Beirdobot, irc.freenode.net

Wednesday, June 11th, 2014, 01:10 UTC

[01:10:20]

superm1 (superm1!uid4318@ubuntu/member/superm1) has quit (Quit: Connection closed for inactivity)

[01:54:58]

fetzerch (fetzerch!~quassel@unaffiliated/fetzerch) has quit (Ping timeout: 240 seconds)

[02:00:34]

fetzerch (fetzerch!~quassel@unaffiliated/fetzerch) has joined #mythtv

[02:19:32]

peper03 (peper03!~peper03@mythtv/developer/peper03) has quit (Ping timeout: 260 seconds)

[02:20:13]

peper03 (peper03!~peper03@mythtv/developer/peper03) has joined #mythtv

[03:16:20]

Warped (Warped!~Warped@108-85-161-113.lightspeed.cicril.sbcglobal.net) has quit (Quit: ChatZilla 0.9.90.1 [Firefox 29.0.1/20140506152807])

[03:18:14]

fetzerch (fetzerch!~quassel@unaffiliated/fetzerch) has quit (Ping timeout: 252 seconds)

[03:19:24]

fetzerch (fetzerch!~quassel@unaffiliated/fetzerch) has joined #mythtv

[03:39:15]

Warped (Warped!~Warped@108-85-161-113.lightspeed.cicril.sbcglobal.net) has joined #mythtv

[05:47:34]

len_ (len_!~quassel@75-161-185-101.mpls.qwest.net) has quit (Remote host closed the connection)

[06:10:37]

FabriceMG (FabriceMG!~Thunderbi@LCaen-156-54-30-212.w80-11.abo.wanadoo.fr) has joined #mythtv

[06:23:44]

jya1 (jya1!~jyavenard@CPE-120-148-103-15.bjzv2.vic.bigpond.net.au) has joined #mythtv

[07:19:01]

jya:

damn…. spent a full day looking into using supposed to be much faster vaapi routine, which unfortunately gives us NV12 video frames.. So have to convert them from NV12 to YV12.

[07:19:24]

jya:

it’s still about 30% slower than the original version :( even SSE optimised

[07:19:36]

jya:

I guess intel does a better job doing that conversion.

[07:27:47]

tris (tris!tristan@2001:1868:a00a::4) has joined #mythtv

[08:12:16]

stuarta:

yeah, it's very interesting the differences you actually find in the processor instruction sets

[08:14:03]

andreaz (andreaz!~andre_000@p5DCA3B0D.dip0.t-ipconnect.de) has joined #mythtv

[08:18:31]

jya:

the interesting stuff (at least for me) is that in the YV12 conversion I added with support for different stripe size, the SSE optimised is actually slower than pure C code

[08:19:00]

jya:

I’ve given up of adding proper support of frames with all planes being on a 16 bytes aligned memory

[08:19:27]

jya:

this would have required being able to display those properly, our display code assumes that the buffer data in VideoFrame is all visible

[08:19:51]

stuarta:

does non visible data really exist?

[08:19:58]

jya:

as it requires to play with memory address boundary, I copy that in a temporary buffer. So while it works with about 16 lines at a time

[08:20:21]

jya:

it is about 6% slower. my guess is the extra memory

[08:20:32]

andreaz (andreaz!~andre_000@p5DCA3B0D.dip0.t-ipconnect.de) has quit (Read error: Connection reset by peer)

[08:20:37]

jya:

stuarta: well, using the VDPAU and VAAPI code to retrieve an image from the GPU

[08:21:01]

jya:

yes, the video frame retrieved has strides != width

[08:21:04]

stuarta:

is the slowdown not the copy itself?

[08:21:11]

jya:

so all lines start on a 16-bytes aligned memory

[08:21:29]

jya:

stuarta: I think it is

[08:21:58]

stuarta:

can we not force alignment with compiler defines? aka __align__ ??

[08:21:59]

jya:

let me find you the article about those video frame, those are all valid ones BTW.

[08:22:10]

jya:

stuarta: that’s just not possible

[08:22:15]

stuarta:

cool thanks, i'd love to understand this stuff better

[08:22:20]

jya:

say your image is YUV, 720x576

[08:22:33]

jya:

the Y plan , each line is 720 wides

[08:22:46]

jya:

that’s good, and can sit on a 16 bytes boundary

[08:23:35]

jya:

but the U and V plan, are 310 pixels wide (the YUV is 12 bit pixel depth: 8 bits for Y, and 4 bits for U/V)

[08:23:50]

jya:

so now the first U line is 16 bytes aligned

[08:24:19]

jya:

the 2nd line isn’t (as 360 isn’t a multiple of 16)

[08:24:26]

stuarta:

ah makes sense

[08:24:28]

jya:

as typically you work line by line

[08:24:39]

jya:

your 2nd line isn’t aligned, your 3rd is, 4th isn’t etc..

[08:25:03]

jya:

when you call the FFmpeg sws_convert utility on our frame : you get a nice segfault

[08:25:27]

stuarta:

so if we supported, for example, putting 720x576 content into a slightly bigger container, where every line was aligned, then things would be better?

[08:25:59]

jya1:

stuarta: http://msdn.microsoft.com/en-us/library/windo . . . s.85%29.aspx

[08:26:15]

jya:

stuarta: yes, it would work much better.

[08:26:20]

stuarta:

ie. pad each 360 line to 368

[08:26:36]

jya:

in fact, I wouldn’t be surprised if some of the video filters found in ffmpeg 2.2, which are very heavily SSE optimised

[08:26:43]

jya:

aren’t likely to crash

[08:26:45]

stuarta:

that first picture is exactly what i was trying to describe

[08:27:09]

jya:

I modified mythframe.h, so the stride with create are all 16 bytes aligned

[08:27:28]

jya:

so the Y plan, is itself 32 bytes aligned (as it’s twice the resolution than U and V)

[08:27:50]

jya:

so you store your 720x576 image, in a 768x576 frame

[08:28:07]

stuarta:

makes sense

[08:28:11]

jya:

that displays well with Xv and Qt rendering, there’s full support for strides there

[08:28:38]

stuarta:

more importantly does it help us with VDPAU, VAAPI etc?

[08:28:43]

jya:

(except PiP, because the resize code itself is just a loop with memcpy, and ignore the stride side)

[08:29:10]

jya:

but the OpenGL render, doesn’t handle those. as it considers an image to be continuous, it all looks garbage

[08:29:21]

jya:

VDPAU, VAAPI it makes no difference

[08:29:32]

jya:

we never see the frames, it stays in the GPU

[08:29:45]

stuarta:

i wonder if the pip video frames could be composited into the underlying frames

[08:30:00]

jya:

it does help if we want to extract the frame from the GPU, because that frame, from my tests have strides != width

[08:30:27]

jya:

stuarta: the vdpau code does it that way: it renders the PiP directly in the smaller window

[08:30:34]

stuarta:

nice

[08:31:14]

jya:

but the VAAPI code, uses the basic render, it extracts the frame from the GPU, feeds it back as if it was a SW decoder, and the default PiP code does the rest

[08:31:30]

jya:

my guess is that MarkK did it for two reasons:

[08:31:57]

jya:

1- the old vaapi code could only work in a single OpenGL texture. That’s not the case anymore

[08:32:25]

jya:

2- the nullvaapi code is an obvious rip of the VLC code.. it’s almost identical line for line, and that’s the way they did it

[08:32:56]

stuarta:

i would have initially suspected #1

[08:32:56]

** MythLogBot http://code.mythtv.org/trac/ticket/1 **

[08:33:50]

jya:

tbh, I’m not sure it was fully tested, it just couldnt work before

[08:34:11]

jya:

because the frames returned is never the same as the frame we copy into

[08:34:42]

jya:

not unless it was tested with some videos that happened to have stride == width / 16-bytes aligned

[08:35:10]

jya:

hmm… 1920x1080 fits the bill, so it’s possible

[08:35:57]

jya:

allright, back to writing some unit test for my frame conversion, just to measure the speed

[08:36:08]

jya:

I went that far, better make use of the code!

[08:36:42]

stuarta:

:)

[08:37:19]

jya:

it all started because in the VLC mailling-list I saw someone submitting a patch to use new vaapi API..

[08:37:28]

jya:

opened a can of worms

[08:37:52]

stuarta:

and you just had to go and investigate. i know how that feels

[08:39:18]

jya:

it’s touching topics I knew nothing about, and that’s knowledge we need in the future if we don’t want myth to die a slow death.

[08:39:28]

jya:

still need to find someone with expertise in that area

[08:39:55]

jya:

I can fix Qt/Xv, OpenGL on the other hand....

[08:41:07]

jya:

it’s so weird that ffmpeg SSE optimised conversion tool do not work on the frame they can decode..

[08:41:09]

jya:

makes no sense

[08:55:52]

dekarl (dekarl!~dekarl@p4FCEFDC6.dip0.t-ipconnect.de) has joined #mythtv

[09:01:18]

jya:

hmmm… there’s a bug in the calculation of frame plane offset in mythframe.h.. how could this never cause an issue ?

[09:01:40]

stuarta:

luck

[09:03:40]

jya:

that routine is probably not used… can’t think of another reason.. was added in 2008

[09:56:26]

dekarl1 (dekarl1!~dekarl@p4FE859B9.dip0.t-ipconnect.de) has joined #mythtv

[09:57:49]

dekarl (dekarl!~dekarl@p4FCEFDC6.dip0.t-ipconnect.de) has quit (Ping timeout: 252 seconds)

[10:04:34]

paul-h (paul-h!~Paul@2.221.230.194) has joined #mythtv

[10:11:28]

paul-h:

stuartm: Don't suppose you know if the jqueryGrid can have images in the cells? Not finding anything in the docs to suggest it can have anything other than text

[10:14:48]

paul-h:

jya: I sent my email to the dev list before I saw yours that puts a new perspective on things :(

[10:15:05]

jya:

paul-h: it does

[10:15:27]

stuartm:

paul-h: does this help? http://stackoverflow.com/questions/1896192/ca . . . 8273#1898273

[10:15:32]

jya:

should have sent it earlier.. been in my inbox for a few days

[10:17:54]

paul-h:

stuartm: Looks promising I'll give it a try

[10:24:33]

stuartm:

jya: odd, I never received that email from warpme yet it was addressed to both of us?

[10:25:07]

jya:

it was set to stuart@taste..

[10:25:14]

jya:

sent

[10:25:24]

stuartm:

oops

[10:25:33]

dekarl1 is now known as dekarl

[10:25:47]

stuartm:

someone at taste.co.uk is probably a bit confused then

[10:26:04]

** stuarta chuckles **

[10:27:20]

jya:

what is that email?

[10:33:37]

stuartm:

should be stuart@tase.co.uk , not taste

[11:13:57]

seld_ (seld_!~seld@h7n7-rny-a12.ias.bredband.telia.com) has joined #mythtv

[11:14:28]

seld (seld!~seld@h7n7-rny-a12.ias.bredband.telia.com) has quit (Ping timeout: 265 seconds)

[11:14:30]

kormoc (kormoc!~kormoc@mythtv/developer/kormoc) has quit (Ping timeout: 265 seconds)

[11:14:30]

nyloc (nyloc!~quassel@nyloc.de) has quit (Ping timeout: 265 seconds)

[11:14:54]

kormoc_ (kormoc_!~kormoc@mythtv/developer/kormoc) has joined #mythtv

[11:15:01]

kormoc_ is now known as kormoc

[11:16:23]

nyloc (nyloc!~quassel@nyloc.de) has joined #mythtv

[11:16:23]

eee-blt (eee-blt!~nb0yjxtr@ma.sdf.org) has quit (Ping timeout: 265 seconds)

[11:16:54]

sraue (sraue!~stephan@xbmc/staff/sraue) has quit (Ping timeout: 265 seconds)

[11:23:22]

sraue (sraue!~stephan@xbmc/staff/sraue) has joined #mythtv

[11:23:22]

sraue (sraue!~stephan@xbmc/staff/sraue) has quit (Excess Flood)

[11:23:36]

ElmerFudd (ElmerFudd!~le@87-55-166-130-static.dk.customer.tdc.net) has quit (Ping timeout: 265 seconds)

[11:24:04]

nyloc (nyloc!~quassel@nyloc.de) has quit (Ping timeout: 265 seconds)

[11:24:06]

ElmerFudd (ElmerFudd!~le@87-55-166-130-static.dk.customer.tdc.net) has joined #mythtv

[11:28:22]

nyloc (nyloc!~quassel@nyloc.de) has joined #mythtv

[12:08:30]

jya1:

i found SSE optimised code in VLC to copy YV12 frames...

[12:08:56]

jya1:

it's 3.9 times slower than my plain C on my desktop

[12:19:04]

stuartm:

heh

[12:23:49]

stuarta:

doh

[12:30:51]

** stuarta builds new frontend **

[12:40:07]

sphery:

stuartm: Yeah, I did work on it in its original form. It's been modified a bit since then with the changes to UTC storage.

[12:41:25]

dekarl:

jya1 have you looked at liboil?

[12:41:26]

sphery:

I'm guessing you have questions about the myth proto usage... Ideally it would be gone, now, and we'd be using the services API to request mbe zone information, but we don't yet have a services/soap/json client with which to do so. :(

[12:41:40]

jya1:

dekarl: what's that?

[12:41:59]

dekarl:

a collection of "optimized inner loops" for stuff like colour space conversion etc

[12:42:06]

dekarl:

http://liboil.freedesktop.org/wiki/

[12:42:34]

jya1:

and you could only point to this now ? :)

[12:42:48]

sphery:

The biggest problem with the use of myth proto is that we have to create a temporary connection to check it since we don't yet have our real connection since it's done early in startup.

[12:43:47]

dekarl:

uhh, that would be due to ... nursing dementia :D http://liboil.freedesktop.org/documentation/

[12:44:59]

** dekarl wonders why every other projects seems to hide links to their documentation... **

[12:45:55]

jya1:

sphery: there should be a system connection in gCoreContext that can use for that type of things

[12:45:58]

jya1:

no need for new connections

[12:46:30]

jya1:

IMHO, using myth protocol and the helper in gCoreContext, at this stage is *much* easier and is very fast, and run in a different thread

[12:47:35]

jya1:

dekarl: i wonder if it's memcpy on mac that is extremely well optimised, I get massive speed improvement replacing the VLC copy with memcpy

[12:47:48]

sphery:

jya1: the time check is done before the context is set up

[12:47:57]

sphery:

which is why we need a temp connection\

[12:48:12]

sphery:

(and why we'd be better off not using myth proto and just using a services request)

[12:54:07]

aloril (aloril!~aloril@dsl-tkubrasgw2-54f80b-12.dhcp.inet.fi) has quit (Ping timeout: 265 seconds)

[12:55:09]

dekarl:

jya1: yes, I think OSX, the Intel compiler and some others have good implementations

[12:55:45]

jya1:

dekarl: did you check the presentation?

[12:55:57]

jya1:

they say to take ICC generated code and copy the assembly :)

[12:56:29]

stuartm:

sphery: personally I question whether the public Services API will ever be the right one for frontend/slave/backend communication but that's another conversation, right now I'd like to know if there was any particular reason why the timezone check makes it's own connection with a hostname of 'tzcheck' instead of the hostname of the fe/slave

[12:57:12]

stuartm:

making the connection I understand, just not why it's using a fake/invalid hostname

[12:57:34]

dekarl:

jya1: aye, quickly zapped through it

[12:59:07]

dekarl:

all this "we can do malloc/free/memcpy better then the host OS" leads to all kinds of "interesting" effects...

[13:00:15]

stuartm:

sphery: if there was no particular reason for it then I can fix the problem reported by Paul by switching it to use the hostname

[13:00:56]

aloril (aloril!~aloril@dsl-tkubrasgw2-54f80b-12.dhcp.inet.fi) has joined #mythtv

[13:01:36]

dekarl:

jya1: I'm following the clang/llvm lists and it makes me jealous of xcode users

[13:02:36]

stuartm:

one possibility is that it's to prevent spurious 'connection' system events, if that's the case I can fix it by adding a new arg to ANN instead

[13:02:45]

jya1:

dekarl: I don't get much good out of it with myth unfortunately. I only use it to debug and edit the code. So don't enjoy any of the auto-completion, compilation as you type etc

[13:02:58]

stuartm:

which seems neater than overloading the hostname

[13:03:27]

jya1:

it is nice that with just a few lines of python, you can display the Qt objects content

[13:03:42]

jya1:

done so for QString, QList, QMap etc

[13:06:31]

sphery:

stuartm: that part was all done by Anduin – https://github.com/MythTV/mythtv/commit/a96f7d22b83

[13:07:49]

sphery:

I don't know why he chose that (for better reporting of what the connection is for/that it's not the "real" connection from the system, or because it was easier than pulling the hostname before we had the context set up, or ...)

[13:07:59]

sphery:

Captain_Murdoch may have some ideas

[13:08:21]

sphery:

he understood the connection issue/how myth proto connections are supposed to be use better than I ever did

[13:09:11]

sphery:

though is there some reason why it's not closing properly or why the mbe isn't recognizing the close?

[13:09:22]

stuartm:

atm a frontend for example makes at least 2 connections, and when several transferring files from the backend it can be many more

[13:09:28]

sphery:

it should be very short lived

[13:12:38]

stuartm:

sphery: mbe was deliberately ignoring disconnects for 'tzcheck', which only affects my code because of where I chose to monitor the connects/disconnects, I could move my hook outside that if (hostname != 'tzcheck') but I thought I'd try to understand why it's doing what it is since it seemed unnecessary to be using a false hostname as it does – certainly won't help with logging because there's no way from the mbe log to tell which systems those

[13:12:40]

stuartm:

connections are coming from

[13:13:43]

sraue (sraue!~stephan@xbmc/staff/sraue) has joined #mythtv

[13:13:59]

stuartm:

since there's no obvious reason given in the original commit I'll just change it to use the hostname, don't expect anything to break as a result

[13:14:35]

sphery:

sounds like a plan

[13:15:45]

sphery:

oh, and as far as the services api usage--I'm not advocating using it for everything, but just for those things where we have to work around the myth protocol's design (such as the temporary monitor connections made for tz check and some other connections we do (don't we do similar for some file transfer or remote file stuff?))

[13:18:05]

sphery:

the current workarounds are just confusing, and removing basically every bit of code in [a96f7d22b83] and replacing it with a simple getJSON("http://mbe:6544/Myth/GetTimeZone"); would be a lot simpler, IMHO

[13:18:46]

sphery:

after all, stateless connections are perfect for replacing these temporary monitor connections

[13:18:56]

sphery:

(that don't receive events)

[13:23:25]

stuartm:

sphery: you may have a point there :)

[13:29:29]

jya1:

dekarl: same deal on linux ubuntu 14.04... SSE is 4.47x slower

[13:32:31]

stuartm:

Captain_Murdoch: I'm moving the CLIENT_CONNECTED/CLIENT_DISCONNECTED system event messages into backend context so that they are only sent once per client and not for every single client connection, does that sound reasonable?

[13:33:40]

stuartm:

so there will be one event for the first connection and one event for the last disconnection

[13:37:18]

eee-blt (eee-blt!~nb0yjxtr@ma.sdf.org) has joined #mythtv

[13:39:00]

Warped (Warped!~Warped@108-85-161-113.lightspeed.cicril.sbcglobal.net) has quit (Quit: ChatZilla 0.9.90.1 [Firefox 30.0/20140605174243])

[13:41:49]

jya1:

stuartm that sound good to me... what about having a keep alive method also, so the backend knows something went wrong with the backend

[13:52:10]

jya1:

dekarl: that is interesting... with the pure SSE copy, CPU usage is 98.4% with C code (that in itself runs 4 times faster) is at 135% CPU usage

[13:52:15]

jya1:

what do you think of it?

[13:55:13]

dekarl:

gtg and take care of the runners at http://www.jpmccc.de/

[13:55:24]

dekarl:

jya1: SSE code is using different units in the CPU

[13:55:41]

jya1:

that isn't monitored with top ?

[13:56:06]

stuarta:

cpu usage > 100% strikes me as being multithreaded

[13:56:29]

jya1:

stuarta it is... one player = two threads

[13:56:58]

jya1:

the conversion NV12->YV12 is done in the decoder thread

[13:57:00]

dekarl:

I don't think so. hyperthreading does funny things to CPU monitoring, too :)

[13:57:02]

dekarl:

laters

[13:57:42]

jya1:

so what are we better of using? the code that gives lower CPU usage, or the code that runs faster in unit tests?

[13:59:03]

stuartm:

jya1: makes most sense to me, but I don't know the background to those events been added in the first place and whether there was a good reason they were called for each socket opened

[13:59:26]

jya1:

something is buggy in the PiP code... if you play in a PiP an existing recording. When the recording in the PiP reaches the end, it just stops

[13:59:43]

jya1:

stuartm IRC, the backend answer with the version number

[13:59:45]

superm1 (superm1!uid4318@ubuntu/member/superm1) has joined #mythtv

[14:00:05]

jya1:

in the RemoteFile class, it tests the answer for every new transfer, and check if the backend is of the right version to continue

[14:00:32]

jya1:

(I'm no fan of those IMHO useless "security/safety" tests

[14:01:50]

stuartm:

jya1: the events in question are the ones for users to hook scripts into, so I'm not sure that the version tests have any bearing

[14:05:39]

jya1:

LOL... turned off the VAAPI code, playback + 4 PiP takes 15% less than with vaapi :)

[14:06:06]

jya1:

it's very buggy opengl PiP, you see regularly frames of one PiP showing into another

[14:06:25]

jya1:

there's a race there in the opengl context switching is my guess

[14:06:36]

jya1:

anyway... two days working on optimisations for nothing....

[14:06:56]

** jya1 off to bed **

[14:08:51]

Captain_Murdoch:

stuartm, sounds ok, I don't have a big preference one way or the other. If we can't differentiate between the apps making connections then there isn't much reason to alert every time.

[14:11:21]

stuartm:

Captain_Murdoch: we can only differentiate by hostname, and since each frontend can open multiple socket you'd see multiple CLIENT_CONNECTED for each, it would be especially noisy when browsing videos etc or anywhere we fetch images from the backend

[14:12:27]

Captain_Murdoch:

yeah, I agree. at one point we talked about sending the app name in the connect message to differentiate clients. based on your comments just now, I prefer your new solution. :)

[14:15:12]

stuartm:

changes pushed, will be doing the exact same for slaves at some point and mediaservers too

[14:24:50]

jya (jya!~jyavenard@mythtv/developer/jya) has quit (Quit: jya)

[14:28:30]

jya (jya!~jyavenard@mythtv/developer/jya) has joined #mythtv

[14:42:59]

andreaz (andreaz!~Andreaz@tmo-107-87.customers.d1-online.com) has joined #mythtv

[14:58:16]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has joined #mythtv

[15:07:46]

FabriceMG (FabriceMG!~Thunderbi@LCaen-156-54-30-212.w80-11.abo.wanadoo.fr) has quit (Quit: FabriceMG)

[15:33:39]

andreaz (andreaz!~Andreaz@tmo-107-87.customers.d1-online.com) has quit (Read error: Connection reset by peer)

[16:25:18]

laga (laga!~laga@h1626373.stratoserver.net) has quit (Ping timeout: 240 seconds)

[16:42:05]

jpabq (jpabq!~quassel@mythtv/developer/jpabq) has quit (Remote host closed the connection)

[16:58:37]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has quit (Ping timeout: 272 seconds)

[17:01:53]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has joined #mythtv

[17:02:10]

natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has joined #mythtv

[17:08:10]

Weaselweb (Weaselweb!~quassel@77-64-221-251.dynamic.primacom.net) has joined #mythtv

[17:15:49]

Steve-Goodey (Steve-Goodey!~steve@host86-134-63-27.range86-134.btcentralplus.com) has joined #mythtv

[17:18:33]

Weaselweb (Weaselweb!~quassel@77-64-221-251.dynamic.primacom.net) has left #mythtv ("http://quassel-irc.org - Chat comfortably. Anywhere.")

[17:20:06]

Steve-Goodey (Steve-Goodey!~steve@host86-134-63-27.range86-134.btcentralplus.com) has quit (Client Quit)

[17:20:47]

Steve-Goodey (Steve-Goodey!~steve@host86-134-63-27.range86-134.btcentralplus.com) has joined #mythtv

[17:36:58]

andreaz (andreaz!~andre_000@p5DCA3B0D.dip0.t-ipconnect.de) has joined #mythtv

[17:38:17]

Warped (Warped!~Warped@108-85-161-113.lightspeed.cicril.sbcglobal.net) has joined #mythtv

[17:39:47]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has quit (Ping timeout: 245 seconds)

[17:40:43]

Steve-Goodey (Steve-Goodey!~steve@host86-134-63-27.range86-134.btcentralplus.com) has quit (Quit: Konversation terminated!)

[18:00:17]

SteveGoodey (SteveGoodey!~steve@host86-134-63-27.range86-134.btcentralplus.com) has joined #mythtv

[18:47:39]

Jordack (Jordack!~Jordack@users.twp.ypsilanti.mi.us) has joined #mythtv

[19:15:00]

superm1 (superm1!uid4318@ubuntu/member/superm1) has quit (Ping timeout: 260 seconds)

[19:15:20]

superm1 (superm1!uid4318@ubuntu/member/superm1) has joined #mythtv

[19:24:32]

Steve-Goodey (Steve-Goodey!~steve@host86-134-63-27.range86-134.btcentralplus.com) has joined #mythtv

[19:41:28]

andreaz (andreaz!~andre_000@p5DCA3B0D.dip0.t-ipconnect.de) has quit (Read error: Connection reset by peer)

[19:52:29]

caelor:

stuartm – if you're making changes that affect mythmediaserver, be aware that #11862 (if it still affects master) may impact the ability to test

[19:52:29]

** MythLogBot http://code.mythtv.org/trac/ticket/11862 **

[19:54:33]

caelor:

of course, if you find it no longer affects master, then it's one less ticket that's needed...

[20:11:06]

Warped (Warped!~Warped@108-85-161-113.lightspeed.cicril.sbcglobal.net) has quit (Quit: ChatZilla 0.9.90.1 [Firefox 30.0/20140605174243])

[20:13:19]

jpabq (jpabq!~quassel@97-123-207-7.albq.qwest.net) has joined #mythtv

[20:13:20]

jpabq (jpabq!~quassel@97-123-207-7.albq.qwest.net) has quit (Changing host)

[20:13:20]

jpabq (jpabq!~quassel@mythtv/developer/jpabq) has joined #mythtv

[20:15:52]

gigem (gigem!~david@mythtv/developer/gigem) has quit (Quit: WeeChat 0.4.3)

[20:20:44]

Steve-Goodey (Steve-Goodey!~steve@host86-134-63-27.range86-134.btcentralplus.com) has quit (Quit: Konversation terminated!)

[20:21:10]

gigem (gigem!~david@mythtv/developer/gigem) has joined #mythtv

[20:21:16]

Warped (Warped!~Warped@108-85-161-113.lightspeed.cicril.sbcglobal.net) has joined #mythtv

[20:22:51]

paul-h:

caelor: That bug was still a problem last time I tried the mythmediaserver and I've not seen anything to suggest that bug has been fixed

[20:29:40]

SteveGoodey (SteveGoodey!~steve@host86-134-63-27.range86-134.btcentralplus.com) has quit (Quit: Konversation terminated!)

[20:39:53]

unforgiven512 (unforgiven512!~unforgive@oxymorphone.unforgivendevelopment.com) has quit (*.net *.split)

[20:39:54]

Seeker (Seeker!~cjo20@unaffiliated/seeker) has quit (*.net *.split)

[20:39:54]

coling (coling!~colin@cpc7-sgyl36-2-0-cust267.18-2.cable.virginm.net) has quit (*.net *.split)

[20:39:54]

wagnerrp (wagnerrp!~wagnerrp_@mythtv/developer/wagnerrp) has quit (*.net *.split)

[20:39:57]

Captain_Murdoch (Captain_Murdoch!~cpinkham@mythtv/developer/CaptainMurdoch) has quit (*.net *.split)

[20:39:59]

Jordack (Jordack!~Jordack@users.twp.ypsilanti.mi.us) has quit ()

[20:40:01]

coling (coling!~colin@cpc7-sgyl36-2-0-cust267.18-2.cable.virginm.net) has joined #mythtv

[20:40:10]

Seeker (Seeker!~cjo20@host86-150-48-131.range86-150.btcentralplus.com) has joined #mythtv

[20:40:11]

Seeker (Seeker!~cjo20@host86-150-48-131.range86-150.btcentralplus.com) has quit (Changing host)

[20:40:11]

Seeker (Seeker!~cjo20@unaffiliated/seeker) has joined #mythtv

[20:40:19]

Captain_Murdoch (Captain_Murdoch!~cpinkham@c-67-170-1-245.hsd1.wa.comcast.net) has joined #mythtv

[20:40:20]

Captain_Murdoch (Captain_Murdoch!~cpinkham@c-67-170-1-245.hsd1.wa.comcast.net) has quit (Changing host)

[20:40:20]

Captain_Murdoch (Captain_Murdoch!~cpinkham@mythtv/developer/CaptainMurdoch) has joined #mythtv

[20:40:22]

unforgiven512 (unforgiven512!~unforgive@oxymorphone.unforgivendevelopment.com) has joined #mythtv

[20:40:32]

wagnerrp (wagnerrp!~wagnerrp_@mythtv/developer/wagnerrp) has joined #mythtv

[20:47:56]

caelor:

I was the originator of the ticket, and it still affected me last time I tried, but if there's work on changing the connect/disconnect logic, then a different code path may make it no longer apply

[20:48:35]

caelor:

that said, I think the changes are at the event notification level, and the bug is at a lower level network connection level

[20:59:28]

andreaz (andreaz!~andre_000@p5DCA3B0D.dip0.t-ipconnect.de) has joined #mythtv

[21:02:58]

stuarta:

really must fix that, if you haven't got the qt mysql driver installed we don't provide a nice error message, we just go around in a loop of barf

[21:04:52]

stuarta:

and the auto join throws an error on the backend

[21:05:11]

stuarta:

"HttpServer73 servicehost.cpp:143 (Invoke) – MethodInfo::Invoke – An Exception Occurred: Not Authorized"

[21:18:25]

paul-h:

Those unhandled exceptions are a pain in the arse. I looked at getting the master backend to startup without a DB which was easy enough except the media server would throw exceptions causing the BE to abort if the wrong thing was passed to the services api

[21:18:27]

unforgiven512 (unforgiven512!~unforgive@oxymorphone.unforgivendevelopment.com) has quit (*.net *.split)

[21:18:28]

kc (kc!~Casper@unaffiliated/kc) has quit (*.net *.split)

[21:18:28]

moparisthebest (moparisthebest!~quassel@cpe-74-140-228-202.swo.res.rr.com) has quit (*.net *.split)

[21:18:29]

poptix (poptix!poptix@poptix.net) has quit (*.net *.split)

[21:18:33]

poptix (poptix!poptix@poptix.net) has joined #mythtv

[21:18:51]

kc (kc!~Casper@pool-71-185-186-159.phlapa.fios.verizon.net) has joined #mythtv

[21:18:51]

kc (kc!~Casper@unaffiliated/kc) has joined #mythtv

[21:18:51]

kc (kc!~Casper@pool-71-185-186-159.phlapa.fios.verizon.net) has quit (Changing host)

[21:19:25]

stuarta:

paul-h: you still got the code for allowing the backend to startup without a db? that's something i was keen to get working, so that then you can bootstrap the backend via the webserver

[21:20:55]

unforgiven512 (unforgiven512!~unforgive@2600:3c03:e001:281a::31) has joined #mythtv

[21:21:34]

moparisthebest (moparisthebest!~quassel@cpe-74-140-228-202.swo.res.rr.com) has joined #mythtv

[21:22:42]

paul-h:

I think I still have it stashed away somewhere. That's why I started to look at it so the BE could be started without a DB allowing the webfrontend to be used to create the DB and DB user etc

[21:23:50]

stuartm:

paul-h: dblain believes the exception crashes are a QT issue, there is an exception handler but it doesn't catch them, so he was going to replace the throw/exception stuff with a different error handling mechanism

[21:24:37]

paul-h:

At the moment they go unhandled and cause an abort

[21:24:58]

stuartm:

I'll ask him about what he had in mind so that we can start working on it, because we can't think about a release as long as using the WebFrontend runs a risk of the backend crashing in the event of an error

[21:25:21]

stuarta:

paul-h: we need to create a branch to track that work

[21:25:59]

stuartm:

paul-h: yeah, as I said, there is a handler, but perhaps because QT is running the QtScript stuff in another thread (just speculation) our handler never catches them

[21:26:54]

stuarta:

sheesh our test sound is static?

[21:28:24]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has joined #mythtv

[21:54:11]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has quit (Ping timeout: 252 seconds)

[21:59:47]

natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has quit (Ping timeout: 245 seconds)

[22:01:21]

jya:

caelor: have you tried with 0.27.1 ? there was some changes on how reference counting are managed for the connection handler.

[22:01:24]

jya:

done to prevent just that

[22:01:34]

** jya so someone does use mythmediaserver ! youhou **

[22:02:06]

andreaz (andreaz!~andre_000@p5DCA3B0D.dip0.t-ipconnect.de) has quit (Read error: Connection reset by peer)

[22:02:56]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has joined #mythtv

[22:08:38]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has quit (Ping timeout: 240 seconds)

[22:08:44]

stichnot1 (stichnot1!stichnot@nat/google/x-semxobwbqgkowhoj) has joined #mythtv

[22:15:55]

stichnot1 (stichnot1!stichnot@nat/google/x-semxobwbqgkowhoj) has quit (Ping timeout: 272 seconds)

[22:18:39]

stichnot1 (stichnot1!~stichnot@216.239.45.130) has joined #mythtv

[22:25:17]

paul-h (paul-h!~Paul@2.221.230.194) has quit (Quit: Konversation terminated!)

[22:41:53]

stichnot1 (stichnot1!~stichnot@216.239.45.130) has quit (Ping timeout: 272 seconds)

[22:46:13]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has joined #mythtv

[23:14:18]

stichnot1 (stichnot1!~stichnot@adsl-68-125-54-22.dsl.pltn13.pacbell.net) has quit (Ping timeout: 264 seconds)

[23:14:26]

stichnot1 (stichnot1!stichnot@nat/google/x-mcjshdfjleoqlphn) has joined #mythtv

[23:15:50]

knightr (knightr!~Nicolas@69-165-170-178.dsl.teksavvy.com) has joined #mythtv

[23:15:51]

knightr (knightr!~Nicolas@69-165-170-178.dsl.teksavvy.com) has quit (Changing host)

[23:15:51]

knightr (knightr!~Nicolas@mythtv/developer/knightr) has joined #mythtv

MythLogBot@irc.freenode.net :: #mythtv

Current users (65):