Back to Home / #mythtv / 2002 / 11 / Prev Day | Next Day
#mythtv IRC Logs for 2002-11-23

00:20<mdz_>witten: via an output from the sound card?
00:21<Chutt>computer's line out to my receiver
00:21<mdz_>witten: personally, I have my computer connected to my receiver, not the TV
00:21<witten>hmm ok
00:21<Chutt>tv speakers suck
00:21<Chutt>as a general rule
00:21<mdz_>TV speakers generally don't sound so good
00:21<witten>but not everyone has a bitching sound system :)
00:21<mdz_>knock it off, chutt, you keep beating me to the punch
00:22<witten>so how would one go directly from a pc to a tv for audio, if one was so inclined?
00:22<mdz_>witten: even cheap speakers sound better than the speakers built into most TVs
00:22<Chutt>mdz, oh, changing things to do the epg from inside the tv class was harder than i thought, so i went and did it =)
00:22<witten>last I checked most TVs don't have 1/4" audio inputs :)
00:22<mdz_>Chutt: you went ahead and did it because it was harder?
00:22<Chutt>mine does
00:22<Chutt>mdz, yeah, qt threading issues
00:22<witten>Chutt: heh ok
00:22<mdz_>Chutt: 1/4", really?
00:23<Chutt>well, no
00:23<Chutt>it has rca
00:23<Chutt>but, that's just a cable adaptor
00:23<mdz_>witten: why do you want 1/4" inputs? I assume your sound card has 1/8" connectors
00:23<witten>mdz: er, 1/8". that's what I meant
00:23<mdz_>witten: as Chutt says, you can get a cable which goes from 1/8" stereo to dual RCA
00:23<mdz_>witten: at pretty much any computer or stereo store
00:24<witten>okay cool
00:24<witten>just wanted to make sure there was a way
00:24<witten>now the only thing I haven't found is how to do TV out in linux
00:24<witten>video, that is
00:27<mdz_>then you've gotten to the hard part
00:27<witten>it would just pain me to plunk down $125 on an external scan convert if there's some pci card that can do it in linux
00:29<mdz_>an external converter will more than likely work out of the box, though
00:30<mdz_>and I can say with some confidence that you will not have an easy time getting TV output on a card to work under Linux
00:30<witten>even with proprietary vendor-supplied drivers?
00:30<mdz_>Chutt: I was thinking about taking a stab at per-recording quality settings this weekend if I have some time
00:30<mdz_>Chutt: unless you want to avoid a database change
00:30<Chutt>that'd be appreciated
00:30<Chutt>naw, go for it
00:31<mdz_>witten: if you find something which works easily, a lot of people would like to know about it
00:31<Chutt>the nvidia binary tv out stuff works easily
00:31<witten>mdz: what chutt said
00:32<mdz_>I hear a lot of complaints from users who have problems getting the drivers to work
00:32<mdz_>which isn't to say it's entirely the drivers' fault, but it doesn't seem trivial for everyone
00:32<witten>it's true
00:32<mdz_>Chutt: you may not represent an entry level Linux user :-)
00:32<witten>the nvidia drivers can be a bitch
00:32<witten>especially with agp
00:32<Chutt>i'm not an entry level user?
00:33<mdz_>I've revoked your membership card
00:38<Chutt>i'm still pondering how to break stuff apart
00:38<mdz_>which stuff?
00:38<Chutt>the multiple boxes support
00:39<Chutt>of course, thinking about code doesn't really happen while i'm playing metroid =)
00:40<mdz_>better make sure that the database config stuff has a 'host' field when it goes in :-)
00:41<Chutt>it's probably going to be easiest to make even the single computer case semi-separated
00:41<mdz_>will it be a requirement that all mythtv boxes sharing a database also share the same RecordFilePrefix?
00:42<Chutt>i just want a player to say to the master box 'give me this file', and it tells the player where it is, etc
00:42<Chutt>the only time you really need close 2 way communication is while watching live tv
00:44<lichen_>wow, watching live tv over the network.. brilliant!
00:45<mdz_>Chutt: I assume capture still only happens locally, though :-)
00:47<witten>is there such a thing as an svideo to coax converter? and is it just a cheap cable?
00:47<mdz_>are you thinking of separating out the single computer case just from a code structure perspective, or running independent processes?
00:48<Chutt>probably independent processes
00:48<Chutt>which'll also have the nice side effect of fixing all the smp issues in libavcodec
00:48<mdz_>witten: I don't know, but my guess would be no
00:48<witten>hmm ok
00:48<Chutt>just have a backend
00:48<Chutt>and a frontend
00:48<mdz_>Chutt: or providing an excuse not to fix the SMP issues in libavcodec
00:48<witten>well then a card with svideo out doesn't do me any good
00:49<Chutt>that too
00:49<mdz_>witten: you don't have composite input even?
00:49<Chutt>i just need to sit down and write up a simple little communications protocol
00:50<witten>mdz: what does composite input look like?
00:50<mdz_>for the single computer case, I'd much rather have the communication of bulk data over the filesystem than some IPC mechanism
00:50<mdz_>witten: an RCA connector
00:50<witten>mdz: oh, yeah, I'm sure the tv has rca inputs
00:51<mdz_>witten: use that, then
00:51<Chutt>mdz, right
00:51<witten>oh, there are cheap s-video to rca cables?
00:51<Chutt>as it is right now, there's very little data that's communicated between the player and the recorder
00:51<Chutt>that i can just do via sockets, and have everything else be files
00:52-!-lichen_ [] has quit ["thats what life is, one big series of down notes"]
00:52<Chutt>it'll complicate the ringbuffer a _little_ bit
00:52<mdz_>so to watch a recording in progress, it would just communicate the path to the file
00:52<Chutt>but not all that much
00:52<Chutt>it'll make live-tv more like that, really
00:52<witten>mdz/chutt: why don't you guys use NFS or something? :)
00:52<Chutt>because i don't want to use nfs
00:52<mdz_>in fact they might even be exactly the same
00:53<Chutt>well, live tv needs to know that it's a ring buffer
00:53<mdz_>on the player side anyway
00:53<Chutt>and when to wrap around
00:53<Chutt>other than that, yup.
00:53-!-witten [] has quit ["ircII EPIC4-1.1.2 -- Are we there yet?"]
00:53<mdz_>why should the player need to know? shouldn't it just be able to read the buffer forever and never come to the end?
00:53<Chutt>if it's a file
00:54<mdz_>I'm getting a hang when some recordings get previewed
00:54<mdz_>I thought it might be one problematic recording, but it just happened with another apparently
00:55-!-lichen_ [lichen@] has joined #mythtv
00:55<mdz_>seems to be easily reproducible, so I'm going to take a look at it now
00:55<mdz_>this is with CVS from 2 days ago
00:55<mdz_>didn't seem to happen with 0.7, but I haven't switched back to try it
00:55<Chutt>i can hang it myself
00:56<mdz_>maybe I don't have to rebuild for debugging then :-)
00:56<Chutt>not in gdb, though
00:58<mdz_>I am attached to one that's hung
00:58<mdz_>it's weird though
00:58<mdz_>when I attach, gdb apparently sends itself a SIGTSTP and it gets suspended :-)
00:58<mdz_>Reading symbols from /lib/ debugging symbols found)...done.
00:58<mdz_>Loaded symbols for /lib/
00:58<mdz_>[2]+ Stopped gdb /usr/bin/mythfrontend 17382
00:59<Chutt>have you heard of a bug in libc or libpthreads in debian?
00:59<mdz_>one thread is sleeping in #3 0x08054ec0 in QScrollView::mouseReleaseEvent ()
00:59<Chutt>static bins segfault in threads in libc functions?
00:59<mdz_>a lot of statically linked binaries break with glibc 2.2->2.3
00:59<Chutt>newly compiled
00:59<mdz_>all due to NSS issues that I know of, though
00:59<mdz_>no, haven't seen that
01:00<mdz_>why would it sleep on a mouseReleaseEvent?
01:00<Chutt>which thread, though?
01:00<mdz_>can't tell without debugging symbols
01:01<mdz_>thread 4 :-)
01:01<Chutt>i got it, i think.
01:01<mdz_>crap, I didn't build a source package for this CVS snapshot
01:01<mdz_>it happens when scrolling down the list of recorded programs
01:02<Chutt>i think i've got it
01:02<mdz_>doesn't seem to always happen on the same program, but after a certain amount of frames have been displayed or something
01:02<Chutt>race condition
01:02<Chutt>fixing it now..
01:02<mdz_>if moving down the list at the same speed every time, it tends to hit at the same time
01:02<mdz_>sound like the same bug I see?
01:03<mdz_>back in a bit
01:03<Chutt>i'll have a fix checked in shortly
01:37<blahblechblam>Everything is configured now. The channels are loaded and the program listings are loaded. Now I get a lime green screen when running mythtv
01:38<blahblechblam>On any channel.
01:41<lichen_>i was just watching tivo
01:41<lichen_>its so smooth and nice :(
02:06<blahblechblam>I get a lot of these errors "VIDIOCMCAPTURE1: Invalid argument"
03:29-!-witten_ [] has quit [Ping timeout: 14400 seconds]
05:36-!-nevertheless [] has joined #mythtv
07:09<-- blahblechblam( has left #mythtv
09:18-!-Universe [] has quit [Read error: 104 (Connection reset by peer)]
10:32-!-sefudier [] has quit ["Leaving"]
11:31-!-nevertheless [] has quit [Remote closed the connection]
11:55-!-nevertheless [] has joined #mythtv
12:40-!-nevertheless [] has quit [Remote closed the connection]
12:47-!-nevertheless [] has joined #mythtv
13:10-!-hadees [] has joined #mythtv
13:15-!-nevertheless [] has quit [Remote closed the connection]
13:23-!-nevertheless [] has joined #mythtv
13:34<mdz_>Chutt: around?
13:34<mdz_>I'm trying to work out when it should load the recording parameters from the db
13:35<Chutt>meaning, either in the tv class, or in the frontend?
13:35<mdz_>which method of programinfo, actually
13:36<mdz_>or applyrecordstatechange?
13:36<Chutt>not there
13:36<Chutt>or there =)
13:36<mdz_>by the way, what does query.exec(foo) do when query is a newly initialized QSqlQuery?
13:36<mdz_>where does it get a database connection from?
13:36<Chutt>if it doesn't specify a database connection, it uses the default connection
13:37<Chutt>static variable in the query object
13:37<mdz_>I already have TV loading the parameters based on a profile id
13:37<mdz_>I just need to figure where to load the profile id for a particular recording
13:38<mdz_>(which I added as a column on each recording table)
13:38<Chutt>so you added a recordingoptions table
13:38<Chutt>to each recording table
13:38<hadees>anyone working on a news/stock/whatever ticker?
13:38<Chutt>hadees, no.
13:38<mdz_>no, I added a recordingprofile column to each recording table
13:38<Chutt>mdz, can you do a quick diff?
13:38<mdz_>which references a recordingprofile table
13:39<hadees>any reason or just cause no one wants too?
13:39<Chutt>hadees, i don't particularly want to
13:39<Chutt>if someone else does, i don't know
13:40<hadees>hehe, just wondering, i think i might give it a crack
13:40<mdz_>hadees: why would you want that to be part of mythtv?
13:40<hadees>why not?
13:40<mdz_>there are already plenty of those sorts of programs which could be run next to mythtv
13:41<hadees>i want it so it looks like it does on like msnbc
13:41<hadees>where its ontop of the image as you watch tv
13:41<hadees>i like to keep up to date with news
13:41<mdz_>so put a KDE kicker or GNOME panel on top of mythtv
13:42<Chutt>mdz, so, hm
13:43<hadees>i guess that would work
13:43<hadees>just thought it would be nice
13:43<Chutt>mdz, what'd i would do is create a 'default' encoding quality
13:44<hadees>from the standpoint of having a machine that only does mythtv so you install just mythtv and nothing else
13:44<Chutt>check if the programinfo has a profile set or not
13:44<Chutt>and fall back to that
13:44<mdz_>I thought we talked about this and agreed that the defaults should just be a profile
13:44<Chutt>in the tv class
13:44<Chutt>so it is just a profile
13:45<mdz_>are you saying that the profile in the recording should be able to be null?
13:45<mdz_>and then we load the default profile?
13:45<mdz_>I was thinking profile would just default to the id of the default profile
13:45<mdz_>and when converting the db, just set them all to that
13:45<mdz_>or make it always id 0
13:45<Chutt>that works too
13:46<Chutt>as for setting the profile id otherwise
13:46<Chutt>that's your question?
13:47<Chutt>probably just add a helper function to the programinfo class
13:47<mdz_>my question was about loading the actual profile column
13:47<Chutt>ah, hm
13:47<mdz_>it seems like it would belong in getprogramrecordingstatus except that is returning a value
13:47<Chutt>in the scheduler
13:48<Chutt>getprogramrecordingstatus is only used in the epg, i believe
13:48<mdz_>which you are apparently using in the epg
13:48<Chutt>the scheduler does all the loading of data into programinfo classes that actually get used for recording
13:49<mdz_>hmm, the scheduler is part of the frontend, yes?
13:49<mdz_>this seems like it belongs in the lib
13:49<Chutt>soon to be part of the backend
13:50<mdz_>ah, I see it in scheduler now
13:51<Chutt>sorry for not understanding your question right off =)
13:51<Chutt>bbiab, though
13:51<mdz_>fillercordlists only handles timeslotrecording
13:51<mdz_>oh, no it doesn't.ok
14:11-!-nevertheless [] has quit [Remote closed the connection]
14:18-!-nevertheless [] has joined #mythtv
16:08-!-witten [] has joined #mythtv
16:38-!-jason_ [] has joined #mythtv
16:41<Chutt>nevertheless, are you here?
16:41<Chutt>can you change something with that shutdown patch?
16:41<Chutt>i'd rather have the shutdown logic in the mythfrontend's main.cpp, not themedmenu.cpp
16:42<nevertheless>hmm, ok, ill have a look
16:42<Chutt>just the stuff that does the dialog
16:43<Chutt>should be a fairly simple modification
16:43<Chutt>other than that, looks goo
16:43<nevertheless>ok, how is it about the 3 lines making a new scheduler? is there a way to get rid of em?
16:44<Chutt>ah, not really
16:44<nevertheless>ok, i just thought its silly to make a new one at the end of the program, but it is quite fast though
16:44<Chutt>yeah, it shouldn't be that big of a deal
16:46<Chutt>it _is_ kinda silly, but that's the way it works =)
16:46<jason_>Hi. I'm getting a bright green screen when I start MythTV on Redhat 8.0. Any suggestions?
16:47<nevertheless>its a year ago that i worked with pthreads so I do not remember in detail how the interaction could be between the treads, but I remember this beeing complicated
16:47<mdz_>Chutt: I've got the profile stuff working
16:47<mdz_>Chutt: no way to change it from the UI, though...what are your thoughts on that?
16:47<Chutt>jason, any error messages?
16:47<Chutt>mdz, cool
16:48<Chutt>mdz, not sure, exactly
16:48<mdz_>I wrote a little migration program to take the old settings and stuff them into the database as a default profile
16:48<Chutt>i'm thinking the fix recording conflicts screen could do it
16:48<mdz_>and do the alter tables
16:48<Chutt>have like a radio button on top 'fix conflicts' 'edit record settings'
16:49<Chutt>left/right arrow change that, up/down change programs as currently
16:49<mdz_>what about letting the user select a profile when selecting a program to record?
16:49<Chutt>how would you add that with the current interface?
16:49<jason_>I'm getting errors like: "VIDIOCMCAPTURE1: Invalid argument"
16:50<Chutt>jason, capture card?
16:50<jason_>All-in-Wonder Rage 128
16:50<Chutt>it's unsupported.
16:50<mdz_>Chutt: dunno, would it require extending DialogBox?
16:50<jason_>I should buy a Hauppage WinTV then?
16:51<Chutt>mdz, not thinking of implementation, but just the ui
16:51<mdz_>Chutt: ah
16:51<Chutt>how it'd work
16:51<mdz_>Chutt: I was thinking a radio button for the profile, and then a button (as it is now) for each type of recording
16:51<Chutt>it's not a button now =)
16:51<mdz_>sure looks like a button :-)
16:51<Chutt>that's a select box
16:51<Chutt>list box
16:52<Chutt>most often, though, people will use the default settings, right?
16:52<Chutt>and you could probably come up with a bunch of different profiles
16:52<mdz_>I guess so
16:52<Chutt>so you'd need to present them in a list
16:52<mdz_>it also occurred to me that this makes it easy to have different settings for live TV vs. scheduled recording
16:53<Chutt>two lists on one page doesn't quite work
16:53<mdz_>trivial even
16:53<Chutt>so if it were on the same page as the select recordings box, you'd need to click through 2 pages to select a recording
16:53<mdz_>is Markie's configuration stuff ready? we'll need a UI for configuring the profiles too
16:53<Chutt>i had to rewrite most of what he sent in
16:54<Chutt>the framework's there, though
16:55<Chutt>to add in more config modules
16:55<Chutt>well, not modules, but dialogs
17:13<nevertheless>Chutt: hmmm, ok, i checked it now, but I have no idea how to accomplish your request. how would i move the stuff to the main.cpp?
17:29<nevertheless>well, i have a solution now
17:29<nevertheless>which i don't relly like
18:04<witten>I've got my mythtv HTPC all spec'd out
18:04<witten>but I was curious.. what kinda cpu power is needed for realtime tv watching with mythtv at 352x240?
18:05<nevertheless>my tbird 800 runs at about 30-60 % cpu
18:05<nevertheless>at 384x288
18:05<witten>wow cool
18:05<witten>IDE harddrive?
18:06<nevertheless>ide with UDMA 66
18:06<nevertheless>hdparm tells 45MByte/sec
18:07<witten>well I think I've got to tone down these specs a bit then
18:07<nevertheless>ill go for a XP 2000+, so i have a little space left
18:08<witten>those are like $90
18:09<witten>a 1.1 ghz tbird is $34
18:09<nevertheless>i don't get tbirds anymor here in germany
18:09<witten>how about durons?
18:10<nevertheless>i don't run a duron on a system which only lives from cpu power
18:11<nevertheless>there you could take the C3 too, this is bad stuff
18:11<witten>hehe true
18:13<witten>anyone willing to sell me a home-brew infrared receiver?
18:18<Markie>witten: heh
18:18<Markie>mdz: yea chutt fixed all my crap..
18:19<Markie>i bet he cant wait for my next set of changes :^)
18:35<mdz_>Chutt: what do we do about getting the default profile set up on a new installation (not upgrade)?
18:36<mdz_>Chutt: add it in mc.sql?
18:42<Chutt>mdz, sure
18:42<Chutt>if it can be added there
18:45<mdz_>Chutt: mailed you the latest diff, displays the profile name in the conflicts screen
18:57<Chutt>all looks good
19:26<mdz_>I don't really want to check this in until there is a way for the settings to be changed, though
19:27<mdz_>otherwise that functionality is lost
19:28<mdz_>where is the new setup stuff?
19:29-!-hadees [] has quit [Read error: 110 (Connection timed out)]
19:30<mdz_>I see the menu XML, but nothing attached to it
19:43-!-pml [] has joined #mythtv
19:50-!-pml [] has quit [Remote closed the connection]
19:58<Chutt>nothing is yet
19:59<Chutt>except for the theme switching
19:59<Chutt>and that doesn't get saved at all
20:02<mdz_>grr, I just got a crash during playback
20:02<mdz_>slice end not reached but screenspace end (6 left EC3BEF)
20:02<mdz_>concealing errors
20:02<mdz_>Bus error
20:03<Chutt>that'd be in libavcodec somewhere
20:03<Chutt>reproducible at all?
20:04<mdz_>trying it now with debug
20:04-!-jason_ [] has quit ["Client Exiting"]
20:05<mdz_>seeking isn't working right for me all the time
20:05<mdz_>if I seek twice in quick succession, it doesn't start playing again
20:06<mdz_>and if I seek while it is in that state, it only advances 1 second
20:06<mdz_>it's especially apparent when it's under gdb and built slow for debugging
20:08<Chutt>you can't seek again until it's done now
20:09<mdz_>that wouldn't be a problem except that seeking seems to take a long time
20:09<mdz_>I'll look at that when I've tried to reproduce this crash
20:12<mdz_>hmm, didn't happen for me at the same point in the recording
20:12<mdz_>I did get these, though
20:12<mdz_>1. marker bit missing in 3. esc
20:12<mdz_>Error at MB: 106
20:12<mdz_>concealing errors
20:12<mdz_>I don't _think_ the file is corrupted
20:13<Chutt>no idea, unfortunately :(
20:14<mdz_>hmm, it's saying it has an invalid seektable
20:14<mdz_>that would explain the poor seeking behaviour
20:15<Chutt>which error?
20:15<Chutt>maybe something killed the recording before it could be written completely?
20:15<Chutt>i've had a few reports of seektables not getting written
20:15<Chutt>but from noone who was able to debug things at all
20:16<mdz_>the file is pretty close to the right size
20:16<mdz_>haven't watched it all the way through yet
20:17<mdz_>looking at the end of the file, it sure looks like there's a seektable there
20:17<mdz_>bunch of Q frames
20:17<Chutt>maybe the offset's wrong?
20:18<mdz_>hmm, there's supposed to be one
20:18<mdz_>I can't tell if they're valid frames, of course, just looking at a dump
20:19<mdz_>I can't reproduce that error from libavcodec though
20:20<mdz_>playing over the same spot where it happened
20:21<mdz_>position saving doesn't seem to be working...does that only work with a seektable?
20:21<mdz_>I don't see why it would...
20:21<Chutt>just faster to seek to the point
20:22<mdz_>so it doesn't even save the position if there is no seektable?
20:22<Chutt>if you can figure out why its not seeing the seektable, that'd be great
20:22<mdz_>or if it really has a seektable
20:23<mdz_>it looks like it probably doesn't
20:24<mdz_>(gdb) print extradata.seektable_offset
20:24<mdz_>$1 = 2150842253
20:24<Chutt>does it actually say 'invalid seektable'?
20:24<mdz_>the file is 2150902721 bytes
20:24<mdz_>yes, it does
20:25<mdz_>that offset looks pretty reasonable to me
20:25<Chutt>ok, so that error says there's no Q at that offset
20:25<mdz_>I'm going to find out what is there
20:26<Chutt>is there a Q?
20:26<Chutt>well, anywhere near there
20:26<mdz_>$2 = {frametype = -1 '\xFF', comptype = -5 '\xFB', keyframe = 120 'x',
20:26<mdz_> filters = 100 'd', timecode = -1007546624, packetlength = -385756380}
20:27<mdz_>there's no Q in the entire thing it thinks is the frame header
20:27-!-sefudier [] has joined #mythtv
20:28<mdz_>hmm, actually there is a Q as the very next character in the file
20:28<mdz_>but not in the block that was read
20:29<Chutt>the very next character?
20:29<Chutt>after that frameheader?
20:29<mdz_>might be that character
20:29<mdz_>I used tail -c +<num>
20:29<mdz_>I think that counts from 1
20:30<mdz_>mizar:[~] tail -c +2150842253 /space/video/mythtv/2126_20021117180500_20021117193000.nuv | od -A d -c |head -1
20:30<mdz_>0000000 213 Q \0 \0 \0 \0 \0 \0 \0 ( \xEC \0 \0 224 \t \0
20:30<mdz_>so I think 213 is at offset ...52 and the Q at ...53
20:31<mdz_>I think I know what it is
20:31<Chutt>might it not be seeking far enough?
20:31<mdz_>there are no LFS flags on the compiler command line
20:31<Chutt>_GNU_SOURCE defines all those
20:31<mdz_>are you sure?
20:32<mdz_>this flie is just over 2GB
20:32<Chutt>unless that's changed in the past few months
20:32<mdz_>and the stuff it's reading does not seem to be the stuff at the right offset, even though it tries to seek there
20:32<mdz_>how big is extradata.seektable_offset?
20:32<Chutt>long long
20:36<Chutt>i dunno
20:36<Chutt>maybe it's not redirecting lseek to lseek64 like it should be
20:36<Chutt>try renaming the function call in ringbuffer
20:37<mdz_>(gdb) print ringBuffer->Seek(0, 1)
20:37<mdz_>$9 = 1144
20:37<mdz_>that's after it's supposed to have seeked to the seek_frameheader
20:37<mdz_>it never went anywhere
20:37<mdz_>does it check the return code on seek?
20:38<Chutt>go into ringbuffer.cpp and change all lseek calls to lseek64
20:38<Chutt>see if that helps
20:39<mdz_>some thing
20:40<Chutt>you changed all lseeks?
20:40<Chutt>sure you ran that binary? =)
20:40<mdz_>`/usr/bin/mythfrontend' has changed; re-reading symbols.
20:40<mdz_>it was different than the one I ran before at least
20:41<Chutt>there's 4 lseeks
20:41<Chutt>the 3rd one is the important one
20:41<mdz_>I changed all 4
20:41<mdz_>aw, crap
20:41<mdz_>I had two RingBuffer buffers open
20:43<mdz_>I did install the binary I built though :-P
20:43<mdz_>it just didn't have the change
20:43<mdz_>(gdb) print ringBuffer->Seek(0, 1)
20:43<mdz_>$1 = 2150842253
20:43<mdz_>works now
20:43<Chutt>it should seek fast now, too
20:44<Chutt>and that explains the couple bug reports i've had
20:44<mdz_>g++ -c -pipe -Wpointer-arith -Wwrite-strings -Wmissing-prototypes -Wall -W -g `freetype-config --cflags` -D_REENTRANT -D_GNU_SOURCE -DPREFIX=\"/usr\" -DEXTRA_LOCKING -DMMX -DQT_THREAD_SUPPORT -I/usr/include -I/usr/local/include -I.. -I/usr/share/qt/include -I/usr/share/qt/mkspecs/linux-g++ -o RingBuffer.o RingBuffer.cpp
20:44<mdz_>I really don't think that _GNU_SOURCE sets LFS
20:44<Chutt>since i compress down to < 2 GB / hour, and i never record more than an hour
20:44<Chutt>it does
20:44<Chutt>else you wouldn't be able to write files that large
20:45<mdz_>the doc says it does
20:45<yebyen>Chutt: hey
20:45<mdz_>I dunno what's wrong then
20:45<yebyen>Chutt: you really haven't seen that codec not found error?
20:45<Chutt>it may not be defining whatever it needs to redirect lseek to the lseek64 call
20:45<Chutt>yebyen, right
20:46<yebyen>Chutt: it's not consistent, but some of the times when I let videos play all the way to the end I get it...
20:46<Chutt>never seen it
20:46<yebyen>Chutt: even with the debs now
20:46<Chutt>it's only at the end, though?
20:46<Chutt>that might help a bit..
20:46<mdz_>that stuff is conditional on __USE_FILE_OFFSET64
20:46<yebyen>Chutt: yeah, do you hit escape before stuff ends usually?
20:46<mdz_>in unistd.h, which includes features.h before any of that
20:47<Chutt>yebyen, no
20:47-!-nevertheless [] has quit ["Client Exiting"]
20:47-!-nevertheless [] has joined #mythtv
20:47<yebyen>Chutt: hm
20:48<sefudier>ahh.. finally, mythtv at descent. just installed it on a xp 2.0 box, the hdd is still old though (only 7MB/s)..
20:49<yebyen>xp 2.0 box?
20:49<yebyen>athlon xp?
20:49<yebyen>didn't know there was a 2.0
20:49<yebyen>of course, i'm still running a Slot A 700mhz Classic Athlon
20:49<sefudier>works very good with the latest rls version at 640x480/mpeg4, using rtjpeg its not good.. whats the average MB/s required for rtjpeg?
20:49<sefudier>yebyen: yeah, there's even 2.8 now
20:50<sefudier>of course thats some major $$
20:50<sefudier>no.. 2000+ thing
20:50<yebyen>i thought like a revision 2.0 :)
20:50<sefudier>their naming stuff
20:50<mdz_>Chutt: seek: Invalid argument
20:50<Chutt>what'd you do?
20:51<mdz_>just switched back to lseek and had it print the error
20:51<Chutt>_GNU_SOURCE defines __USE_LARGEFILE64
20:51<Chutt>but it does not define __USE_FILE_OFFSET64
20:51<Chutt>meaning, it needs to use lseek64 manually
20:52<mdz_>ah, that makes sense
20:52<yebyen>Chutt: you need a slow-ass box to do testing on :)
20:53<yebyen>Chutt: I tend to think that it's the cause of most of my random little niggling problems
20:53<mdz_>I didn't think _GNU_SOURCE would change the size of off_t because that changes binary compatibility
20:53<Chutt>mdz, exactly
20:53<Chutt>yebyen, naw
20:53<Chutt>that'd be silly =)
20:53<mdz_>so we need -D_FILE_OFFSET_BITS=64
20:53<Chutt>i just use lseek64 =)
20:54<yebyen>Chutt: would it? :) could be bugs that only happen with dropped frames, which you'd almost never experience
20:54<mdz_>are you going to commit that then?
20:54<mdz_>looks like it's only that file which uses lseek
20:54<Chutt>and jim's patch to fix the delete bug that he made
20:55<yebyen>Chutt: problems like there being a file that plays back strangely (no audio and it plays ultra-fast), or a file that causes mythtv to hang for a minute or so when I go over it in the browser
20:55<Chutt>and erik's patch to add the xawtv adding
20:55<mdz_>actually that's the only change to my RingBuffer right now, I can just commit it
20:55<Chutt>i've already got it in
20:55<Chutt>well, my tree
20:55<yebyen>Chutt: what's your plan on the next mythtv release? any major features going in? :)
20:55<Chutt>and going to commit in a sec
20:56<mdz_>let me know when you have, so I can finish watching my movie :-)
20:56<Chutt>yebyen, ability to have multiple boxes
20:56<yebyen>oh, like one for encoding and one for playing?
20:57<Chutt>as long as the encoding box has a tuner card :p
20:57<yebyen>uh, yeah :_
20:57<yebyen>well what were you referring to?
20:57<Chutt>big beefy encoder box in my office
20:57<Chutt>silent wimpy box out in the living room
20:58<Chutt>mdz, it's in there (tm)
20:58<Chutt>was that empty commit message from a few hours ago you adding a directory or something?
20:58<yebyen>Chutt: i'm thinking of upgrading my shit, so I can get it to go at 640x480
20:58<yebyen>but I just ordered a wireless access point, so that's not happening right now
20:58<mdz_>Chutt: probably, I did add a directory
20:58<Chutt>yeah, that'd do it
20:58<mdz_>and I don't make empty commit messages :-)
20:59<Chutt>i should fix that script one of these days, but i'm lazy, and it's been broken like that for years
20:59<mdz_>I hate that CVS makes you add directories to do a proper diff
20:59<Chutt>oh, it was totally empty, no changed files or anything
21:00<mdz_>yay, no conflicts
21:00<yebyen>speaking of conflicts
21:00<yebyen>"You have time conflicts"
21:01<yebyen>isn't quantum leap on scifi channel anymore
21:08<Chutt>answering email
21:08<Chutt>some guy asks if i'm planning a windows version
21:10<_shad>Chutt: so when you have a box that will encode seperately, the cpu req's will go down a bit?
21:12<_shad>how much cpu time does it take to just play back the video?
21:14<Chutt>depends on the video, of course
21:16<_shad>say I have a seperate box for encoding and playing, I can have a slightly slower box because i won't be playing and encoding at the same time, right?
21:17<Chutt>but, then you need two boxes
21:17<_shad>I have 9 computers here
21:17<_shad>all kinda crappy :)
21:18<mdz_>you'll still need one with a fairly fast CPU to do encoding
21:18<mdz_>depending on the quality you want
21:18<_shad>I'll use the p3-450 for encoding
21:19<_shad>and max out the quality on that
21:19<mdz_>won't take much :-)
21:19<_shad>then I'll use a p2-350 for playback
21:19<_shad>mdz: I know :)
21:19<_shad>but it's better than what it is right now
21:20<_shad>and it also means that I can put many computers around the house to watch tv with one cable line. :)
21:20<_shad>I even have enough puters to put on in the crapper. Hehe
21:20<sefudier>how's the current cvs? usable? any numbers on the performance improvement over the last release?
21:20<Chutt>current cvs is quite useable now
21:20<Chutt>it may not be for long
21:21<Chutt>so if you want to try it out, now's the time
21:21<sefudier>great :)
21:22<mdz_>Chutt: what are you going to break?
21:22<Chutt>well, not 'soon'
21:22<Chutt>but i'm thinking i'll be starting on the splitup next week
21:24<mdz_>illegal 3. esc, esc 2 encoding possible
21:24<mdz_>Error at MB: 295
21:24<mdz_>concealing errors
21:24<Chutt>you got it back
21:24<mdz_>it seems to have successfully concealed the error this time
21:24<mdz_>(didn't crash)
21:25<mdz_>completely different position
21:25<mdz_>and it didn't report an error at the previous position
21:25<Chutt>i wonder what it is
21:25<mdz_>so I don't think it's the stream
21:26<mdz_>I wonder if we don't have this dr buffer stuff quite right yet
21:26<Chutt>it may not be =)
21:26<mdz_>I want to find at least one program which actually uses it
21:27<mdz_>I can't find anything besides mplayer though
21:27<Chutt>it was probably written for mplayer's use
21:28<mdz_>then you'd think the mplayer guys would know what to set dr_ip_buffer_count to
21:31<mdz_>so about configuring the settings in the I to understand that there really is no substantial code yet for the settings dialogs?
21:32<Chutt>there's pretty much nothing
21:32<Chutt>want to learn qt?
21:32<mdz_>doesn't look hard
21:32<mdz_>I probably already know enough Qt to do it
21:32<mdz_>but I'd rather not write something from scratch just to do the recording settings
21:33<mdz_>presumably we want some reusable stuff for various kinds of settings
21:33<Chutt>i think the recording stuff is going to be pretty one-shot stuff, anyway
21:33<Chutt>since it's fairly different from all the other settings
21:33<mdz_>I had thought about defining the valid codec parameters in the db
21:33<mdz_>and using those to build config dialogs
21:33<Chutt>most of those are just on/off type things
21:33<mdz_>but that's a pain to deal with on upgrades
21:34<Chutt>the recording stuff is all values and the like
21:34<mdz_>right, all of the recording parameters are numeric currently
21:34<mdz_>except for the names of the codecs themselves
21:34<Chutt>and, you'd have to do like 'edit profile'
21:34<Chutt>'new profile'
21:34<mdz_>but most of them have valid ranges
21:34<Chutt>'delete profile'
21:35<mdz_>isn't there a GUI design tool for Qt? what's it called?
21:35<Chutt>i've not used it
21:35<mdz_>maybe I'll lay out some stuff and see what it looks like
21:35<mdz_>I've used glade
21:35<mdz_>it's gotta be better than glade
21:35<Chutt>not a fan of design tools
21:36<Chutt>since glade is crap =)
21:36<mdz_>I don't use them for the final product
21:36<mdz_>but they're very nice for quickly putting together layouts to look at them
21:36<Chutt>it's just always said something to me that the glade people don't use glade to develop glade
21:36<mdz_>using glade to develop itself would make my head hurt
21:37<mdz_>is it a widget or a meta-widget?
21:37<Chutt>throw in a few GTK_BLAH_BLAH() macros, and you're set
21:38<mdz_>ah, it's called qt Designer
21:38<mdz_>oh yeah, forgot to cast it
21:38<mdz_>it's in qt3-tools, I already had it installed
21:39<Chutt>well, then it probably work
21:39<Chutt>if trolltech wrote it
21:39<mdz_>how do you feel about wizards? :-)
21:39<Chutt>like em
21:39<Chutt>long as they're easy to use
21:39<Chutt>and they'd probably be the best format for this stuff
21:40<mdz_>I agree
21:40<Chutt>especially if it's trying to keep everything mostly useable with a remote
21:42<Chutt>i envisioned a wizard like thing for replacing the text setup program
21:44<mdz_>oh wow, designer has widgets which can be tied to the db
21:44<Chutt>those are in qt, actually
21:45<mdz_>seems like more trouble than it's worth for this particular thing
21:45<mdz_>I wonder if a tabbed notebook would be usable with a remote
21:45<mdz_>if left/right switched tabs and only up/down moved within the tab
21:46<mdz_>pretty much equivalent to a wizard where left/right mapped to back/next
21:51<mdz_>heh, I have some radio button groups set up, but I can't find a simple damn text label
21:51<mdz_>I want to put a description next to the different codecs explaining the tradeoffs
22:05* yebyenwaits impatiently for wireless access point
22:06<mdz_>Chutt: what do you think, independent width/height controls, or predefined WxH choices?
22:12<_shad>predefined to keep aspect ratio?
22:32<Chutt>depending on the tv format -- pal/ntsc, etc =)
22:36<mdz_>right now I have full-size and half-size (which are calculated from the format PAL/NTSC)
22:36<mdz_>and spin boxes for a custom setting
22:41<yebyen>o/~ while research on successful drug policy shows that treatment should be increased... AAAAAAHHHHH... and law enforcement decreased while abolishing manditory minimum sentences... AAAAAAAHHHHHHH o/~
23:04<mdz_>mockups -
23:05<Chutt>i think a select box might work better for the codec
23:06<Chutt>like, hit the space/enter to select it, pops up a selection
23:06<Chutt>list thingie
23:06<Chutt>looks good, though
23:06<mdz_>a button saying "Select codec..." or something?
23:07<mdz_>I thought a radio button was nice because it triggered other changes on the screen
23:07<Chutt>not a button
23:07<mdz_>(the parameters at the bottom)
23:07<Chutt>but a select list box
23:07<Chutt>like, it shows one thing
23:07<mdz_>oh, gotcha
23:07<Chutt>then you select it, and it pops up a list of possible selections
23:07<mdz_>same semantics, different visual
23:07<mdz_>I don't know about the popping up, though
23:07<Chutt>easier to add more
23:08<mdz_>pop up into a box in the existing window, or pop up a window?
23:08<mdz_>I'm concerned that it might be too many settings on one page
23:08<mdz_>I thought about moving the codec-specific stuff to a separate page
23:08<mdz_>because the user has to navigate past all of the settings to go to the next page
23:09<mdz_>do you mean a list box for the codec, like there is for the sound device?
23:09<mdz_>or really something which does not display all the options until it's selected?
23:09<Chutt>is what i mean
23:09<mdz_>my idea of a combo box is what is on the audio page for selecting the sound device
23:10<mdz_>but i'm not goot with this terminology
23:10<Chutt>there's an example on there
23:10<mdz_>found it
23:10<mdz_>(in designer)
23:10<mdz_>ah, a drop down box
23:11<Chutt>there's so many different things to call standard widgets =)
23:11<mdz_>hmm, with a combo box I can't have the explanation next to it
23:11<Chutt>sure you can
23:11<mdz_>inside the selection?
23:11<Chutt>just make the stuff a lot of text
23:11<Chutt>like, the whole line there can be the box
23:12<mdz_>ok reload it
23:13<Chutt>i like it
23:13<Chutt>what do you think?
23:13<mdz_>I don't like that LCD number widget very much, but I didn't see anything better
23:13<mdz_>I think the combo box is a better fit
23:13<mdz_>but I wish I could right-align the descriptive text
23:14<Chutt>can you make it just a qlabel (the number, that is)?
23:14<mdz_>I can try that and see how it looks
23:14<mdz_>unless you like the LCD :-)
23:14<Chutt>hell no =)
23:15<mdz_>the LCD shows pretty unambiguously that it's strictly numeric
23:15<Chutt>it's ugly
23:15<mdz_>I don't like the way a plain label looks
23:15<mdz_>it doesn't look associated with the slider
23:15<mdz_>it's just some text sitting there which happens to be a number
23:16<mdz_>I could use a spin box
23:16<mdz_>like the luma/chroma thing
23:16<mdz_>but that has arrows and would be confusing next to the slider
23:16<mdz_>reload for a screenshot with the label
23:17<mdz_>I just want a non-editable number in a box
23:17<mdz_>is there one of those that I'm not seeing?
23:17<Chutt>might be able to use an editbox
23:17<mdz_>yeah was just going to try that
23:17<mdz_>I guess if I make it not focusable it might work
23:18<mdz_>ok updated the screenshot
23:19<Chutt>looks good to me
23:24-!-chris__ [] has joined #mythtv
23:27<mdz_>do you think it'll be navigable with a remote/
23:27<Chutt>should be
23:27<mdz_>it sucks with a keyboard
23:27<Chutt>will probably have to futz with qt's signals a bit
23:28<mdz_>if you nav to the combo box, up/down change the combo box
23:28<Chutt>to make focus change properly
23:28<mdz_>then you can't escape without tab
23:28<Chutt>yeah, that's something that'll have to be fixed =)
23:29<mdz_>when I preview it, the radio buttons don't act like radio buttons
23:30<Chutt>they probably need to be in a group
23:30<mdz_>I put them in a buttongroup
23:30<Chutt>dunno then
23:30<mdz_>oh, maybe not
23:30<mdz_>there are groupboxes and buttongroups and they look exactly the same
23:32<mdz_>that didn't fix it either
23:33<mdz_>oh well
23:33<Chutt>implementation issues
23:37<mdz_>you think tooltips would be usable?
23:37<mdz_>or just obnoxious?
23:37<mdz_>it depends on how it would place them
23:37<mdz_>if I could move some of the explanatory text into tooltips, it would clean up the screen
23:37<mdz_>but if it placed the tooltip over the input widget, it would be bad
23:37-!-nevertheless [] has quit [Read error: 110 (Connection timed out)]
23:37<Chutt>well, that's doable
23:37<Chutt>make a section a qlabel
23:38<Chutt>then every other widget send a signal to that when it gets focused
23:41<mdz_>and put it at the bottom or something?
23:41<Chutt>or somewhere, i dunno
23:55<mdz_>I don't really know where to put the deinterlace setting
23:56<mdz_>I guess I should leave it in a separate section
23:56<mdz_>since these will be reused for different profiles