrapper code	10:43
linuxstb	lja: You could ask him what's wrong with "char x = 200"; Or how to printf the value of a pointer	10:43
	← cheasee left	10:43
lja	adamsutton: not for the job... I need him to be an expert in python, but I need to test his truth about what he wrote in the CV	10:43
adamsutton	linuxstb: the first one would depend on if its C or C++ :p	10:43
	linuxstb doesn't know C++	10:43
jwh	c++ is evil	10:44
	linuxstb would like to know how to printf the value of a pointer...	10:44
adamsutton	lja: I don't tend to care if they know this language or that, though obviously knowing C helps if you're going to be writing lots of C	10:44
jwh	kill it with fire	10:44
	linuxstb: %p	10:44
	:P	10:44
adamsutton	linuxstb: %p :p	10:44
jwh	adamsutton: get out!	10:44
adamsutton	lja: are you really interested in if they're a C/C++ expert, or just able to code and/or problem solve etc...?	10:44
lja	the second	10:45
adamsutton	if the latter, set them something from the google challenges	10:45
	I was going to do that for my interviews, but never got around to it	10:45
	I used to do them just for fun, and they can be a useful test of someones ability to solve problems and write basic code	10:45
lja	he also claims he does programming with his RPi	10:46
	interesting	10:46
jwh	nobody does programming on their rpi	10:46
adamsutton	lja: http://code.google.com/codejam/	10:46
jwh	they just install xbmc on it	10:46
adamsutton	well, some do ;p	10:46
	:p	10:47
lja	haha, tell that to linuxstb	10:47
jwh	although, I'm unbreaking some stuff on rpi atm	10:47
	did I mention gstreamer is heinous and needs to die?	10:47
	linuxstb failed to install xbmc on it, the first time he tried (raspbmc). OE worked though.	10:47
amet	sraue: will love you for that	10:47
adamsutton	raspbmc was a pig	10:47
amet	was?	10:48
adamsutton	ironically I read a review recently (from twitter) that said raspbmc was "easier" to install than "OE", I nearly choked on my coffee!	10:48
linuxstb	They probably have a point-and-click Windows installer	10:48
amet	yeah, depends who pays the reviewer more in HW	10:48
adamsutton	it took me 30 mins to get raspmbc to boot the installer, it still had to then download crap from the internet for final installation and it just kept failing	10:49
	10mins after I started teh download of OE it was running :p	10:49
linuxstb	Sounds like my experience.	10:49
	Exactly ;)	10:49
	This was a year or more ago though.	10:49
adamsutton	and 9 mins of that was the download and write to uSD ;)	10:49
amet	lol, monty python and the holy grail never gets old...	10:50
adamsutton	that's true, which is why I said "was a pig", I've no recent experience, but I doubt its much better	10:50
amet	just sayin	10:50
	→ serafean joined	10:51
lja	assembly quick question: fastest way to multiply by 7!	10:52
	you all fail	10:53
linuxstb	Precisely which CPU?	10:53
lja	lol	10:53
	no need to use real mnemonics	10:54
	and no... no floating point	10:54
adamsutton	depends on the req's :p and no I don't know a decent answer :p	10:55
linuxstb	I'm guessing a left-shift (*8) and a subtraction.	10:55
adamsutton	I'd have assumed something similar	10:55
linuxstb	But maybe a simple MUL r1,r1, #7 would be the same.	10:55
adamsutton	I'm sure lja will give us the answer soon....	10:56
lja	thats what linuxstb said	10:56
adamsutton	left shift and sub?	10:56
lja	yes	10:56
adamsutton	x << 8 - x?	10:57
linuxstb	But couldn't a MUL operation be just as fast?	10:57
adamsutton	x << 3 - x	10:57
linuxstb	(depending on the CPU)	10:57
lja	don't know any that is	10:57
	adamsutton feels inferior and goes back to the horrible task of DB archiving (really not my thing)	10:58
lja	tbh I don't know timmings for newer x86(_64) cpus	10:58
linuxstb	If this is an interview question, perhaps ask how to multiply something by 7 without using the multiplcation operator?	10:58
adamsutton	what sorta level?	10:58
lja	last interview I got the answer: "sum it up 7 times"	10:59
linuxstb	It's not wrong.	10:59
lja	but its not the fastest	10:59
adamsutton	neither is left-shift / sub ;)	11:00
	linuxstb has to make a big decision now - what to eat for lunch	11:00
adamsutton	mmm, food	11:00
	actually ignore me	11:01
	adamsutton crawls under his DB rock	11:01
adamsutton	and hopes there is food there!	11:01
	← serafean left	11:01
linuxstb	DB archiving == "pg_dump databasename \| gzip -9 > archive.sql.gz" ?	11:01
	→ serafean joined	11:02
adamsutton	not quite	11:02
	we have some DBs that collect historic readings of data, and they're growing unwieldy someone had the idea of chopping data from the live DB after X months and dropping into read-only (ish) DBs	11:03
	actually it's a completely nonsense idea, as in reality no-one will ever look at the archives, but I lost taht argument and was dumped with some code someone had written	11:03
	which I used to feed my habit, and rewrote!	11:03
	it might be a shit idea, but at least teh code that performs the shit idea will be good!	11:04
	linuxstb goes out to forage on the streets of London	11:05
lja	btw, I bought a skystar usb yesterday	11:09
	but	11:09
	I done it from that cheap store	11:09
	in UK	11:09
	http://extak.com/	11:10
	after doing the payment I they sent me to a 'rating' page to rate the store	11:10
	and... I didn't like what I read	11:10
	does anyone live near Birmingham to go there and kick their ass?	11:11
	basically -> http://www.trustpilot.co.uk/review/www.extak.com	11:11
linuxstb	lja: Unfortunate.	11:14
lja	there is still hope!	11:14
	(but not much)	11:14
	→ sIRwa2 joined	11:15
lja	the hilarious thing is that you make the pursuase and after you pay they redirect you to rate their store at that site	11:15
	← sIRwa2 left	11:16
linuxstb	They are probably too incompetent to disable that feature on their website...	11:16
lja	like if they were telling you: "hey now that you paid, go check how screwed you are"	11:16
linuxstb	lja: I mostly just buy things from amazon nowadays - not always the cheapest, but mostly they are competitive, and delivery always reliable.	11:21
lja	true, sometimes paying a bit more is safer	11:21
linuxstb	Although the Skystar USB (S2 version) is £57.84 there.	11:23
lja	I paid £38	11:23
linuxstb	Yes, but for what? ;)	11:24
lja	for a headache!	11:24
linuxstb	lja: Have they given you an estimated delivery date?	11:29
lja	14 oct	11:29
linuxstb	That's a long time.	11:31
lja	I'll be happy if the item arrives :)	11:33
	← hadifarn_ left	11:46
	→ hadifarnoud joined	11:46
adamsutton	isn't that the same place linuxstb was looking at? I read the reviews, looked like they were a bit poor on delivery time, but like most people said, you get what you pay for	11:47
linuxstb	adamsutton: Don't think it was me.	11:51
adamsutton	ok, mind playing tricks on me as usual, hopefully this sandwich will help!	11:52
	that and the can of Coke :p	11:52
linuxstb	Mmmm, Coke...	11:52
adamsutton	indeed	11:53
linuxstb	Logs say andyb2000 mentioned that site.	11:53
mpmc	adamsutton: [rewrite] I've deleted the network and added it again (added a mux too) but it isn't doing anything at all.	11:53
adamsutton	linuxstb: ah yes, that was it...	11:54
	mpmc: dvb-s? setup satconf?	11:54
mpmc	I didn't delete the satconf, it's still there.	11:55
adamsutton	but no longer connected to the network....	11:55
mpmc	So I have to re-create the satconf too even though it still exists?	11:58
adamsutton	no, you have to reconnect it to the network	12:01
	it's become disconnected, since you deleted what it was linked to	12:01
	linuxstb is looking forward to this going into master ;)	12:01
	← sirdancealot left	12:02
adamsutton	yeah yeah I know	12:02
	amet and clandmeter have plans to redo the satconf interface	12:03
	or more likely have plans to beat me up until I do	12:03
	but I do wonder sometimes....	12:03
linuxstb	So you need a default satconf? Couldn't tvh default to a universal LNB and "no diseqc" ?	12:03
mpmc	adamsutton: Ah! >.< I was using the edit button in the satconf pane which has no options! You edit the network via a dropdown.. REALLY confusing.	12:04
adamsutton	yeah, as I've said time and time again, the satconf UI is utter SHITE ;)	12:04
mpmc	It's not shitty as such, just confusing as hell.	12:05
adamsutton	no, its shite :p	12:05
	I wrote it, I get to call it what it is	12:05
mpmc	You're too hard on yourself.	12:05
adamsutton	it's fine if you know what you're doing, but the whole point is its confusing	12:05
	it's completely illogical, because it's illogical under the hood	12:06
amet	we have started ... round 2 coming up	12:06
adamsutton	as in you've got new ideas?	12:06
	or better yet, you've rewritten all the code?	12:06
amet	no, all ideas are the same... we need to continue beating	12:07
mpmc	One thing I have noticed though, if you try and watch a service, it'll sometimes fail because it's busy doing init scanning/epggrab. You have to request it twice so it'll release the frontend.	12:07
amet	in rewrite?	12:08
mpmc	Yes	12:08
adamsutton	what? never	12:08
amet	yeah, I have seen it too	12:08
adamsutton	don't believe either of you	12:08
amet	wa told I am dreaming it up ;p	12:08
	*was	12:08
adamsutton	you are!	12:08
amet	lol	12:08
	anyone running xbmc master?	12:08
mpmc	don't worry, I usually feel like I don't exist.. I'm used to it xD	12:08
adamsutton	mpmc: please provide logs with --trace mpegts,linuxdvb,subscription for stuff like that	12:09
	adamsutton won't believe it till he see's it!	12:09
amet	I cant get tvh pvr addon to play anything	12:09
adamsutton	great, xbmc just gets better and better ;)	12:09
	let me guess its because we don't output an i-frame on start?	12:10
amet	no idea	12:10
adamsutton	that's usually the "excuse"	12:11
amet	opdenkamp: is ignoring me so I have no idea.. its on OSX, so it coudl be audio related as XBMC is going through some shit with audio.... again	12:11
mpmc	adamsutton: That won't help tbh as there is 'no free input', I think it fails because the tuner is already tuning to a different mux at the same time as you request a service.	12:11
linuxstb	adamsutton: Why doesn't the tvh pvr addon just set normts?	12:11
	At least that would remove the excuse...	12:12
adamsutton	linuxstb: don't ask me ;) but I'm hearing this complaint for people with timeshift enabled	12:12
	so I'm dubious anyway...	12:12
amet	oh, wait...	12:12
	let me try that	12:12
adamsutton	mpmc: what won't help?	12:12
mpmc	adamsutton: getting logs etc.	12:12
adamsutton	it will, it'll show me what's happening and then I might believe you ;)	12:13
amet	adamsutton: dsabling transcoding helped	12:14
adamsutton	transcoding? what you doing with that enabled?	12:14
amet	hehe	12:14
	need to feed low powered boxes	12:14
	→ sirdancealot joined	12:15
adamsutton	ok, no idea about transcoding, I've not looked at that code at all	12:15
amet	yeah, disabling that fixed the issue	12:15
	i'll check with opdenkamp when I see him	12:15
linuxstb	I found it often doesn't work, or will only work with specific codec settings (e.g. AAC audio wasn't working for me, just MPEG2AUDIO (which is actually MPEG-1, but …))	12:16
	← serafean left	12:17
	← linuxstb left	12:19
	→ zombor joined	12:23
	← hadifarnoud left	12:32
	→ hadifarnoud joined	12:32
	← hadifarnoud left	12:37
mpmc	adamsutton: Doing a service mapping, I'm getting "2013-10-04 13:49:01.152 service: 12422H/Al Jazeera Eng si 0x7f299812bc10 weight -1 prio -1 error 402" Whats the issue do you think?	12:47
adamsutton	pastebin a log, it's probably related to the stuff I was changing last (priority etc...)	12:48
	someone has already reported an issue with service mapper	12:48
mpmc	adamsutton: OK the mapper is still going even though I'm actually watching the service, just a load of the above error while list cat caid info etc. Anyway to stop this?	12:51
	listing*	12:52
adamsutton	honestly can't remember, restart TVH for now	12:52
	there is an API call to stop the service_mapper, but not sure it's actually accesible from teh UI	12:53
	hold on....	12:53
mpmc	haha, okay :p	12:53
adamsutton	try /api/service/mapper/stop	12:53
mpmc	all I got was, "{}" and its still going xD	12:54
adamsutton	hmm, ok	12:55
	probably broken then :p	12:55
mpmc	I _just_ love watching Al Jazeera Eng!	12:57
adamsutton	logs please	12:58
	→ hadifarnoud joined	12:58
mpmc	adamsutton: I've restarted it, I was being sarcastic ;)	12:59
adamsutton	I know, but I'm still waiting for logs ;)	12:59
mpmc	For which? the mapping bug or the tuning fail?	12:59
adamsutton	tuning fail	13:00
	epggrab/initscan blocking etc..	13:00
mpmc	Ok, give me a moment, I need to redo the config in that case as init has finished, unless I can start it again somehow?	13:01
	Lol, yeah I've just realised, it'll init scan on restart xD	13:03
	← sirdancealot left	13:06
mpmc	Blarg, it wont do it now >.<	13:07
	→ linuxstb joined	13:46
	→ Gaston\|Home joined	13:55
adamsutton	mpmc: ??	13:57
jwh	Services:	14:01
	2043	14:01
	Muxes:	14:01
	mmm	14:01
	ALL THE SERVICES	14:01
	→ serafean joined	14:05
adamsutton	?	14:11
jwh	just checking whats on this new dish	14:15
	gstreamer is winding me up now	14:16
	time for another cig	14:16
	if libav/ffmpeg zealots failed less, it would have openmax support	14:16
	← serafean left	14:19
	→ JoeSixpack joined	14:26
	← NisseDILLIGAF left	14:39
lja	adamsutton: "lcn" means local channel number?	14:52
	in the services config file	14:53
BtbN	ffmpeg doesn't support any hardware acceleration	15:04
	why should it support omx?	15:04
jwh	because then it would be less shit	15:06
	but they're too busy having domestics	15:06
	→ DexterF joined	15:13
DexterF	hi	15:13
	can tvhe generate a playlist for mplayer, vlc and others?	15:15
BtbN	jwh, ffmpeg is a codec collection. If you want a media framework, use gstreamer or vlc	15:22
	← tyldis left	15:35
	→ tyldis joined	15:35
	← JoeSixpack left	15:36
adamsutton	lja yes	15:37
lja	DexterF: yes	15:37
	at least for vlc	15:37
DexterF	lja: well - how?	15:38
jwh	BtbN: what	15:39
lja	adamsutton: if I want to create a new channel I just need to create a new file with a unique hash and link it to the hash of the service?	15:39
	in the dvbrewrite	15:39
jwh	BtbN: what do you think libav is?	15:39
adamsutton	lja: can't you just hit the ad button?	15:39
BtbN	well, the same as ffmpeg, just forked by some upset developers	15:39
adamsutton	or is that not working? or missing?	15:40
BtbN	causing a bunch of pointless trouble to everyone	15:40
jwh	yes, but most importantly, ffmpeg is pretty much just a frontend to the underlying library	15:40
lja	DexterF: http://tv_headend_ip:9981/playlist/channels	15:40
BtbN	i'm talking about the whole thing, not the ffmpeg cli tool	15:40
jwh	nm, almost got gstreamer behaving anyway	15:40
BtbN	gstreamer is great	15:41
jwh	it is certainly handy	15:41
BtbN	don't have time to finaly complete my gst tvh implementation	15:41
jwh	pipes/cli syntax is a total fucker though	15:41
BtbN	but it still works fine	15:41
	yeah, gst is not realy intended as a end user cli tool	15:41
	you can still build nice stuff with it	15:41
jwh	yeah	15:41
	only using it for omx	15:42
	cluster of rpis for transcoding	15:42
BtbN	i'm successfully using it for full hardware transcoding	15:42
jwh	shame there isn't any hardware audio encoding	15:42
BtbN	decoding, deinterlacing, encoding	15:42
jwh	yup	15:42
lja	adamsutton: was that "half-question" form me? if it was, I didn't get it!	15:42
	for	15:42
jwh	mpeg2 decoding, h264 encoding on the rpi	15:42
	mp2->aac in software	15:42
BtbN	not on the pi	15:42
	intel HD2500 in my case	15:42
jwh	ah	15:43
BtbN	can handle ~10 1080p transcodes in realtime	15:43
	then it runs out of vram	15:43
jwh	does it output decent quality video?	15:43
BtbN	it's not as good as x264, but it's realy not bad	15:43
jwh	might be worth looking into	15:43
BtbN	die intel devs have a strange habit of still developing their reference implementations against gst-0.10	15:44
	no idea why	15:44
	but it should be ported to 1.0 soon	15:44
jwh	old :(	15:44
BtbN	https://gitorious.org/gst-plugins-vaapi/gst-plugins-vaapi	15:45
	most commits go to the 0.4 branch	15:45
	which is gst-0.10	15:45
	no idea why	15:45
jwh	Internal data stream error.	15:45
	useful as ever, gst	15:45
BtbN	run it in debug mode	15:46
jwh	doesn't give me anything more	15:46
	:(	15:46
	obviously pipe ordering is broken	15:46
BtbN	huh? It usualy gives you a hell lot of information if set high enough	15:47
jwh	above 1 I think it is, its just noise	15:47
	but yeah, working out the proper pipe order is a right mindfuck	15:47
BtbN	never had any problems with that	15:48
	it's usualy always the same	15:48
jwh	doesn't seem to be in any logic order	15:48
	like	15:48
	sink comes before encoding on the examples I've found	15:48
	logic dictates output sink should be last	15:48
BtbN	hm? That doesn't make sense, the output sink always is last	15:50
	it's usualy inputsourcewhatever ! mpeg2/h264parse ! decoder ! encoder ! outputsink	15:50
jwh	I know	15:50
	I need to play with it some moe	15:51
	more	15:51
	but now	15:51
	its hometime!	15:51
BtbN	it's possible that you got multiple pipelines there	15:51
	the syntax for that is quite confusing	15:51
jwh	it is	15:53
	nm, work it out monday	15:54
	← tyldis left	15:54
	→ tyldis joined	15:55
	→ NisseDILLIGAF joined	16:06
manio	adamsutton: ping	16:11
	adamsutton: i had a busy day at work and now i'm reading your capmt converstion with bow...	16:12
	adamsutton: currently it work like this: every channel opens a new connection to a camd.socket	16:12
	and there is a back UDP connection - one per adapter	16:13
BtbN	yes, it's an extremely... broken protocol	16:13
	i think the socket could just be used bidirectional	16:14
manio	i am not saying it's ok - i just used the code in wrapper to be compatible with	16:14
	← hadifarnoud left	16:14
	→ hadifarnoud joined	16:16
	← Skeptix left	16:20
	→ Skeptix joined	16:26
	← andreasb- left	16:39
	← zombor left	17:02
	→ zombor joined	17:03
	→ xhaggi joined	17:37
adamsutton	manio: yeah I get that, its just it does a bad job of managing the connections IMHO. Also there is this bug, so I'm probably going to rewrite it, I'll keep junks related to capmt protocol handling. But there are several things I think could be improved.	17:37
manio	adamsutton: thing which you should know:	17:38
	currently it uses one connection per channel	17:38
	when a connection is closed, oscam is automatically stop decrypting (one channel)	17:38
	but there is new interface for this, maybe it's worth to consider using it:	17:39
	http://openpli.org/wiki/caPMT	17:39
	i was playing with it but didn't see any time benefits	17:39
	32ms to 2ms is not visible to end user	17:39
	anyway - this way we can use only one connection to camd.socket - instead of each for a channel	17:40
	adamsutton: also BtbN has a good idea - to handle key handing on the same camd.socket connection instead of UDPs	17:41
BtbN	well, that's nothing that can just be done in tvh	17:43
manio	BtbN: i know - it's a thing for oscam, and i was also thinking about it	17:43
	anyway - there has to be some protocol for this	17:43
	currently it's CAPMT standard for -> oscam direction	17:44
	what about the replied keys - pack it into capmt frames?	17:44
	introduce something new?	17:44
	anyway - it's something which can be introduced in the future	17:45
	s/which/what/	17:46
adamsutton	.	17:48
	manio: one connection would be better yes, but as BtbN points out, that requires updates to OSCAM, at this stage I'd rather just improve the code based on something known to work/exist	17:49
manio	adamsutton: sure, anyway i will try to improve it (currently in my own sandbox) :)	17:50
adamsutton	cool	17:50
manio	adamsutton: in a way using simple camd.socket connection (and back)	17:50
adamsutton	I've already done some work moving common elements of cwc/capmt into a set of routines etc... I'll be going a step further now I better understand this stuff	17:51
	I've also been playing with a small buffer to overcome occasional latency in key responses, works quite well, but code needs cleaning up	17:52
	manio: I'll attack things from the point of view of improving the structures in TVH, if you fancy creating a nicer protocol for oscam (single port etc..) then we can see about combing our efforts	17:54
	manio: is the intention to make everything route through camd.socket?	17:55
	that would be how I'd do it	17:55
manio	adamsutton: yes, this is what i was planning	17:57
DexterF	welll.	17:58
manio	adamsutton: but it needs some channel state management in tvh	17:59
DexterF	I was pointed at http://tv_headend_ip:9981/playlist/channels , but vlc 2.1.0 does not play that, gives me tons of errors	17:59
manio	adamsutton: and some changes in oscam too	17:59
DexterF	how do I use the list generated there? smplayer would take it but played some random channel	17:59
mpmc	DexterF: I've used tried it and it works fine here.	18:01
	just*	18:01
adamsutton	manio: what management? one thing I will be doing is actually fixing all the broken structures related to this, so I might already be covering your issues	18:01
manio	adamsutton: i just need to know what channels are currently 'subscribed' (decoded)	18:02
adamsutton	yeah, well that's already known	18:02
manio	adamsutton: at any time (in case of disconnection)	18:02
adamsutton	that's already known within capmt code	18:02
manio	adamsutton: + PMT updates	18:02
DexterF	mpmc: just oben that in vlc as a playlist?	18:02
adamsutton	no change required there, though I'll be fixing some minor issues	18:02
DexterF	*open	18:03
mpmc	DexterF: Yeah.	18:03
manio	adamsutton: stopping channel/starting channel - this kind of events	18:03
adamsutton	manio: what do you need PMT updates for? you just want changes in CAIDs?	18:03
manio	adamsutton: yes, CAID/ECM PIDS, PROVIDER IDs	18:03
adamsutton	ok, you don't actually need that	18:04
manio	adamsutton: you have it in local config, right?	18:04
	adamsutton: and it's updated after channel switch (opening filter)	18:04
adamsutton	we get the necessary info indirectly, via the table (ECM) input	18:04
	If its not, it should	18:04
	don't worry though, all that's pretty trivial	18:05
	the info is already there, just a case of sending it to oscam	18:05
bow	there are 2 sets of ca descriptors in the pmt	18:05
DexterF	mpmc: ok, it asks now for credentials - wouldn't do that when I passed the m3u directly at the cli	18:05
adamsutton	??	18:05
	bow:??	18:05
bow	the global one and then possibly others per individual stream	18:06
DexterF	it does not accept the login, but I'll see about that now...	18:06
adamsutton	yeah, they're treated as one	18:06
bow	i.e a specifc audio/video pid can have its own capid	18:06
adamsutton	I think it just enumerates the lot	18:06
mpmc	DexterF: tried it as http://user:pass@....?	18:06
bow	really? the cam will be expecting things to look as it would in the ts... for example for seca there are workarounds in place in oscam for that kind of thing	18:06
DexterF	mpmc: I guess that's the problem, that exported m3u has no credentials in the stream address	18:07
adamsutton	tvh just sends SID,PID,CAID,PROVID	18:07
	I think	18:07
manio	adamsutton, bow: but it's packed into special type of data	18:07
	oscam is then extracting it	18:07
mpmc	DexterF: Yeah thats got me a few times xD	18:07
manio	it's called CA descriptors	18:07
adamsutton	exactly	18:07
bow	that data (descriptors of type 0x09) are the same as in the raw pm	18:07
	so the intention is tvh should be just passing those along as is	18:07
DexterF	mpmc: well, in terms of security it is a good idea not to plain-text export user creds I guess	18:08
adamsutton	yes	18:08
	it doesn't look inside the ECMs if that's what you mean?	18:08
bow	ecms dont exist at this level	18:08
	ecms are what the cam reads later on, once its decided which of the capids it wanted	18:08
adamsutton	ok, sorry, my mistake, still getting the terminology right	18:09
bow	its only because tvh has a built in client in cwc that it also does this to some extent	18:09
manio	yes, the only information needed by oscam is capmt table, it's easy to filter it from dvb adapter and i am using it in vdr (when i don't have capmt table from vdr)	18:09
bow	I wrote pmt parsing for oscam for the ecm harvesting test, and I was able to use ca-descriptor code already in there (same exact function used to parse the capmt traffic, but used for the actual pmt read directly from a separate filter)	18:10
	its actually quite messy and complex because of the stream specific extra descriptors	18:10
	and the workarounds to handle that when say one of the audio streams has a different set of ecms	18:11
adamsutton	I imagine if that happened, TVH wouldn't handle it	18:11
bow	and besides the 0x09 ca descriptors there are others in there, but probably the extra enigma private descriptor thats added on is the only important one (the others are logged by oscam but ignored)	18:11
	yeah its pretty rare	18:12
adamsutton	tvh assumes one set of keys per service	18:12
bow	oscam sort of generates a dummy service, for the other audio or video pid	18:12
	on the fly	18:12
adamsutton	not much use if TVH only stores one set of keys ;)	18:13
	anyway I'll try and do some updates based on what I know (or think I do :p) and see where I get	18:14
manio	adamsutton: if you interested: http://www.dvb.org/resources/public/standards/En50221.V1.pdf page 30	18:14
adamsutton	manio: not really ;) I'll leave the complex bits to you :p I'll just fix the internal structures (a bit)	18:15
manio	adamsutton: this is what oscam is expecting - creating this frame is already done in tvh	18:15
	sure :)	18:15
bow	there are timing issues too hehe	18:15
	but since its working atm probably not critical	18:16
adamsutton	bow: what timing?	18:16
bow	well tvh shouldn't for example filter for ecms first and then do the capmt parts	18:16
adamsutton	how do you mean?	18:17
bow	it should be in response to the pmt being read	18:17
	i.e first thing after zap	18:17
	but probably is?	18:17
adamsutton	how do you mean by filter for ECMs and then do capmt?	18:18
bow	well you said something that made it sound that ecms were a factor in the tvh capmt internals hehe	18:18
	+like	18:18
adamsutton	did I?	18:18
bow	i.e info needed by capmt being available through the ecms :P	18:18
adamsutton	ok, I'm lost	18:19
bow	but there aren't any ecms, tvh shouldn't be filtering for them at all unless cwc is in use, and then that would be an entirely separate thing from capmt	18:19
adamsutton	maybe I just used the wrong terminology	18:19
	I think I was mixing ECMs and CA descriptors?	18:19
bow	ok	18:19
manio	guys	18:19
adamsutton	where exactly do the ECMs come from? just so I get it right in future	18:20
bow	they're read from the capids indicated by those ca descriptors	18:20
manio	does tvh have all information about caid/ecm pid and provider id for a channel?	18:20
	in it's local config?	18:20
adamsutton	yes, but its also rebuilt if it changes etc...	18:20
bow	manio: not in its config, that is the information that is in the pmt	18:20
manio	so i think it's best to put it as soon as possible to oscam	18:20
bow	in those ca descriptors	18:21
adamsutton	the PMT structure is basically cached in config, there are some reaosns for this, but mostly its not needed ofc	18:21
manio	bow: i know it's in PMT but i am asking about the local storage, which is cached in tvh	18:21
adamsutton	yes it is	18:21
bow	but it shouldn't really rely on that	18:21
adamsutton	the last PMT is cached to disk	18:21
	it's not relying on it	18:21
	using it as a best guess until a PMT update is fine though	18:22
bow	the cam wants to see the pmt as it was in the stream, so it can deal with all the nonstandard crap the providers put there	18:22
	so parsing it and rebuilding it for capmt is likely to cause problems unecessarily	18:22
manio	bow: yes, the standard is telling that only ca descriptors should be there	18:22
adamsutton	well if it wants the real PMT it can get it directly	18:22
bow	but sending it raw (as originally intended) is no longer an option because of the now mandatory enigma extra descriptor that needs to be injected :)	18:23
	yeah but the ca descriptors are often misleading in subtle ways	18:23
	there are extra ones, decoys... and pids "encoded" to make the ecms hard to find	18:23
adamsutton	well, we can only work with what we're given	18:23
bow	the cam already knows how to work with all this, hence the need for passthru	18:24
manio	i was only trying to speed things up :)	18:24
	i mean channel zapping	18:24
adamsutton	if oscam wants the raw PMT, we can send it the raw PMT, if it doesn't we won't	18:24
manio	because if we have cached PMT, we could even send it to oscam before the filters are set...	18:24
adamsutton	manio: we can, and do I think	18:24
bow	yeah it might help some	18:24
adamsutton	capmt definitely sets up the structures, I was looking at the code today	18:25
bow	but it could create artifacts that are very difficult to troubleshoot hehe	18:25
adamsutton	I didn't trace to see where it sent it though	18:25
	tbh, if it works 99% of the time (or more correctly 99% of the time for me) I'm happy :p	18:26
bow	thered does seem to be a lot of unecessary code in there atm	18:26
	-d	18:26
	→ Rudde joined	18:26
adamsutton	where, capmt in TVH?	18:27
manio	bow: just curious - do you have some experiences with pmt_mode=6 (reverse capmt)?	18:27
bow	yeah... but still might be good to have as much manual control over what gets sent as possible	18:27
	its the default in openpli yes	18:27
	its used transparently	18:27
manio	you noticed some speed up when comparing to 'legacy' mode?	18:27
bow	instead of camd.sock opened by oscam, there's .listen.camd.sock opened by enigma	18:28
	and oscam connects to that instead of the other way around	18:28
	the only difference ought to be that multiple cams can connect to the same dvb app	18:28
	which makes more sense than the other way around since there is only ever going to be one dvb app	18:28
manio	yes - the only benefit i see now - is that it is using one single capmt connection (to provide PMT for all decrypted channels)	18:28
	instead of separate connection per channel	18:29
bow	I thought oscam only used one regardless?	18:29
	or it opens the same socket multiple times?	18:29
adamsutton	bow: I'd definitely not want to use that!	18:29
manio	currently it opens new connection per channel	18:29
adamsutton	one thing I definitely need is multiple apps to oscam, much easier for me!	18:29
bow	are you sure? I mean there can only be one /tmp/camd.sock file	18:30
	or are you talking about the udp stuff now?	18:30
adamsutton	yes, and multiple clients should be able to connect through that	18:30
manio	i cannot do it differently, because PMT updates on single connection closed decrypting the old one on PMT update	18:30
adamsutton	or does it artificially limit it?	18:30
bow	enigma doesn't care how many connect to its .listen.camd.sock	18:30
adamsutton	I'm sure I've used capmt from both my TVH instances at the same time, but maybe I wasn't	18:30
manio	then i see how it's done in vu+ and i did it the same way, now when a connection is closed, a simple channel is stopped on oscam	18:31
	bow: i am talcking about /tmp/camd.socket	18:31
bow	and you're saying oscam has multiple connections to that?	18:31
	or tvh does?	18:32
manio	tvh is connecting multiple times to the same socket	18:32
bow	ah ok	18:32
manio	every new channel - new connection	18:32
bow	yeah thats probably not ideal	18:32
	but thats just tvh internals	18:32
manio	can't be done without this new pmt_mode=6	18:32
	so fat	18:32
	so far	18:32
	now we can :)	18:32
bow	course it can hehe	18:33
	tvh would just maintain a single connection, have a separate thread + lock for it	18:33
adamsutton	bow: not if the connecting operates i the other direction!	18:33
bow	that only affects the initial connection... who establishes it	18:34
	its still one way communication either way	18:35
	just a matter of who is the server that opens the socket file and who is the client that connects to it	18:35
	as I said the only reason for reversing it is so you can have multiple cams... and enigma has both at the same time (i.e it will connect to camd.sock if one exists, but still open .listen.camd.sock and allow anyone to connect there)	18:36
adamsutton	yes, but if I want N arbitrary TVH instances it certainly makes more sense (to me) for TVH to connect to oscam	18:36
	→ cheasee joined	18:37
bow	yes I don't think there will be any advantage to the reverse, until someone demands support for running cccam, scam and oscam simultaneously :P	18:37
manio	:)	18:37
adamsutton	anyway, I would reject any commits if it removed the current mode :p	18:38
	and I don't really see the benefit myself	18:38
	but anyway, I'm outta here	18:38
manio	as i told you - the only benefit is see is one global connection instead of one per channel	18:38
bow	but you can do that anyway	18:40
	the connection is supposed to be global	18:40
manio	afaict oscam not handled it correctly	18:41
bow	enigma only connects once, so tvh can too	18:41
	or so I've assumed anyway	18:41
manio	bow: http://www.streamboard.tv/wbb2/thread.php?postid=398960#post398960	18:42
	bow: but maybe it changed, it was some time ago	18:42
bow	but thats old	18:42
	yeah	18:42
	pretty sure it has	18:42
	now oscam logs exactly what each pmt update is about, and what it decides to do with it	18:43
	at level 128	18:43
manio	ok	18:43
	→ JoeSixpack joined	19:15
	← SpiceTV left	19:59
	← NisseDILLIGAF left	20:07
	← JoeSixpack left	20:47
	→ lja_ joined	21:10
	← zombor left	21:10
	← DexterF left	21:23
	← xhaggi left	21:28
	→ zombor joined	21:35
	← zombor left	21:35
	→ zombor joined	21:35
	← zombor left	21:47
linuxstb	Anyone using Freeview in the UK? Can you set a series record for Jools Holland on BBC Two HD tonight? It worked for me for the SD channel. but not HD...	21:58
lja_	it worked for me	22:05
linuxstb	Can you set it on both the HD and SD version of BBC Two?	22:05
lja_	only tried hd	22:06
	nop, now I cant set for the SD	22:07
linuxstb	OK, not just me then. I did SD first, and then couldn't set it on HD.	22:07
lja_	hmm, on the autiomatic recorder it is set	22:07
	"Automatic Recorder" tab	22:07
linuxstb	Ah yes. Three times for me ;)	22:08
lja_	but the "Upcoming recordings" only shows the HD one	22:09
	(the one I done first)	22:09
linuxstb	Yes that's what I saw.	22:09
lja_	Homeland season3 starts this sunday on channel 4, right?	22:20
	← IvanG left	22:32
	→ SpiceTV joined	22:35
lja_	linuxstb: what's the state of pidvbip ?	22:52
	progress	22:52
	adamsutton: ping	22:57
linuxstb	lja_: Yes (re: Homeland). It started in the US last week.	22:57
	I haven't done anything on pidvbip recently. But I'm using it. Distracted with my modulator.	22:58
lja_	did you got to implement / try image scaling?	22:58
	for lower resolutions?	22:58
linuxstb	No, I haven't done anything.	22:59
lja_	I'm wondering if adamsutton will make further changes to the way the new code stores the services, channels, and all that stuff	23:05
	linuxstb moans that the subtitles on Later… with Jools Holland are now badly auto-transcribed instead of proper synced lyrics as they used to be in the past	23:06
lja_	hehe	23:07
	← hadifarnoud left	23:07
lja_	I watched a bit and I noticed that	23:07
	→ hadifarnoud joined	23:07
	← hadifarnoud left	23:12
	← Rudde left	23:44

Logs Search ←Prev date Next date→

Channels Documentation