Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

#rockbox log for 2021-04-08

00:02:31 Quit dconrad (Remote host closed the connection)
00:06:14***Saving seen data "./dancer.seen"
00:15:42fs-bluebotBuild Server message: New build round started. Revision c77cd7027f, 298 builds, 10 clients.
00:26:53fs-bluebotBuild Server message: Build round completed after 671 seconds.
00:26:56fs-bluebotBuild Server message: Revision c77cd7027f result: All green
00:48:52 Join bonfire [0] (
01:16:16 Join dconrad [0] (~dconrad@
01:20:27 Quit dconrad (Ping timeout: 252 seconds)
01:44:32 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:b9fd:c03d:3404:8cf7)
02:06:15***Saving seen data "./dancer.seen"
02:45:07 Join kakaka [0] (~koniu@gateway/tor-sasl/koniu)
03:16:54 Join dconrad [0] (~dconrad@
03:21:06 Quit dconrad (Ping timeout: 240 seconds)
03:50:40 Quit Galois (Quit: Leaving)
03:51:32 Join Galois [0] (
04:06:17***Saving seen data "./dancer.seen"
04:11:53 Join lebellium [0] (
05:00:07 Quit kadobanana (Quit: Idle for 30+ days)
05:42:56 Join atsampson [0] (
05:48:18 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
06:06:20***Saving seen data "./dancer.seen"
06:19:57 Quit atsampson (Quit: new kernel)
06:33:08 Join atsampson [0] (
07:17:30 Join dconrad [0] (~dconrad@
07:21:41 Quit dconrad (Ping timeout: 240 seconds)
08:04:03 Join Saijin_Naib [0] (
08:06:23***Saving seen data "./dancer.seen"
08:11:04 Join cockroach [0] (~blattodea@pdpc/supporter/active/cockroach)
08:17:06 Quit Saijin_Naib (Ping timeout: 246 seconds)
08:25:54 Join massiveH [0] (
08:29:40 Join dconrad [0] (~dconrad@
08:30:13 Quit dconrad (Client Quit)
08:31:10braewoodsonce I get my H10 back from bilgus, i just had an idea for improving the scroll pad
08:54:11braewoodsthe old one is a PITA to use
08:54:27braewoodsi noticed the code for the HDD6330 port uses a similar method for the scroll pad it has
08:54:51braewoodsso i thought, why can't i try to implement a similar algorithm for interpretting the scroll pad data?
08:55:01braewoodsit might work better than the current one
08:55:06braewoodswhich is a pita to use
09:26:28 Join tomato3 [0] (t0mato@gateway/vpn/mullvad/tomato)
09:26:46 Quit tomato (Ping timeout: 240 seconds)
09:26:47 Nick tomato3 is now known as tomato (t0mato@gateway/vpn/mullvad/tomato)
09:36:46 Quit tomato (Ping timeout: 240 seconds)
09:37:54 Join tomato [0] (t0mato@gateway/vpn/mullvad/tomato)
09:49:55_bilgus speachy looking at your patch for addresssanitizer do I just do configure with the switch then? −−with-address-sanitizer && −−sdl-threads?
09:50:11speachynope, just −−with-address-sanitizer
09:50:17speachyit'll enable SDL threads for you
09:50:43braewoodsisn't address sanitizer only for SDL native?
09:51:01braewoodsyou can't really use AS in embedded afaik
09:51:21speachyit's only used in sim builds, yes.
09:51:38speachy(most of our "native" targets lack SDL too)
09:51:48speachywhoops, "hosted"
09:55:16_bilguswell the sim is for sure faster but I haven't seen anything come through −− false positive or otherwise
09:55:52_bilgusdoes it do it like valgrind in the terminal or does it do it after quit?
09:56:22speachylike valgrind
09:56:38_bilgusugh the whole not shutting down the sim while charging really annoys me some days
09:56:53speachysome stuff is only reported upon exit (eg leaks)
09:57:09_bilgusuh oh sim locked up
09:59:27_bilguseh I think it might just be that button ovfl bug
09:59:42_bilgusit just doesn't segfault with address sanitizer running
10:01:20braewoodsspeachy: i see the ipod video units have an unsupported video decoder?
10:01:25braewoodsspeachy: what does that do?
10:01:29braewoodsvideo output?
10:01:57speachympeg2 (ie dvd) in hardware, IIRC
10:02:10braewoodsif that's it then not worth fretting about
10:02:26_bilgusI think it was detrmined to not be easy enough for the marginal benefit
10:02:47braewoodsi was wondering if it was required for tv out
10:02:48speachyI _think_ sufficient info is out there to actually drive the chip but it's honestly rather pointless. even when it was new
10:03:15speachy(I have similar opinions of rockbox's mpegdecoder plugin fwiw)
10:03:17braewoodsin any case i think i should start with the tv out for what i have
10:03:18_bilgusoh thats cool the os.exit(stmt) from lua prints out in the terminal
10:03:49_bilgusI guess it probably always did that
10:03:49braewoodswhen i get my h10 back i plan to do some IO tests
10:03:55braewoodssee if these fixes change the speeds any
10:04:16braewoodswould be hilarious if the H10 is suddenly faster
10:04:30braewoodsjust weird how my other PPs can do 13MB/s but the H10 isn't
10:04:34_bilgusI'm really interested to hear if we have any less problems with ipods and iflash
10:05:05_bilgusI mean what a cool series of events that your old h10 lead to a 10 year old bug
10:05:25braewoodsit was obviously above my paygrade
10:05:31_bilgusmine too lol
10:05:38braewoodsi may know C well but not the PP hardware
10:05:50braewoodsi had no chance of knowing how to fix such a thing
10:05:51_bilguslittle help from our friend speachy
10:06:26***Saving seen data "./dancer.seen"
10:06:53braewoodsat the very least it appears that these usb to dock cables can allow audio transmissions over usb
10:07:15braewoodsin theory, car audio streaming
10:07:21braewoodsand perhaps more
10:07:44_bilgushow many support more than usb drive reading though?
10:07:52braewoodsno idea
10:07:59braewoodsbut if UMS is all they support
10:08:02_bilgusthats probably who we should talk to
10:08:06braewoodsmay as well just use a real UMS thing
10:08:12_bilguscar audio head units
10:08:19_bilgusthe software sucks ass
10:08:43_bilgus$35 for a headunit is nuts even if its shit
10:09:59_bilgusthe dual rockbox din & double din hardware buttons a ok screen and a good dac only ll out for $$
10:10:25braewoodsit's kinda sad when you can easily modify an old mp3 player with more storage than a modern smartphone
10:10:27braewoodslike wtf
10:11:03_bilgusno amp onboard comes with bt, headphone port, and the latest stable of RB
10:11:43_bilgusthe car audio market hasn't dried up like personal mp3 players yet
10:12:01_bilgusand the tooling is probbly already done
10:12:22braewoodsand we have reference source code we can use
10:12:37_bilgusI mean it has buttons and a screen with head phone port and usb sounds like 'insert car audio producer here'
10:13:38_bilguswithout the amp ad wasted space you could make it 2 inches deep and sell low mid high amps to bolt onto the back
10:13:54_bilgusin the remaining case..
10:14:26braewoods_bilgus: you have any experience with analog video outputs?
10:14:42braewoodscoding wise
10:14:52_bilgusnot my forte
10:15:13braewoodsi'll see what i can find ut
10:15:29braewoodslucky me, one of the chips has test mode so i can use that for helping see if i'm getting somewhere
10:15:57braewoodsplus i can plug them into video capture device to test it on my host
10:16:42_bilguspersonally I think video is probably pointless in this age but your time as you see fit :) plus its fun
10:17:40braewoodsperhaps so
10:17:51braewoodsbut it's a way for me to learn embedded code writing
10:18:05braewoodsi prefer to focus on something that someone hasn't already done
10:18:15braewoodsfor rockbox in this case
10:18:53braewoodsbut yea i think i've got all the units i'm going to get for the foresseable furture
10:18:59_bilgusI think the barr group nigel so and so has a blog that gives you a nice overview of some of the common pitfalls when coming from x86 to embedded
10:19:08braewoodsi got enough for general USB
10:19:22braewoods2 units for tv out testing
10:19:32braewoodsone for remote control work
10:20:46_bilgusnope the sim locked up again on shutdown
10:21:00braewoods_bilgus: weird question though, is there a point to headphones vs line out?
10:21:04braewoodsthey seem very similar
10:21:16_bilguswell typically lineout is tied to 0
10:21:34_bilgusno volume control just straight line level
10:21:50braewoodsso raw audio
10:21:50_bilgusso think of it like the record players of old
10:22:20_bilgusyou had a line level that was the voltage off the crystal maybe slightly tweaked
10:22:31_bilgusthen you plugged it into an amp
10:23:35_bilgusnow something that hasn't been done in a long time is actually checking these players for no clipping at 0db
10:24:09_bilgusprobably saratoga knows the last time
10:25:25_bilgusso thats typically how you would do line out turn it all the way up and do volume through your playback device
10:28:27_bilgusthats also why having line out and plugging your headphones in hurts
10:32:19 Quit massiveH (Quit: Leaving)
10:36:17speachycar audio headunits are pretty highly integrated affairs. much like the mp3 market, the middle ground is largely gone because most cars are too highly integrated to easily replace just the radio
10:36:36speachybut on the low end they're single-chip-does-everything affairs, little-to-no RAM
10:37:02speachyon the high end they're full multi-core Android
10:37:13speachyand next to nothing left in the middle
10:37:53speachybut doing our own hardware is easier for a car thing, except the power supply and filtering has to be a _lot_ more robust...
10:38:21_bilgusSteps to repro in sim with-address-sanitizer start sim wait recharging icon and try to close the sim 1x
10:38:44_bilguswhen the bar gets full the sim locks
10:38:48speachybraewoods: headphone is just a high-impedence speaker, you're actually driving _power_ through the wire.
10:39:13speachyline out is intended to drive nearly no power, used strictly as a voltage reference.
10:39:21braewoodsi see
10:39:53speachys/power/current/ if that will make more sense.
10:40:18speachyline is intended to be consumed by an amplifier.
10:41:16_bilgusbut thats the common way its different in daps
10:41:31speachy_bilgus: one data point, the eros q/k hardware drives its line out _way_ too hot.
10:41:41_bilgusIIUC they turn it up to 0db to get rid of audio path processing
10:42:03speachybut there's no actual attenuation in the audio path so all we can do is scale the digital input down
10:42:34_bilgus^which kinda sucks
10:43:44_bilgusI ended up doing it with an audio transformer IIRC it eneded up being 2:1 @ 32k?
10:44:05_bilgusIDK its in my truck still
10:46:54_bilgusI just figured on the car audio stuff that we could get a din shell made easier and stuff your own hardware
10:47:29_bilgusplenty of room for the player stuff and a compact class d amp kit
10:47:52_bilgusenough for some real 100wx4
10:48:07_bilgusnot this 10w = 1000w peak
10:48:50_bilgusbut yeah thats true the last 10 years of cars have no provision for a radio
10:49:24_bilgusI hope they go back but its just too cheap to run wires and a touch screen compared to hardware
10:50:06speachyor just use bluetooth.
10:50:19speachy(which honestly makes a _lot_ more sense)
10:50:44speachyit's not like you're going to be playing doom when driving. :D
10:51:06_bilguswell the rocker is on its way back if you can find someone thats wants to define the low level I can finish the high
10:51:27speachypretty sure that someone defaults to me
10:51:41_bilguswhen you have time :D
10:52:25speachya native stack is pretty much a pipe dream right now. heck, if we came up with one it would be a sellable product in its own right.
10:53:07_bilgusI plan to better integrate the lua stuff during this key remap adventure its so much easier to deal with data processing tasks there
10:53:43_bilgusI think I want to better integrate it in core −− the open plugin stuff is key there
10:54:44_bilguslater I think I can get lua working in core by using a scheme from the archos
10:55:12_bilgusmake a slimmed down version that can be used as a loadable overlay
10:55:34speachy"open source car head unit" == "shoehorn android auto onto RPis"
10:56:08_bilgusall except the heat in the summer
10:56:56speachytrue, automotive electronics have to handle extended temp ranges (-40 through +125C)
11:02:20 Join Rower [0] (
11:05:25braewoodsi think there's a mpio hd300 in this listing
11:06:17speachygood price for the lot
11:07:31speachyeven if only a couple ultimately work
11:08:22speachythis one is a _lot_ more interesting IMO.
11:09:32_bilgusa lot of straight junk mixed in that second one
11:09:55speachya lot less junk in this one. decent price for the set IMO
11:11:19braewoodslooks like it has a gogear sa9200
11:12:24_bilgussame number of fuze v2 in the secons and 3rd listing think they are the same seller?
11:13:47speachyif the concensus is there I'll throw a bid up on one (or more) of these with the rb slush fund.
11:14:23braewoodsthe one with a mpio hd300 isn't a bad option, if we want one of those for testing that target
11:14:32_bilgusI think 300 is too much for whats there
11:14:36braewoodsit also have an iriver h10 of some kind in there
11:14:51_bilgusthe fuze v1s are pretty much meh
11:15:22braewoodsi see 2-4 possible rockboxable units in the ebay listing i put up
11:15:47braewoodsnot one we support
11:15:48_bilgusand the Fuze v2 is a NICCCEE player but only if handled well the front is plastic that scratches
11:15:49speachyfuzes are going for 60-80 on ebay these days. thoug there's a new-old-stock fuze+ for $40, hmm.
11:16:04_bilgusthe fuze+ is crap
11:16:16speachygood hardware, gawdaful touchpad
11:16:18_bilgusnice hardware
11:16:48braewoodsthat first one just caught my eye because of how rare the hd200 and hd300 have become
11:16:54_bilgusthose v2s are one of my fav big format players
11:17:14braewoodsif we want to continue supporting them we need hardware to test them
11:17:37_bilguslets talk to geaverts maybe he will put some out on load backed by our fund
11:17:40braewoodsbut as far as ports go they're pretty bland
11:18:12braewoodsthe hd300 is a pretty basic one
11:18:36_bilgusthe zip is worth $50
11:18:48_bilgusbut we have all of these at hand in me alone
11:19:07_bilgusnot the v1 but like I said meh
11:19:27braewoodsit's funny how toshiba gigabeats never turn up in these lots...
11:20:23_bilgusspeachy you wanna take a chance on that we should be able to recoup our investment
11:21:05_bilgusthose v2s are probably a lost cause in the front
11:21:06braewoodsisn't replacing the batteries in those a bit difficult?
11:21:10speachyI'll throw a bid at that first one that's like half junk.
11:21:39_bilgusthe v2 is simple but ya know
11:21:49_bilgusthe zip is a PITA but i'm experienced
11:21:55_bilgusthe v1 no clue
11:22:28_bilgusand the zune no clue (or if it was even ever a port)
11:22:31speachythere's a pile of sonys in there that wouldbe handy. and I can always resell the sansas assuming they work.
11:22:47speachythe fuze v1 shares the same case foibles as the v2.
11:23:21speachythat #$@#!$ fragile cable
11:23:22_bilgusor send me a box of pieces and i'll return the franken dap
11:23:58speachyI have a pile of fuze v2s (and one v1) parts. everything works except for the controls. :/
11:24:41_bilgusyou can send those my way sometime too Ive done some pretty 'interesting' repair work on old shit
11:24:42braewoodsi find it hilarious that people are selling early flash players with < 1GB of storage for $30+
11:24:54braewoodsdo they seriously think anyone would want something that crappy?
11:25:44braewoodsends soon
11:25:46speachyif someone pays them for it, sure. if not, then maybe they'll realize it's not wirth the sale.
11:26:22braewoodsmaybe to a collector but no one else
11:27:24_bilguswhatd ya bid on the mixed lot?
11:27:27speachyrecognize the philips model in that lot? other than that and the sansas, the rest is junk.
11:27:46speachyunless you're a hanna montana fan
11:28:10braewoodsspeachy: not really but i do know it's not one we have a port for
11:28:26braewoodsthe gogears we support look a lot different
11:28:33braewoodsfor one they're all black
11:28:39braewoodson the front
11:29:03_bilgusapparently lol
11:29:04speachyI bid $73. with 6 days to go. The sansas alone in that lot are easily worth 300 (assuming functional)
11:29:38speachyI think there's a FiiO X3 (2nd gen) in there too
11:30:05_bilgusand a zune
11:30:25braewoodslol zune
11:30:37braewoodsmicrosoft's attempt to be relevant
11:30:41speachyweren't the Zunes basically just the continuation of the Gigabeats?
11:30:53speachyha! instantly outbid.
11:30:54_bilgusthe zune I'm thinking of was
11:31:22speachylasted 2 minutes 21 seconds.
11:31:23braewoodsusually doesn't make sense to bid this early
11:32:05speachyeh, my attitude on such stuff is that I just bid up-front what I'm willing to pay, and if someone else wants it more, so be it.
11:37:01_bilgusI think I could become a fanboi overnight
11:39:50speachymy goal is to expand our readily-accessible testing pool.
11:43:11_bilgushahah thats already over $250
11:44:48_bilgusyep we have just over 50% coverage according t the forum
12:06:27***Saving seen data "./dancer.seen"
12:08:31braewoodsnot surprising the m:robe and mpio are so rare
12:10:26braewoodshuh m:robe has units for sale on ebay, at super inflated prices
12:10:53braewoodsah right it uses that stupid f-flex drive
12:14:11 Quit advcomp2019 (Ping timeout: 240 seconds)
12:15:34 Join advcomp2019 [0] (
12:15:34 Quit advcomp2019 (Changing host)
12:15:34 Join advcomp2019 [0] (~advcomp20@unaffiliated/advcomp2019)
13:19:40speachyhuh. initial experiment with LTO seems to work.
13:20:10speachyat least in a simulator. :)
13:22:21 Join chris_s [0] (
13:24:30chris_s g#3305 addresses the issue discussed here:,53071.0.html   – I'm not sure the old behavior was worth 'optimizing' away since Rockbox now gets stuck on broken tracks in playlists instead of simply skipping over them... ?
13:24:31fs-bluebotGerrit review #3305 at : Skip over broken tracks in playlist by Christian Soffke
13:25:02chris_sThis is the relevant commit from the end of 2017:
13:26:41chris_sI've simply re-added what had been deleted from the old code
13:32:07chris_s(by broken I mean tracks that can't be found / have been deleted)
13:33:11chris_sso it's the playlist that's broken
13:34:26_bilguschris_s could you add a check for entries versus dead entries
13:35:08_bilgusreffing that previous behavior I metioned of churning through a playlist endlessly with no valid tracks
13:36:17chris_syou mean it should only skip until it has reached the last track?
13:37:22speachyI think so, yes. the old behavior was also broken, just differently. :)
13:37:36speachy(I remember getting caught by it on multiple occasions)
13:47:15 Quit chris_s (Quit: Ping timeout (120 seconds))
14:06:29***Saving seen data "./dancer.seen"
14:13:50 Join MrZeus_ [0] (~MrZeus@2a02:c7f:a0aa:4400:1848:2748:ebf:5895)
14:38:38 Join Saijin_Naib [0] (
14:43:01 Quit Saijin_Naib (Read error: Connection reset by peer)
14:45:19 Join Saijin_Naib [0] (
14:47:09 Quit Saijin_Naib (Read error: Connection reset by peer)
15:00:07 Quit pamaury (Ping timeout: 260 seconds)
15:04:22 Join Saijin_Naib [0] (
15:06:47 Quit Saijin_Naib (Read error: Connection reset by peer)
15:20:53 Join ac_laptop [0] (~ac_laptop@
15:32:15 Join chris_s [0] (
15:32:38 Join Saijin_Naib [0] (
15:34:54 Quit ac_laptop (Ping timeout: 268 seconds)
15:36:19chris_sI've changed it so that Rockbox behaves like it currently does for the last track.
15:36:20chris_sTo be honest, though, I'm trying to understand where the (unchanged)
15:36:20chris_s2017-behavior would have been a problem?
15:36:21DBUGEnqueued KICK chris_s
15:36:21chris_sWouldn't you want't Rockbox to finish the playlist
15:36:21chris_sinstead of sitting on the last track if it doesn't exist?
15:36:22***Alert Mode level 1
15:36:22chris_sOr am I missing a scenario where
15:36:22***Alert Mode level 2
15:36:22chris_sRockbox would have entered into some kind of infinite loop?
15:36:30 Join ac_laptop [0] (~ac_laptop@
15:37:00speachyif the entire playlist is shot (eg you remove the files, but tell it to resume playback)
15:37:50 Quit Saijin_Naib (Ping timeout: 258 seconds)
15:43:44chris_syeah, I guess it could be unexpected that nothing appears to happen in that case
15:44:20 Join Saijin_Naib [0] (
15:46:23***Alert Mode OFF
15:51:02chris_sby the way, I seem to be getting buffer overflows in the Simulator using the address sanitizer when the keyboard is brought up on the iPod4g or video simulator. Not sure I'm up for debugging that though
15:52:54chris_salso when entering the recording screen using the default skin
15:54:26speachyor launching the sim without running 'make install' first
15:56:39 Quit Saijin_Naib (Ping timeout: 246 seconds)
15:57:40 Join Saijin_Naib [0] (
16:02:37 Quit Saijin_Naib (Ping timeout: 250 seconds)
16:06:32***Saving seen data "./dancer.seen"
16:17:16 Join Saijin_Naib [0] (
16:23:36 Join daswf8523 [0] (~daswf852@unaffiliated/dwf)
16:26:53 Quit daswf852 (Ping timeout: 268 seconds)
16:26:53 Nick daswf8523 is now known as daswf852 (~daswf852@unaffiliated/dwf)
16:35:47speachyok, the hifiman hm-602 showed up
16:38:01 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
16:38:23speachybattery's nearly completely dead, I'lll let it charge up before experimenting.
16:41:06 Quit chris_s (Quit: Ping timeout (120 seconds))
16:46:46 Quit akaWolf (Ping timeout: 240 seconds)
16:54:41 Join akaWolf [0] (~akaWolf@unaffiliated/akawolf)
17:02:50 Quit Saijin_Naib (Remote host closed the connection)
17:17:03 Quit ZincAlloy (Quit: Leaving.)
17:18:28 Join Saijin_Naib [0] (
17:48:59 Join PimpiN8 [0] (~PimpiN8@
17:52:50 Join ZincAlloy [0] (
18:04:59braewoodsspeachy: i usually replace batteries in units and storage if it's HDD. old expendable parts have been a source of phantom bugs for me in the past.
18:06:36***Saving seen data "./dancer.seen"
18:08:34speachyinternal flash and full-size SD on this thing.
18:08:55speachythough it's as thick as a 1st-gen ipod...
18:09:19braewoodsfunny when you put it that way
18:09:39braewoodsif we ever manage to do our own hardware, we should really socket our main storage and battery
18:09:53braewoodseither sd card or some kind of ATA connector
18:10:00speachydrives up the $$$
18:10:12braewoodstrue but those are the most used up
18:11:00braewoodsi'm just thinking it should have the most commonly used up parts be socketed for easier repairs
18:11:13braewoodsthe rest can probably be soldered
18:11:19braewoodsand given today it'd be a sd card socket
18:11:21braewoodsmost likely
18:11:51speachybattery on a connector, sure. but I have a hard time seeing how primary storage will be anything other than uSD.
18:12:02braewoodsthe HDD era has passed then
18:12:19braewoodssure, fair enough
18:13:33braewoodsone idea, just use a large enough ROM chip to install the entire rockbox file set in ROM :D
18:13:58braewoodsor just hold our bootloader
18:14:37braewoodsinteresting how there's pretty much no packard bell vibe 500s for sale like ever
18:14:43braewoodsi wonder if it was that unpopular
18:15:45speachygo with whatever's cheaper but just a bootloader will suffice. 1MBit ought to be plenty but due to market volumes larger could easily be cheaper.
18:16:07braewoodsor maybe the SoC ROM, if any, would suffice
18:16:44braewoodsjust need a simple way to bootstrap the CPU
18:18:47 Quit lebellium (Quit: Leaving)
18:18:59 Quit pamaury (Ping timeout: 248 seconds)
18:32:15 Quit Saijin_Naib (Ping timeout: 258 seconds)
18:38:26 Quit ZincAlloy (Quit: Leaving.)
19:31:28 Join coderex [0] (
19:34:10coderexAre 24 hours that I'm trying to make work a 5.5gen iPod Classic, a nightmare...
19:35:00coderexAnyone can say me how can useful help the community with bug reporting?
19:35:25coderexI would resolve my problem helping others
19:35:50_bilgusor the forums
19:36:53coderexI'm following this post,52560.0.html but it is really hard for me to understand, I'm new in this world
19:36:59 Join dconrad [0] (~dconrad@
19:40:03_bilgusfirst you should update to the newest dev version before doing any bug reporting
19:43:40coderexOk, understood
19:48:41 Join Saijin_Naib [0] (
19:51:09 Quit coderex (Quit: Connection closed)
20:06:40***Saving seen data "./dancer.seen"
20:21:16fs-bluebotBuild Server message: New build round started. Revision ad55da5f87, 298 builds, 10 clients.
20:23:36dconradspeachy: did you get that file I linked to on 3259?
20:23:48dconrader, g#3259
20:23:50fs-bluebotGerrit review #3259 at : ErosQ: Set Line Level Output volume by Dana Conrad
20:24:06speachydconrad: yep!
20:24:22dconrad(btw, am I doing the screenname mentioning thing right?)
20:24:55speachyI won't be able to do the tests until this weeknd (probably) but it seems prudent to allow the level to be "backed" down.
20:25:17dconradit makes a lot of sense to me to let it be user-defined, for sure
20:26:49dconradalso, I appreciate you A/Bing them, they're probably the same, but...
20:27:19speachy...real data is always prefererred to guesswork.
20:30:07 Join MrZeus__ [0] (~MrZeus@
20:32:51 Quit PimpiN8 (Quit: Textual IRC Client:
20:32:51fs-bluebotBuild Server message: Build round completed after 695 seconds.
20:32:55fs-bluebotBuild Server message: Revision ad55da5f87 result: All green
20:34:15 Quit MrZeus_ (Ping timeout: 260 seconds)
20:42:07 Quit Saijin_Naib (Ping timeout: 250 seconds)
20:42:46 Quit MrZeus__ (Ping timeout: 240 seconds)
20:45:24speachydconrad: as a random aside, the H2 v1.3 comes in a smaller box than the older one
20:45:50dconradhah, well less waste is good
20:47:56dconradI haven't even tried the headphones it came with, but they look kind of nice
20:53:05speachyI'm getting a panic when booting a current build
20:56:05dconradon the H2?
20:56:52dconradI think I remember getting a panic when I had just received mine when I tried to play music for the first time, but I rebooted it and it was fine? probably not the same thing
20:57:06speachyIt's trying to open "/mnt/sd_0//dev/snd/controlC0"
20:58:41 Quit ac_laptop (Ping timeout: 240 seconds)
21:00:17dconradthe "//" a problem?
21:01:02speachyno, it shouldn't be prepending /mnt/sd_0/
21:01:11dconradoh, sure
21:02:31speachyfunny thing is opens the correct file twice, closes it, then tries the bogus one.
21:03:09dconradlol it just /wants/ to fail
21:03:20dconrad"nah dog"
21:03:24speachy(running it with strace in adb)
21:43:12speachyok, latest dev build goes boom too. so it's not my end.
21:43:21speachybacking off to oldest nightly
21:43:37speachyI have a fix identified but I don't know what changed that caused it to be needed
21:44:16speachyok, 03-12 worked.
21:47:53dconradthat's pretty old, I first started playing with stuff, looks like 3/14 was my first gerrit patch I uploaded
21:48:55speachyI'm bisecting
21:49:33dconradI'm curious if you find that they break before 3-25, that would've likely been the last time I made a build locally
21:50:39dconrader, no sorry apr. 3
21:50:56speachy4-5 is still good
21:51:47speachyhmm, let me try fully reverting that line out volume patch. I thought I'd backed out the changes by hand but just in case...
21:52:08speachyok, that's it
21:52:42speachyapparently pulling in settings.h is the culprit.
21:55:04speachyayep. that pulles in rbpaths.h which is what breaks everything
21:55:17 Join ac_laptop [0] (~ac_laptop@
21:57:32dconradso it successfully opens it, closes it, does that cycle again, and then decides to open it with /mnt/sd_0/ for some reason?
21:57:52speachythe previous opens are from alsa-lib
21:57:53dconradthat's wild, it's hardcoded in as /dev/snd/controlC0
21:58:21dconradoh, not hw_open()?
21:59:08fs-bluebotBuild Server message: New build round started. Revision e17337c9aa, 298 builds, 10 clients.
21:59:39speachyopen() gets remapped to the internal rockbox app_open() which mangles paths.
21:59:59speachygotta love side-effects from #includes
22:00:12speachyanyway, this build has a fix.
22:03:17dconradboy, rockbox is way more complicated than the pic stuff I'm used to working on at work
22:04:12dconradwhere I might separate out my functions into a separate file, if I'm feeling fancy
22:06:41***Saving seen data "./dancer.seen"
22:08:27dconradits quite the whiplash
22:08:45speachyok, I'm seeing ~ +-3v on the v1.2 unit, call it 6Vpp, which lines up with your observations.
22:09:07dconradso, same
22:09:41dconradI don't know why on earth it's so crazy hot
22:10:34fs-bluebotBuild Server message: Build round completed after 685 seconds.
22:10:36fs-bluebotBuild Server message: Revision e17337c9aa result: All green
22:10:53speachysame on the v1.3
22:12:01dconradI probably mentioned this on gerrit, but I wonder if maybe they took the line out from post-headphone amp
22:12:33dconradeither just because or they had some filter circuitry they didn't want to duplicate and drive up cost
22:12:35speachythere's some sort of internal switch going on; only one output is driven at a time
22:14:07speachythe DAC lacks any built-in amp; it's purely line-level output
22:14:33dconradso it is line-level though, doesn't need amping up to line level, I was wondering about that
22:14:41speachyand is 2VRMS.
22:16:29speachywhich I suppose is close enough to 3Vpp
22:17:13dconradboy it's been a while since I've had to convert between peak and rms, but...
22:18:11dconrad5.6Vpp? yeah that lines up huh
22:18:12speachythere's no built-in analog or digital attenuation (beyond soft muting)
22:18:25dconradso it's TI's fault! haha
22:18:38braewoodsspeachy: so what unit are you working on?
22:19:00braewoodsthe hifiman?
22:19:06speachyeros q (hifiwalker h2, agptek h3, surfans/irulu f20, and eros k)
22:19:13braewoodsoh i see
22:19:43dconrador lazyness/cost-cutting on the circuit designers
22:24:37dconradis that the right datasheet? the rockbox wiki reads 5102
22:27:14speachywell, that's also a 2V RMS thingey
22:27:25speachy2.1V, more precisely
22:27:26dconradyeah true
22:27:51dconradI got excited though because it looked like the 512x had a bunch more features than the 5102 haha
22:27:58 Quit ac_laptop (Ping timeout: 240 seconds)
22:28:00speachyI have a 5122 plugged into one of my RPis
22:28:30speachyextra features or not, we have no ability to twiddle it directly thanks to the lack of kernel sources
22:28:38dconradwell, someday...
22:29:24speachythough if we (==I) ever get around to making that a native port to the Eros Q platform..
22:32:12dconradI guess I just don't get why TI would design it to output such high voltages
22:32:20speachyhmm. if we natively supplied this thing 24-bit audio we could avoid losing dynamic range when we scale down
22:32:38dconradmaybe they're expecting 50% level at most, for the hottest line spec?
22:33:09dconradaccording to wikipedia (great source, right...) pro stuff goes up to 3.5Vpp
22:33:09speachywell, true full scale audio signals are quite rare, even before you consider 24/32-bit sample data
22:34:04dconradright, but it's probably a good edge case test, don't want clipping
22:34:25speachyso what we _could_ do is ditch the soft-vol stuff in the rb core, and fake volume control in the driver, converting/scaling the data to 32bpp output
22:35:06dconradthat sounds like a significant ...paradigm?... shift
22:35:21speachynot really, that's already done for most sonys
22:35:37speachyand the m3k's original linux load
22:35:52dconradlikely what the OF is doing?
22:36:42dconradI sound like a broken record, but that popping stuff just doesn't happen in the OF
22:36:51speachydould be due to that
22:36:54dconradso maybe that's what they're doing
22:37:10braewoodsall we can do is experiment for lack of original source code
22:37:45 Join f1reflyylmao [0] (
22:37:51braewoodsthere's only some many ways to output audio so we're bound to find something that works comparably
22:38:11 Quit f1refly (Ping timeout: 248 seconds)
22:38:11 Nick f1reflyylmao is now known as f1refly (
22:38:27braewoodsthough... popping? is that something a low pass filter would help with?
22:38:36dconradwell, crackling stuff
22:38:39braewoodsi recall emulators would sometimes have that as an option to clean up audio
22:39:14dconradin quiet parts of a track
22:39:31braewoodsthe sound of breakfast cereal
22:39:34braewoodssnap, crackle, pop
22:39:55dconradgood with milk, bad with your favorite band
22:40:10braewoodsi'll admit, the whole math that goes into audio processing it beyond me
22:40:25braewoodsi'm mostly here to add other types of features to rockbox
22:41:01dconradyeah honestly I just wanted to do a few low-hanging-fruit ux fixes
22:41:21braewoodsi wanted to do USB driver development
22:41:38braewoodssomething i can grasp without needing an advanced math degree to even understand it
22:42:31braewoodsspending so much time with C has caused my counting to improve
22:42:35speachydconrad: the 32bpp code in the existing alsa driver is tied to how the sony targets expect things to work −− they actually have some analog attenuation which makes it a bit more complicated.
22:42:55braewoodsi always end up needing to confirm how unfamiliar code counts
22:43:07braewoodsjust a habit I developed :D
22:43:18speachybut that's not going to be too difficult to refactor
22:43:21braewoodsI need to verify the lower levels work as expected before I try to do anything else.
22:44:26dconradto be honest, I'm not familiar at all with this 32bpp
22:44:43braewoodsi only know RGB stuff myself :D
22:44:59speachybps, not bpp. oops.
22:45:10braewoodsbits per sample?
22:45:20braewoodsthat means the raw audio size is
22:45:31braewoods4 * channels * frequency
22:45:47braewoodsit's been awhile since I did counting for audio frames
22:46:07braewoodsoh, and of course * the duration
22:46:39braewoodshow many samples you have is the duration, after you define the rest of the audio stream's properties
22:47:07dconradso the ...PCM?... information prior to this is fewer bits?
22:47:24braewoodsmost audio is stored in at most 16 bits per sample
22:47:37braewoodsi'm guessing he's talking about upscaling it to a higher bit count
22:48:11dconradI guess that makes sense, at low volumes (needed because of the crazy hot dac), you'd get a lot of quantization noise
22:48:27dconradso scale it after you've got more resolution
22:48:41braewoodsit's not unheard of to need to convert audio to meet hardware expectations
22:48:49braewoodsthough best if you don't need to do it at all
23:03:20 Quit cockroach (Quit: leaving)
23:04:33fs-bluebotBuild Server message: New build round started. Revision 54fcb907c1, 298 builds, 10 clients.
23:05:37speachyok! this build has the initial refactoring. and should result in smaller binary sizes for non-32b targets.
23:06:49 Join chris_s [0] (
23:09:29chris_sI think there's a typo, l72 in /firmware/target/hosted/pcm-alsa.c
23:10:18speachychris_s: @%@$%@!!
23:10:29speachyshould have just used global search/replace.
23:11:26speachyok, fixed.
23:11:31speachy(and pushed)
23:11:51dconrad...fiio? does that get used for the eros builds?
23:13:10dconradimmediately beyond my understanding haha
23:15:13fs-bluebotBuild Server message: Build round completed after 640 seconds.
23:15:17fs-bluebotBuild Server message: Revision 54fcb907c1 result: All green
23:15:19fs-bluebotBuild Server message: New build round started. Revision e86d90905b, 298 builds, 10 clients.
23:26:05fs-bluebotBuild Server message: Build round completed after 648 seconds.
23:26:07fs-bluebotBuild Server message: Revision e86d90905b result: All green
23:33:21speachydconrad: nope, not yet.

Previous day | Next day