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 2019-08-04

00:01:26 Quit ZincAlloy (Quit: Leaving.)
00:08:51__builtindanielp3344: Yes. And quake, wolf3d, and duke3d :)
00:12:19 Join dys [0] (
00:25:26fs-bluebotBuild Server message: New build round started. Revision 51d47c5, 280 builds, 10 clients.
00:30:07foolshdanielp3344: Get a fuze over the fuze+ if you can the touch pad is crap on the fuze+ if you want to play doom
00:43:24speachyspeaking of the fuxe+, anyone with one have any comments on g#2087
00:43:26fs-bluebotGerrit review #2087 at : Fuze+ hotkeys and quickscreen by Moshe Piekarski
00:44:18***Saving seen data "./dancer.seen"
00:45:12fs-bluebotBuild Server message: Build round completed after 1186 seconds.
00:45:13fs-bluebotBuild Server message: Revision 51d47c5 result: All green
00:45:14fs-bluebotBuild Server message: New build round started. Revision a18803a, 280 builds, 10 clients.
00:45:55speachywow. the build screen says something I've never seen before.
01:05:45fs-bluebotBuild Server message: Build round completed after 1231 seconds.
01:05:46fs-bluebotBuild Server message: Revision a18803a result: All green
01:11:57 Quit lebellium (Quit: Leaving)
01:14:51foolshspeachy: Seems good to me, fuze+ didn't even have a record button, and those bottom buttons have never had any other use.
01:56:52fs-bluebotBuild Server message: New build round started. Revision 9d79cef, 280 builds, 10 clients.
01:57:14speachyso let's make it so.
02:18:57fs-bluebotBuild Server message: Build round completed after 1326 seconds.
02:19:02fs-bluebotBuild Server message: Revision 9d79cef result: 0 errors 2 warnings
02:44:20***Saving seen data "./dancer.seen"
03:26:20foolshjust got a working sansa c250 lets see if it runs quake!
03:27:13fs-bluebotBuild Server message: New build round started. Revision 4209c09, 280 builds, 10 clients.
03:28:51 Join massiveH [0] (
03:53:20fs-bluebotBuild Server message: Build round completed after 1566 seconds.
03:53:21fs-bluebotBuild Server message: Revision 4209c09 result: All green
04:04:40 Join mendelmunkis [0] (
04:10:34danielp3344foolsh: lol
04:15:47speachyit was gloriously green for two builds, anyway.
04:17:34foolsh__builtin: Got "Undefined instruction pc:018b5c84 sp:40000ad0 bt end" for duke3d and "Undefined instruction pc:01b33944 sp:40000ad0 bt end" for quake halts right after launching the plugin, target has 32mb of ram. I'll see what I can make of it tomorrow
04:18:09foolshsansa c250
04:38:06 Quit mendelmunkis (Ping timeout: 244 seconds)
04:38:16__builtinfoolsh: can you map it?
04:44:24***Saving seen data "./dancer.seen"
05:22:58foolsh__builtin:wolf3d craps out to, I'm looking at the and I can't find the address "01b773d4" the closest one is "0x0000000001b773d0 plugin_start"
05:25:24foolshsame thing with duke3d, let me start another sdl plugin and see if there's something weird going on with sdl on this target
05:26:45foolshnope the sgt-* plugins run fine
05:27:19foolshjust wold3d duke3d and quake all crap out right around plugin start
05:34:37 Quit TheSeven (Ping timeout: 250 seconds)
05:35:05 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
06:01:27 Join mendelmunkis [0] (
06:10:28 Quit mendelmunkis (Ping timeout: 245 seconds)
06:44:27***Saving seen data "./dancer.seen"
06:59:11 Quit massiveH (Quit: Leaving)
07:00:59 Join mendelmunkis [0] (
07:26:54 Quit mendelmunkis (Ping timeout: 246 seconds)
07:34:31 Join mendelmunkis [0] (
07:40:54 Quit mendelmunkis (Ping timeout: 268 seconds)
07:51:17 Join mendelmunkis [0] (
07:53:33 Quit mendelmunkis (Remote host closed the connection)
07:53:57 Join mendelmunkis [0] (
07:58:58 Quit mendelmunkis (Ping timeout: 248 seconds)
08:15:35 Quit speachy (Ping timeout: 260 seconds)
08:36:03 Quit TheSeven (Ping timeout: 264 seconds)
08:37:38 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
08:42:41 Quit TheSeven (Ping timeout: 250 seconds)
08:44:29***Saving seen data "./dancer.seen"
08:45:06 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
09:55:28 Join mendelmunkis [0] (
10:08:46 Quit mendelmunkis (Ping timeout: 245 seconds)
10:16:09 Join lebellium [0] (
10:37:13 Join ZincAlloy [0] (~Adium@2a02:8108:9440:dfc:ec14:291c:83fb:9775)
10:37:28 Quit ZincAlloy (Client Quit)
10:37:38 Join ZincAlloy [0] (~Adium@2a02:8108:9440:dfc:ec14:291c:83fb:9775)
10:37:59 Join mendelmunkis [0] (
10:44:30***Saving seen data "./dancer.seen"
11:17:59 Quit mendelmunkis (Ping timeout: 244 seconds)
11:21:57 Join mendelmunkis [0] (
11:29:04 Quit mendelmunkis (Ping timeout: 268 seconds)
11:40:50 Quit Jinx (Ping timeout: 248 seconds)
11:51:12 Join Jinx [0] (~Jinx@unaffiliated/jinx)
11:54:21 Quit dys (Ping timeout: 268 seconds)
12:00:34 Join dys [0] (
12:09:46 Quit dys (Ping timeout: 268 seconds)
12:30:58 Join dys [0] (
12:44:34***Saving seen data "./dancer.seen"
14:02:31 Quit pystar89 (Ping timeout: 245 seconds)
14:44:36***Saving seen data "./dancer.seen"
15:11:50 Join speachy [0] (d102414d@
16:00:46 Join pystar89 [0] (
16:44:38***No seen item changed, no save performed.
16:57:04fs-bluebotBuild Server message: New build round started. Revision f85df30, 280 builds, 11 clients.
17:11:31 Join petur [0] (
17:11:43 Quit petur (Changing host)
17:11:43 Join petur [0] (~petur@rockbox/developer/petur)
17:12:18fs-bluebotBuild Server message: Build round completed after 914 seconds.
17:12:19fs-bluebotBuild Server message: Revision f85df30 result: All green
17:20:08__builtinfoolsh: sgt-puzzles doesn't use SDL
17:22:52__builtinWhat build are you using? I'd like to see what exactly fault instruction is
17:38:21fs-bluebotBuild Server message: New build round started. Revision 668eed3, 280 builds, 11 clients.
17:38:48__builtinfoolsh: also, which version? c200v1 or v2?
17:39:26__builtinwell, I suppose it must be v1
17:39:34__builtinv2 doesn't have enough memory
17:52:26__builtinfoolsh: I think I know what's going on
17:52:41fs-bluebotBuild Server message: Build round completed after 860 seconds.
17:52:42fs-bluebotBuild Server message: Revision 668eed3 result: All green
17:52:43 Join robertd1 [0] (
17:54:04fs-bluebotBuild Server message: New build round started. Revision d1f349b, 280 builds, 11 clients.
17:59:50__builtinhmm, does ARMv4T have CP15?
18:02:39__builtinfoolsh: I guess a simple fix would be to disable setting the alignment trap in SDL's main.c
18:03:05speachyno, the ARM7TDMI only has a CP14
18:08:37fs-bluebotBuild Server message: Build round completed after 873 seconds.
18:08:38fs-bluebotBuild Server message: Revision d1f349b result: All green
18:14:13speachy... if the SDL (etc) code relies on an there being an alignment trap to fix things up, it's going to fail spectacularly on the ARM7TDMI..
18:19:01__builtinThe trap is just there to catch any unaligned accesses. It does nothing to fix them up.
18:22:58__builtinspeachy: is it present for all ARM_ARCH >= 5?
18:44:41***Saving seen data "./dancer.seen"
19:44:41speachynot sure. let me dig through the docs.
19:45:18speachyI know it's in ARMv6
19:51:45speachyOkay, it's a hard requirement for v6 onwards, and optional in v5 or older.
19:52:37speachythat said it's _usually_ present in v5 implementations, and many v4 too. (notably excluding the ARM7TDMI aka v7T)
19:53:13speachyrule of thumb is that if the pre-v6 processor has an MMU, it has CP15.
19:54:25speachyer, s/v7T/v4T/
19:54:29gevaertsWell, yes, but the first ARM where you could have rules of thumb was the ARM7TDMI :)
19:54:39*speachy groans.
19:54:39*gevaerts hides
19:55:04*speachy is reading the ARM ARM ARM.
19:56:28gevaertsWait, what's the third one?
19:57:10speachyARM-the-company ARM-the-architechure Architecture Reference Manual
19:57:17gevaertsOh, right
19:57:27gevaertsI didn't think of the company
19:57:49speachythere's actually another ARM that can be tacked onto there but I forgot its translation.
19:59:15__builtinApplication, Realtime, Microcontroller?
20:00:03gevaertsI wonder how smug the people were in the meeting where they agreed on "Architecture Reference Manual"
20:00:23__builtinthey sure know their puns
20:00:28speachyThere were definitely some punsters in that lot.
20:01:36speachyand that's just the ones I've met. :)
20:02:18__builtindo you work there?
20:03:20speachyUsed to. they bought the startup I was at, then shut it all down last december.
20:04:18__builtinah, that explains things
20:05:01speachymy group was doing bluetooth/802.15.4 radio designs.
20:06:08Bilgusspeachy do you have a rocker?
20:07:08speachyyeah, broke the case but it otherwise functions
20:07:29Bilgusyou know it needs bt integrated right??
20:07:41Bilgusjust sayin
20:07:46*speachy chuckles.
20:08:00gevaertsWell, speachy might have Opinions about the controller they use :)
20:08:47BilgusI have opinions about the whole thing but glad to have an in production device nonetheless
20:10:01speachyI was doing the realtime radio control and related code, well below the BTLE stack and even further from applications.
20:10:37*gevaerts was just going to suggest ignorance about the upper parts of the stack as a good excuse :)
20:11:41speachybut in the rocker's specific case (or really, any hiby-based unit) just bolting up their existing system server stuff to a rockbox UI would be straightforward.
20:12:27speachyit's when you want to try and extend those pitiful capabilities that things get a lot tricker. If hiby's BT implemetation is anything like what they did to the other Linux layers...
20:12:29*speachy shudders.
20:14:01*speachy grumbles again about flagrant GPL violators.
20:15:01gevaertsI'm not actually sure if prior experience with the stuff makes that much of a difference. I mean, it definitely helps, but what *really* helps is access to debugging and tracing equipment
20:15:27*gevaerts fondly remembers the early days of USB on his sansa c200 :)
20:16:59speachyno real need for low-level debuggers since it's a full Linux system.
20:17:34gevaertsAlthough even then it can help
20:18:33speachywhat would help more −− not just on the rocker but all of the hiby-based units −− is an automated way to patch the OF image.
20:20:09*gevaerts suddenly realises that most of the stuff he did in rockbox is more than ten years old...
20:21:50*__builtin is looking to get in-game music playback in quake
20:23:24Bilguswell it can be automated but not in the way you mean
20:24:00Bilgusmore like we do it first then others can do the patching on their pc
20:24:14speachycan't be worse than trying to automate Sysgen and Vivado.
20:24:15Bilguslike the cracks of yesteryear
20:25:13speachyIIRC we have a method that works well on Linux/etc but it wasn't easily portable to Windows, where most end-users use rbutil.
20:26:53speachybut it's something I need to do for myself if only so I can trust the X3 bootloader builds.
20:37:13speachypaid work comes first though.
20:41:23fs-bluebotBuild Server message: New build round started. Revision 448b97a, 280 builds, 11 clients.
20:44:45***No seen item changed, no save performed.
20:47:46Bilguswho do I ask for flyspray priveliges?
20:48:59speachynot it!
20:49:14 Quit robertd1 (Quit: Leaving.)
20:49:59BilgusI'd settle for someone with them
20:49:59gevaertsBilgus: me, I think
20:50:31Bilguswhen you get a chance Could I get closing privileges?
20:50:42__builtincould we migrate all of FS over to gerrit automatically?
20:51:21Bilgusdamn I ask for an inch __bultin asks for like 150km
20:51:34gevaertsWell. __builtin can work on that :)
20:51:51gevaertsWhat's your flyspray username? Also Bilgus?
20:52:12BilgusWilliam W (Bilgus)
20:52:42gevaertsOK, I did something, so maybe you're in the developer group now instead of the reporter group
20:52:57*gevaerts is not a flyspray expert, merely an admin :)
20:53:15BilgusI was able to close the task as fixed, Thanks!
20:53:18speachyflyspray is still my favorite bug tracker thingey.
20:53:28speachyit doesn't ry to be everything to everyone.
20:56:24BilgusI like it all except that it shows the oldest bugs first
20:56:42fs-bluebotBuild Server message: Build round completed after 920 seconds.
20:56:43fs-bluebotBuild Server message: Revision 448b97a result: All green
20:56:58gevaertsI've only really ever worked with if for rockbox, but I've never really been annoyed by it, which for this sort of software is very good
20:59:31fs-bluebotBuild Server message: New build round started. Revision 7234de5, 280 builds, 11 clients.
21:00:39__builtinfoolsh: can you test SDL with latest git?
21:00:53__builtinI disabled the offending code, hopefully
21:03:21Bilgusanyone object to me closing the bugs from prior to 3.13?
21:07:07*speachy is all for it.
21:07:18__builtinare they fixed?
21:08:27Bilgussome are others who knows
21:08:46Bilgusbut I don't think a task from 3.1 is going to help anyone figure it out
21:09:15foolsh__builtin: sure building now
21:09:29__builtinBilgus: in that case, sure
21:09:58speachyI'd keep patches open though.
21:10:39*__builtin wonders how many devs actually check flyspray for new bugs
21:13:28Bilgustheres what 4 of us?
21:13:30BilgusI do
21:13:35Bilgusspeachy does lol
21:14:03__builtin4 active devs?
21:14:06speachyoh, I don't check it for new bugs.
21:14:16fs-bluebotBuild Server message: Build round completed after 884 seconds.
21:14:17fs-bluebotBuild Server message: Revision 7234de5 result: All green
21:14:27speachytrawling around it for forgotten patches and whatnot..
21:15:56Bilgusoh ok I do, I might say ^ active devs but it all depends what you classify as active
21:19:34__builtinhmm, does anyone have a desire to put out a new release?
21:19:55speachyit would be an interesting exercise.
21:20:17__builtinaround 3.14 there were noises about bumping to 4.0 and purging all the HWCODEC code
21:21:10speachydoes anyone active still have a HWCODEC unit?
21:21:43gevaertspixelma might still have a working ondio
21:22:35__builtindo we have a way to get per-target download stats?
21:25:21BilgusI'd really like to get rid of HWCODEC just the builds first but then all the conditional cruft that abounds
21:26:05BilgusI think she said she hasn't updated it in years though
21:26:36foolsh__builtin: duke3d works, sounds are high pitched sounds like a chipmunk speaking, game play is smooth-ish. wolf3d plays but sounds are choppy, kind of weird considering how well duke3d plays. And quake runs until right before launching the menu with "error: Draw_pic bad coordinates" most likely the screen size thing. And I should spend some time re assigning the button's
21:28:31*speachy would also like to see a toolchain bump on ARM targets
21:29:44speachymore optimization/code size options, and will bring it up to what we're using for MIPS.
21:29:55speachydunno about m68k.
21:30:19__builtinI think you just volunteered youself :)
21:30:53speachyheh, I already have a GCC 4.9.4 rb arm toolchain on this system.
21:31:36__builtinfoolsh: could you record duke and wolf? I haven't seen (heard) high pitched audio from duke
21:31:37speachybefore the flash on my (second) clip+ flaked out, I was using it as a daily driver
21:31:53foolsh__builtin: sure
21:33:49__builtinfoolsh: and of course, I doubt John Carmack had 80-pixel displays in mind back in the day
21:33:57speachydo we have a proper headliner feature or targets for a new release?
21:34:40speachythen again I don't thing we we even have a high-level changelog. That in of itself would be an undertaking.
21:34:48__builtinwe do ^
21:35:20speachyneed to add the various new unstable targets
21:35:36Bilgusthats in several places
21:35:39__builtinwhen did we add Rocker support?
21:36:22speachyI committed the X3 (and IHIFI 770/800) stuff in the late 2018/early 2019 timeframe.
21:36:40*__builtin will go in and backdate some things in MajorChanges
21:37:05speachyyeah, I'll see about doing the same for the stuff I've been hacking on.
21:37:15speachybut that will have to wait until later
21:38:17Bilguslets call it March 05, 2018
21:58:57__builtinfoolsh: definitely playing at twice speed, or something like that...
21:59:59__builtinit looks like it's zeroing the samples at a certain frequency for wolf... I saw a bug like this while porting
22:00:21 Join Strife89 [0] (
22:02:09__builtinhow did you record these? line in?
22:03:04foolsh__builtin: yes with audacity, in duke3d the waveform is choppy like its playing a sample, waiting and then playing the next
22:03:19__builtinyou mean wolf3d?
22:03:28foolshyes dang it
22:03:41__builtinyeah, I suspect the bugs are releated
22:03:55__builtinduke3d is definitely playing at exactly 2x the normal speed
22:04:23foolshtiny voice to match the tiny screen FTW!
22:06:15__builtinok, wolf3d looks like it plays 1024 samples and then silence for 2048
22:07:50__builtinmy guess for duke is that it's failing to set the 22.05KHz sample rate in rockbox, so the samples generated are 22.05KHz but are being played back at 44.1
22:08:29__builtinSomething similar is probably happening to wolf. Might be exhausting the mixing buffer and just playing silence
22:12:00__builtinaha, yep
22:12:13__builtinc200 doesn't support 22.05 rate
22:12:53__builtinI'll have to force it to do 32KHz instead
22:14:25__builtinfoolsh: can you try editing RB_SAMPR in apps/plugins/sdl/include/SDL_config_rockbox.h from SAMPR_22 to SAMPR_32?
22:14:39__builtinhopefully the cpu is powerful enough to handle the bump
22:15:04foolshsure, but in my playback settings it only has 44.1 and 48 kHz
22:15:32__builtinyou could try 44.1 if 32 doesn't work
22:24:16foolsh__builtin: duke3d seem to play fine at 32, game play speed is not affected or so it seems, wolf3d however still less choppy sounds and game play speed is effected
22:24:53__builtinso wolf3d is still choppy?
22:25:45foolshyeah, but less so at 32
22:25:58*user890104 read something about a new release...
22:26:09foolshis there a way to cheaply upconvert 22 to 44 I wonder?
22:26:37__builtinfoolsh: could you try SAMPR_44 and see if that fixes wolf?
22:27:17__builtinuser890104: you're welcome to volunteer as RM :)
22:28:20user890104__builtin: do you still keep your notes/checklist for 3.14?
22:28:36__builtinthere's which I found useful for 3.14
22:30:31user890104i'd like to volunteer for something else first. i wonder if someone can give me moderator-like access to gerrit, so i can mark as abandoned all patches that are already merged as part of other patch or as a commit different from gerrit's
22:31:00user890104i'd have to check them by hand, i guess
22:34:12foolsh__builtin: Alright at 44.1 duke3d still plays very well and I don't hear a difference from 32, wolf seems to stutter just like it did at 32
22:35:03foolshwhich is weird considering how duke3d seems that much more visually intensive
22:35:30__builtinlow framerate, you mean?
22:35:52foolshsound as well as fps
22:36:33foolshpoor little player just cant handle the awesomeness that if wolf3d
22:37:39__builtinI think I know a quick way to fix quake
22:38:00__builtinuncomment the #define BASEWIDTH / BASEHEIGHT 320/240 in sdl/progs/quake/vid_sdl.c
22:38:07__builtinand comment the LCD_WIDTH/LCD_HEIGHT pair
22:38:24__builtinthat'll make SDL simulate a 320x240 buffer that it'll scale down
22:39:13__builtinthat way we'll know if it's just wolf or if quake is affected as well
22:44:47***Saving seen data "./dancer.seen"
22:47:01fs-bluebotBuild Server message: New build round started. Revision 4ad7665, 280 builds, 11 clients.
22:50:08foolsh__builtin: yeah quake runs, poorly. and sampling was still at 44.1khz
22:52:27__builtinyay! :)
22:53:09foolshwhen BASEWIDTH / BASEHEIGHT was still LCD_WIDTH/LCD_HEIGHT it was drawing only a portion of the background image (upper left) before launching the demo, seems like there is some thing drawing to off screen memory causing the "bad coordinates" error message perhaps
22:54:02speachythere are SAMPR_CAP_?? definitions that one can use for sane defaults on various targets..
23:04:13__builtinfoolsh: how poor is the performance with scaling and 44.1?
23:05:05foolsh__builtin: piss poor, even when at 32khz
23:05:34__builtincan you screen record it?
23:06:08__builtinif it's good enough to play a level I might just force scaling for screens smaller than 240x320
23:06:24 Quit lebellium (Quit: Leaving)
23:07:51fs-bluebotBuild Server message: Build round completed after 1250 seconds.
23:07:52fs-bluebotBuild Server message: Revision 4ad7665 result: All green
23:07:53fs-bluebotBuild Server message: New build round started. Revision 08fd69c, 280 builds, 11 clients.
23:08:54speachyokay... that's strange. my last commit saw a 150-300ish byte binary size across the board.. except the Samsung YP-R0, which shrank by 30KB.
23:10:17Bilgussometimes they bounce next build it'll be right perhaps?
23:10:48Bilgusoh 30k hmm that is a lot
23:13:09__builtinthat doesn't make any sense
23:14:01speachyI don't have the toolchain for that target installed on any of my boxes.
23:16:15speachywe'll see if it reverts to the mean after the next build.
23:18:18Bilgushuh I thought I had arm-rockbox-linux-gnueabi-gcc in my toolchain but apparently not
23:18:48__builtinI've got it prebuilt in my docker image, I think
23:26:51speachyso two of the patches I have in my queue bump the sample rate for the mikmod and midiplay plugin to 44.1KHz from a hardcoded 22.05K
23:27:52speachy(or 44K if 22K isn't supported)
23:28:02fs-bluebotBuild Server message: Build round completed after 1210 seconds.
23:28:03fs-bluebotBuild Server message: Revision 08fd69c result: All green
23:28:04fs-bluebotBuild Server message: New build round started. Revision 4d5dfab, 280 builds, 9 clients.
23:30:36speachygiven that music playback is the point of those plugins, I think the bump is warranted. Is it a fair assumption to make that any target capable of handling realtime mp3 playback (ie all of them) shouldn't have a problem doing 44KHz mod playback?
23:36:23 Quit beencubed (Quit: Leaving)
23:37:24BilgusI wouldn't make that assumption
23:38:36Bilgusbut.. one way to find out?
23:39:04speachyeverything I have is relatively beefy.
23:39:50Bilgusfigure mp3 is highly optimized I'd say if they can do realtime in a heavier codec you'd be ok
23:40:19speachyand none of the codec benchmarks published include mod or midi (==mod)
23:42:19Bilgusmidi really isn't terribly complicated so idk maybe make sure there is a way to revert using conditionals and/or just push it and wait for a bug
23:44:47speachythe codec comparison page shows every target handling everything except for HE-AAC and some APEs.
23:45:32 Join advcomp2019__ [0] (~advcomp20@unaffiliated/advcomp2019)
23:46:24fs-bluebotBuild Server message: Build round completed after 1101 seconds.
23:46:25fs-bluebotBuild Server message: Revision 4d5dfab result: All green
23:47:57 Join beencubed [0] (~beencubed@
23:48:24speachyhmm. wonder if the codec_bench plugin will handle mod and midi.
23:48:33speachyguess I'll find out.
23:48:42 Quit advcomp2019_ (Ping timeout: 244 seconds)
23:52:14__builtinmidi isn't implemented through a codec, I believe
23:52:53__builtinand we now have at least 2 separate MIDI implementations in the plugin tree
23:53:59speachywe do?
23:54:12speachyoh for the games you mean?
23:56:46__builtinThere's midiplay and a port of timidity under SDL

Previous day | Next day