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 2020-07-15

00:07:07Strife89Nope, attempting to build gcc 4.5.2 quits for the same reason
00:07:47speachyc'est la vie
00:17:00***Saving seen data "./dancer.seen"
00:33:51 Join mendelmunkis [0] (
00:57:54 Quit reductum (Quit: WeeChat 2.8)
01:22:30 Join __Bilgus_ [0] (41ba23be@
01:22:46 Join ZincAlloy [0] (
01:26:06__Bilgus_RasPI; I had pretty good luck booting from the sdcard with a RO FS w/ Berryboot then the OS runs off a large format SD Card its been fine for going on 2 or 3 years now
01:26:59 Quit ZincAlloy (Ping timeout: 240 seconds)
01:29:19 Join ac_laptop [0] (~ac_laptop@
01:31:53__Bilgus_ g#2534 would be better off in a plugin especially with voices in plugins
01:31:54fs-bluebotGerrit review #2534 at : FS #11541 - Add Voice Announcement of Summary Info to WPS hotkey options by Igor B. Poretsky
01:37:25 Quit ubervison (Ping timeout: 272 seconds)
01:38:17 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:491e:90a8:a8d3:55d5)
01:43:07 Quit ZincAlloy (Ping timeout: 272 seconds)
01:52:45 Join ZincAlloy [0] (
01:53:22efqwDoes rb have any meaningful support for touchscreen at all? Most X1000[E] targets right now are touchscreen devices without meaningful hardware buttons (other than volume and power I guess).
01:57:22 Quit ZincAlloy (Ping timeout: 256 seconds)
02:15:29__Bilgus_efqw there are settings to allow button emulation by zone for some players
02:16:57__Bilgus_like touch on right side = right arrow key press
02:17:04***Saving seen data "./dancer.seen"
02:17:04efqwI've read about it on the wiki but the TouchscreenInterface page was last updated 10 years ago :P
03:21:49 Join sakax [0] (~r0b0t@unaffiliated/r0b0t)
04:17:07***No seen item changed, no save performed.
04:51:44 Quit ac_laptop (Ping timeout: 258 seconds)
04:52:50 Quit gevaerts (Ping timeout: 272 seconds)
05:02:46 Quit mendelmunkis (Ping timeout: 265 seconds)
05:04:17 Join mendelmunkis [0] (
06:17:11***Saving seen data "./dancer.seen"
06:36:23 Join miner49er [0] (5cec5d32@
06:36:54miner49erHi there people! Can I ask a question regarding getting an SDL build of Rockbox on the Pocket Go?
06:38:59miner49erWell, I'll try anyway: My build is having problem loading codecs.
06:39:10miner49erIt's failing with SDL error "Dynamic loading not supported"
06:39:31miner49erWould my theory that the building of SDL on my device simply doesn't have this function implemented?
07:00:52 Join vmx [0] (
07:05:34 Quit __Bilgus_ (Remote host closed the connection)
08:02:06 Quit miner49er (Remote host closed the connection)
08:09:09 Join MrZeus [0] (~MrZeus@2a02:c7f:70d0:6a00:8982:4653:148b:9d8e)
08:10:54 Join ac_laptop [0] (~ac_laptop@
08:17:14***Saving seen data "./dancer.seen"
08:54:14 Join massiveH [0] (
09:03:17 Quit MrZeus (Ping timeout: 272 seconds)
09:16:17 Join gevaerts [0] (~fg@rockbox/developer/gevaerts)
09:29:05 Join petur [0] (~petur@
09:29:05 Quit petur (Changing host)
09:29:06 Join petur [0] (~petur@rockbox/developer/petur)
09:29:23fs-bluebotBuild Server message: New build round started. Revision 8577d5a, 295 builds, 12 clients.
09:34:07 Join jdarnley [0] (
09:34:33 Join __Bilgus_ [0] (41ba23be@
09:35:47 Quit J_Darnley (Ping timeout: 258 seconds)
09:37:06 Join miner49er [0] (5cec5d32@
09:43:08speachy__Bilgus_, I know next to nothing of how plugins get invoked; is it realistic/feasible to have a plugin launched from within the wps that just speaks without pausing the wps while the voice is still going?
09:44:14speachyefqw, there's no "real" touchscreen support, as in treating it as a mouse-like absolute pointer
09:44:40__Bilgus_I believe so yes with the caveat that it be run as TSR as to not have to handle dealing with the WPS
09:45:50speachyefqw plus the entire UI stack has no provisions for non-button-based navigation. so no point-n-click, so to speak.
09:46:34__Bilgus_give me a few days and I'll try to work out a template or at least some POC
09:49:12__Bilgus_I have had lua running TSR as an experiment at one point so I'm at least passingly (more by the day) familar with the TSR stuff
09:49:39speachyoh, __Bilgus_, I dusted off g#1932 for benchmarks on the mini2g
09:49:41fs-bluebotGerrit review #1932 at : opus: shrink stack usage by nearly 700 bytes by Solomon Peachy
09:52:06__Bilgus_that was just moving the structures around right? something was out of iRam on a particular target was that the mini?
09:52:15speachystill have no way of running it on the worst-case coldfires. a while back you fixed the giant stack-sucking seeking-related overflow/crash.
09:53:14mendelmunkis__Bilgus_: what's TSR?
09:53:32__Bilgus_Personally I'd just do the patch and make sure its easily revert-able in case complaints take a year to surface
09:53:45__Bilgus_terminate and stay resident old DOS acro.
09:54:07mendelmunkisah yes. I am familiar with it it just didn't exactly spring to mind.
09:54:19speachymy memory is a little fuzzy now but several of the MIPS targets didn't have enough space for the expanded stack needed to play opus reliably.
09:54:44speachythey'd overflow sometimes doing disk accesses, etc.
09:55:09speachy(even without seeking)
09:55:16mendelmunkis__Bilgus_: if you get TSR working I think a few other plugins can benefit from the treatment.
09:55:24__Bilgus_oh it already works
09:55:41__Bilgus_have a look at the alpine plugin
09:58:07speachyminer49er, yes, I believe you are correct.
09:58:37speachyI find that odd though.
09:59:41speachyanyone with a m68k target around that can do a codec benchmark?
10:12:14speachy__Bilgus_, do you see any inherent reason to not merge that pull-stuff-off-the-stack opus patch? wodz's comment about possible impact on m68k not withstanding.
10:17:18***Saving seen data "./dancer.seen"
10:19:55miner49erspeachy, thanks for that. The only alternative I see is that the codec was built incorrectly - or does SDL_LoadObject just load the file regardless?
10:20:40speachyit's basically a wrapper around dlopen(). if the file was busted then I imagine you'd see an error to that effect.
10:22:19miner49erHmmm, I think I need to look at the source-code for that function as "Dynamic loading not supported" would suggest it's not been implemented. Thanks, that has made me think a bit more on this.
10:22:24 Quit massiveH (Quit: Leaving)
10:22:54speachylibsdl needs to be built with −−enable-dlopen (not the default)
10:23:59speachyminer49er, you're using the generic hosted arm cross-compiler to do the build?
10:25:01miner49erI'm using a cross-compiler specifically for the Pocket Go. I've hacked a MakeFile (for now, I'll fix the configure file properly eventually) to point towards my cross-compiler
10:25:54speachywell, you could always recompile libsdl too
10:26:31miner49erYes, that's on my list of things to try...maybe even statically link it as well?
10:27:09miner49er...or make it start with a script that points at my compiled libsdl.
10:29:00speachypocketgo v1 or v2?
10:30:34speachyah, it's mips-based.
10:32:43miner49erpocketgo V1
10:35:08miner49erAs far as I can tell, it's AllWinner F1C100S ARM9 based
10:40:05speachyI take that back; the v1 is an allwinner ARM SoC. but v2 is an Ingenic jz4770
10:41:27miner49erAh good...that confused me for a moment!
10:41:42speachysurprising they'd swap out the guts like that.
10:42:34speachythen again it's all linux (and opensource software on top too) so there are no software dependencies
10:42:50miner49erI think I shall try just implementing my own SDL_LoadObject that calls try it out. I don't need to rebuild the entire SDL as it's wokring mostly - the Rockbox GUI is displaying, sound is working (if I enable menu-clicking sound)
10:43:59miner49erpocketgo V1 is a very cheap device right now - potentially a pretty good replacement for those who's Rockboxable h/w is failing them.
10:44:27miner49er...but will probably have terrible sound or battery life! I just want to see (and hear!) it running
10:46:26speachyalso on the large side
10:48:17speachywhat you can do instead is build lc-unix.c instead of sdl/load_code-sdl.c
10:48:21speachyin SOURCES
10:48:28speachythat's a native dlopen() implementation
10:48:32miner49erIt's actually pretty dinky to be honest
10:49:20miner49erin the Rockbox source-code?
10:49:47miner49erI don't have it here at the moment...on my work laptop.
10:49:57miner49ercool, thanks.
10:50:03speachyit's relatively tiny but it's quite large for a DAP
10:50:45miner49eryes, same height as my sansa clip but 4 times the width it doesn't have a clip, which is quite useful when cycling.
10:51:38speachygranted the clip is tiny
10:52:30miner49erI love my clip...I'm on my 5th one, so when this one dies, I'll be lost.
10:56:41speachyheh. I have my very own tower of rockbox here. with a cat on top.
10:57:19speachyminer49er, if you haven't already started using the multiboot loader and runing entirely off the SD card, you should do it
10:57:20miner49erMy original archos recorder is still here...but the headphone socket is janky
10:58:14miner49erwhat advantage does that give me? I don't care about the original fact it's annoying when I accidentally cause it to start!
10:58:29speachyit means you won't wear out the onboard flash
10:58:39speachy(that's what killed my original clip+)
11:00:10miner49erAh, the internal storage is being written to...where .rockbox resides?
11:00:27miner49erah, thanks - this is good information!
11:09:12fs-bluebotBuild Server message: New build round started. Revision 82943ea, 295 builds, 14 clients.
11:11:28speachyhmm, never noticed but the X3 buttons in "USB multimedia mode" doesn't don't do anything.
11:13:49Strife89As a bonus, after setting up multiboot, if you turn on "USB Hide Internal Drive" in Rockbox, you can pretty much pretend that the internal storage doesn't exist from that point forward.
11:20:34 Quit ac_laptop (Ping timeout: 240 seconds)
11:20:59fs-bluebotBuild Server message: Build round completed after 708 seconds.
11:21:01fs-bluebotBuild Server message: Revision 82943ea result: All green
11:23:40Strife89I wonder how long the fastest build round was
11:24:27Strife89Maybe not of all time, but in the project's modern history (since, say, release 3.0)
11:27:59speachysince I've been involved I've considerably expanded the number of builds going on
11:31:05Strife89Yeah, you've breathed a lot of life back into Rockbox
11:35:56speachywe can't fix what we don't know is broken
11:45:56fs-bluebotBuild Server message: New build round started. Revision 058ba97, 295 builds, 14 clients.
11:47:22speachyokay, let's see if the irq stack bump helps with the only one still reporting X3 usb stability issues
11:47:58speachyand HID mappings are now there. Excellent.
11:53:28speachylooks like I'm going to have to completely ban the bing crawler
11:54:37speachyalready had to do that for part of the forum, but for some reason it's now up to over 1000 complete crawls of the bug tracker in the past two weeks
11:55:00Strife89That's ... excessive
11:55:40speachyit's responsible for 9% of the main www site traffic all by itself.
11:58:08fs-bluebotBuild Server message: Build round completed after 732 seconds.
11:58:10fs-bluebotBuild Server message: Revision 058ba97 result: All green
12:02:45__Bilgus_ g#2542
12:02:47fs-bluebotGerrit review #2542 at : Voice TSR Plugin Demo by William Wilgus
12:03:05__Bilgus_I'll flesh it out more later but theres the idea
12:04:38__Bilgus_speachy could you clue me in a bit on what all 2534 accomplishes?
12:04:55speachysomeone on tor is trying to mirror the themesite. (wtf..)
12:05:24__Bilgus_I guess it could have a cfg on first run that allows you to set the hotkey and which data you want announced
12:05:49__Bilgus_I've mirrored the themsite before when it was going down alot
12:06:25speachythe main thing is that voices status as one would find on the WPS −− eg date/time, battery life, current track or playlist info
12:06:30__Bilgus_a lot.. maybe they have the same concern or its of nefarious intent though i'm not sure of the eventual payoff..
12:06:31speachyso the plugin would need to be able to access that
12:07:55__Bilgus_easy enough i'll let you know when i've fleshed it out a bit
12:09:46speachyI think everything but playlist_get_display_index() is already in the plugin api
12:10:13speachy(ie which entry of the playlist we're currently on)
12:11:55speachyit currnetly reads what to voice via a config key
12:12:14speachyand would also need to be launched via a wps hoetkey
12:12:54__Bilgus_we can choose any hotkey after the initial launch or the same one
12:13:15speachysame one. the idea is to be transparent
12:14:39 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:6933:271b:1ef7:a8e8)
12:14:54speachyso WPS_HOTKEY action would check to see if the thread is running or not, and if it already is, tell it to run
12:15:24speachyit would always be the same action, so the thread would just block until woken back up.
12:15:31__Bilgus_thats taken careof automagically
12:15:37 Quit petur (Quit: Connection reset by beer)
12:16:25__Bilgus_if the app gets rerun it hands control back to the already 'running thread
12:16:41__Bilgus_if another app is run it exits
12:17:19***Saving seen data "./dancer.seen"
12:17:51speachynow for the actual voicing, would it cause the wps screen to block until the voicing is done or is that all backgrounded/mixed/etc?
12:18:21__Bilgus_i think it doesn't block but i'd have to test it to be sure
12:25:51 Join miner49er38 [0] (5cec5d32@
12:26:13 Quit miner49er38 (Remote host closed the connection)
12:27:26miner49erspeachy, so I tried bypassing the SDL_LoadObject and basically just copied in the function you talked about - I get the same error (it comes from dlerror())
12:28:13miner49erSo...can linux itself not be allowing the loading of the library - or is much more likely I'm actually not building the codecs with the correct compiler...can't understand that though, as the main executable runs fine.
12:28:21speachywas about to say that.
12:28:34speachyrun 'ldd' on the rockbox binary and one of the .codec files
12:28:42speachyor 'file' rather
12:29:17miner49erokay, on the actual device'll take a while with the stupid on-screen keyboard but I shall try!
12:29:34speachyno, you can do it on your pc
12:29:44speachy$ file rockbox.elf
12:29:46speachyrockbox.elf: ELF 32-bit LSB executable, MIPS, MIPS32 version 1 (SYSV), statically linked, with debug_info, not stripped
12:29:53speachy$ file lib/rbcodec/codecs/speex.elf
12:29:54speachylib/rbcodec/codecs/speex.elf: ELF 32-bit LSB executable, MIPS, MIPS32 version 1 (SYSV), statically linked, with debug_info, not stripped
12:30:09speachy(from the x3ii build)
12:30:18miner49eroh, okay...right I'll be back...gonna log out of laptop
12:30:23 Part miner49er
12:31:25 Join miner49er [0] (5cec5d32@
12:33:21miner49erfor rockbox I get: rockbox: ELF 32-bit LSB executable, ARM, EABI5 version 1 (SYSV), statically linked, stripped
12:33:35miner49ercodec: sid.codec: ELF 32-bit LSB pie executable, ARM, EABI5 version 1 (SYSV), dynamically linked, stripped
12:35:04speachywould be interesting to run 'ldd' on that codec executable to see if it's "missing" anything.
12:35:28miner49ermanaged to fiddle with on-screen k/b on device - ldd returns not a dynamic executable!
12:36:15miner49eroh, says same on PC too!
12:36:51speachyeven for the main rockbox binary?
12:36:58speachy(on the PC I'd expect that)
12:37:43speachywhoops, nevermind
12:37:58speachyokay, the problem is that the codec is being linked dynamically
12:38:03 Join J_Darnley [0] (
12:39:10miner49erboth on PC and device - I get not dynamic executable - but rockbox is _just_ an executable so isn't that correct
12:39:31miner49erBut codecs and rocks are meant to be aren't they?
12:39:41speachylook at the top-level Makefile
12:39:47 Quit jdarnley (Ping timeout: 240 seconds)
12:40:23miner49erI have: export SHARED_LDFLAG=-shared
12:40:41speachyand SHARED_CFLAGS
12:40:52miner49erI have: export SHARED_CFLAGS=-fPIC -fvisibility=hidden
12:41:35miner49erI based this makefile off the one generated for the linux SDL build btw
12:43:10speachyon the target, 'ldd' on the plugin/codec says it's not dymanimc?
12:44:01miner49ershould I be using -dynamiclib
12:44:12miner49erfor SHARED_LDFLAG
12:44:31miner49erthat saved me a build then!
12:44:58speachyI'd be willing to bet that some library that the codec relies on is not available for static linking.
12:45:21miner49erso should all the codecs be getting linked with the main executable?
12:45:50speachynope, everything is a loadable module
12:46:09miner49erthat's not how the linux SDL build works - it still dynamically loads the codecs, doesn't it?
12:48:28 Quit michaelni (Ping timeout: 256 seconds)
12:49:52miner49erSo is the result of ldd "not a dynamic executable" (when run on device) correct or incorrect?
12:50:19speachystatic-linked vs dynamic-linked refer to an individual binary. one can dynamically load a statically-linked library.
12:50:57miner49erOkay, that sounds strange but I will take it as given!
12:51:43speachythe "not a dynamic executable" is odd. it very well might not be linked properly.
12:52:21miner49erIt's highly probably that my MakeFile hackery is the cause!
12:52:22speachydynamic-link vs static-link basally determine if the given binary is self-contained
12:52:34speachyie doesn't require anything else in order to load properlyt.
12:52:45speachy(eg the basic C library, SDL, or whatever)
12:53:01speachythe plugins are supposed to completely self-contained
12:54:00miner49erafk for a moment
12:58:31 Join michaelni [0] (
13:00:59miner49erI had to remove -lpulse-simple -lpulse from LDOPTS in order to get it building - could that be the cause?
13:03:10 Join PimpiN8 [0] (~PimpiN8@
13:07:41Strife89lol, I just noticed on the build table that the estimated RAM usage for one device appears to hit 1.5 GB
13:07:47speachyit's probably picking that up from the host pkg-config. if the main executable works then I'd expect the plugins to be okay as well.
13:08:55speachyStrife89, yeah, that happens because of non-linear memory maps
13:08:59PimpiN8what are the updates with the new release?
13:09:52Strife89AFAIK there aren't even soft plans for a 3.16 release
13:10:17speachy shows what's changed since the last release
13:10:17 Join ac_laptop [0] (~ac_laptop@
13:10:26speachythe last couple of months aren't there yet.
13:10:40*speachy pokes everyone
13:10:43*speachy pokes himself
13:13:37 Quit PimpiN8 (Ping timeout: 246 seconds)
13:14:05 Join petur [0] (~petur@rockbox/developer/petur)
13:28:18 Join lebellium [0] (
13:37:05 Join livvy [0] (~livvy@gateway/tor-sasl/livvy)
13:44:24speachywiki updated with most of what's happened since may.
13:44:48speachyit seems simultaneously too verbose and not enough.
13:45:26speachysome of the "improvements" could probably be changed to "new features"
13:47:27speachymendelmunkis, any further movement on g#2428? or g#2469?
13:47:38fs-bluebotGerrit review #2428 at : Add support for ID3 tags embedded in AIFF files by Moshe Piekarski
13:47:38fs-bluebotGerrit review #2469 at : imx233: rtc: Explicitly clear the soft reset bit when initializing by Solomon Peachy
13:47:59Strife89I think you listed a good amount, it's nice to know about so many under-the-hood changes
13:59:56miner49erspeachy, do you have any more insights regarding my codec issue before I give up for the evening?
14:00:14speachynot at the moment. my primary attention for now is $dayjob alas
14:00:40miner49erah, like me earlier!
14:01:28miner49erOkay, well I try and process what you've shared on a low priority background thread for now then. Thanks for you help :-)
14:01:43 Quit livvy (Ping timeout: 240 seconds)
14:05:26 Quit vmx (Quit: Leaving)
14:09:21speachyhey, what was that 1.5GB RAM target?
14:11:06Strife89One of them was ihifi760
14:12:25Strife89creativezenxfistyle: 1,611,905,888
14:12:53speachyfixed the imx233 (eg fuze+) and the hosted binaries
14:13:55speachylet's see if that fix also affected the 760
14:14:17Strife89All of the ihifis have similar numbers
14:15:14Strife89And most of the Creative Zens
14:17:00__builtinspeachy: re MajorChanges, looks good to me
14:17:19__builtinthough on the verbose side for sure
14:17:23***Saving seen data "./dancer.seen"
14:18:09__builtinregarding a release I think there should be some discussion
14:18:51__builtinabout both an upcoming release and how we want to approach them going forward
14:20:24fs-bluebotBuild Server message: New build round started. Revision 650eaa3, 295 builds, 14 clients.
14:20:40speachythis build will show drastic reductions in ram size for several targets.
14:21:50speachyI'd like to see fewer manual steps involved in a release. hopefully the infra changes over the past few months will have helped with that
14:22:34speachybut .. for 3.16 IMO we need to get the rocker/x3ii/x20 into rbutil so they can become "supported"
14:23:09 Quit TheEaterOfSouls (Remote host closed the connection)
14:24:55__builtinthere was talk before doing 3.14 of eventually dropping HWCODEC support for the 4.x branch
14:25:04mendelmunkisspeachy: g#2428 is ready to go I think.
14:25:05fs-bluebotGerrit review #2428 at : Add support for ID3 tags embedded in AIFF files by Moshe Piekarski
14:25:35speachywe can drop hwcodec at any time, really.
14:25:57__builtindo we actually want to, though?
14:25:59mendelmunkis2469 I think is a good idea but I plan to hold off putting it in my prsonal build untill i see the bug again (which may be a few years.)
14:26:21mendelmunkis(untill than==en I can't prove it fixes the relevent bug.)
14:26:27speachyhwcodec isn't actually interfering with anything right now. not even the toolchain bump I'm still working on
14:26:57__builtinexactly, all our new features are #ifdef'd out on HWCODEC
14:26:57 Join TheEaterOfSouls [0] (~souls@unaffiliated/theeaterofsouls)
14:27:50__builtinso they aren't interfering with anything, correct
14:27:57speachythe infra side of things already can handle "retired" targets properly
14:28:16__builtinbut by the same token they haven't gotten any changes for years
14:28:37speachyother than global bugfixes, no.
14:28:38__builtinand we don't even have capability to test on them anymore
14:29:09speachythe recorderv1 hasn't even been compileable since 3.13
14:29:22speachythough with the toolchain bump and a couple more features removed I was able to make it fit
14:29:36 Quit shrizza (Ping timeout: 244 seconds)
14:29:56 Join shrizza [0] (
14:30:53speachyit's probably safe to say that we lack the ability to test 2/3rds of the targets these days.
14:31:39fs-bluebotBuild Server message: Build round completed after 675 seconds.
14:31:41fs-bluebotBuild Server message: Revision 650eaa3 result: All green
14:31:42fs-bluebotBuild Server message: New build round started. Revision ff8cca7, 295 builds, 14 clients.
14:33:22__builtinyes, but we can be reasonably confident because the targets we can't test share most of the mid- to high-level code with other targets
14:33:50speachyokay, binary sizes should be gnerally saner. ihifi shrank by 1.5GB! such progress!
14:34:15__builtinso as long as low-level, target-specific changes are tested by the author, things are usually fine
14:37:06Strife89Oh neat, you *can*add a build client in the middle of a round
14:38:04Strife89i5-M520, 8GB RAM, SanDisk SSD
14:39:12Strife89Debian 10 under WSL
14:42:04speachy speaking of builders −− of the active set, I think only one can do the android and ibasso builds.
14:42:24fs-bluebotBuild Server message: Build round completed after 643 seconds.
14:42:25fs-bluebotBuild Server message: Revision ff8cca7 result: All green
14:42:26fs-bluebotBuild Server message: New build round started. Revision e884140, 295 builds, 15 clients.
14:43:45Strife89I'll try to add those to one of my builders soon
14:44:16Strife89My fastest builder is stuck at home while my slowest builder (the one I just added) is the only one I'm willing to leave at work)
14:47:14TheEaterOfSoulsStrife89: WSL2?
14:47:28TheEaterOfSoulsI'd expect version 1 to be horrendously slow
14:47:37Strife89WSL1. I haven't used this laptop in so long that it's still on 1809
14:47:59Strife89Background upgrading to 1909 right now
14:48:37Strife89I suppose I should just wipe it and load a Linux distro
14:49:23TheEaterOfSoulsWSL2 is nice for what it is, but I'm considering switching to headless VMs in VirtualBox (as I did before WSL) because they behave more like actual Linux systems (mainline kernel and systemd) and I can interact with hardware
14:49:59TheEaterOfSoulsNot 2004?
14:50:55TheEaterOfSoulsI use Windows on the desktop exclusively because of accessibility
14:51:08Strife89I want to get it on 2004, but it's getting fed 1909 automatically
14:51:18Strife89No button to upgrade it to 2004 yet
14:51:34Strife89And I don't have an installer stick handy to jump it straight there.
14:51:42TheEaterOfSoulsa11y on Linux is a mess and may not be a thing for much longer, at least not with GUI toolkits
14:52:00TheEaterOfSoulsYeah, I just used the ISO
14:52:22TheEaterOfSoulsNot downloading that on multiple machines
14:52:42speachythe GNOME folks keep trying on the a11y front but.. what's the saying about herding cats?
14:53:03fs-bluebotBuild Server message: Build round completed after 637 seconds.
14:53:04fs-bluebotBuild Server message: Revision e884140 result: All green
14:53:06TheEaterOfSoulsActually not sure
14:53:12speachy(since dumping on GNOME is how the KoolKids(tm) pass their time these days...)
14:53:29TheEaterOfSoulsWell, not really. part of the problem is that there just aren't many people to maintain it
14:53:43speachyyeah. not unlike rockbox, eh? :)
14:54:54TheEaterOfSoulsMuch, much worse than Rockbox
14:55:09TheEaterOfSoulsA lot of the code still says "Copyright 2001 Sun Microsystems Inc."
14:55:13speachythere are advantages to big top-down corporate budgets.
14:55:41TheEaterOfSoulsWhether they outweigh the cons I can't say
14:56:13TheEaterOfSoulsI would try and help maintain it, but I lack the skills currently
14:56:15speachythe entire traditional desktop stack is slowly bitrotting everywhere, even on Windows.
14:56:58TheEaterOfSoulsAnd in its place is the modern web, which is so much worse
14:57:22speachyyeah. and the most recent abobination, "electron apps"
14:57:41mendelmunkiswhat is ally?
14:57:46TheEaterOfSoulsI don't do Electron. Just won't use anything written in it
14:57:52 Nick mendelmunkis is now known as mendel_munkis (
14:58:26mendel_munkishow about a hardware driver tha the only semi usable version I found was electron!
14:58:51mendel_munkisjust why.
14:58:53speachymendelmunkis accessability. aka a[11 letters]y
14:59:06speachynot unlike i18n (internationalization)
14:59:15TheEaterOfSoulsOh, I was confused about that question
14:59:31mendel_munkisTIL: my font has 1 and l too similar.
15:00:10TheEaterOfSoulsBut yeah, it's way too buggy for me on Linux. Almost all the attention of the one Orca dev (screen reader) is focused on web handling, and it's still terrible
15:00:56TheEaterOfSoulsI'm grateful people at least try, but the current stack won't work in GTK4. There's an issue to rework it, but I really don't think it'll happen
15:02:15TheEaterOfSoulsConsole accessibility (without a display server) is better, even though the most lightweight screen reader is in the kernel, where IMO it does not belong
15:02:31speachyplumbing layers everywhere are being starved. why work on that when you can build a datamining app instead?
15:02:35TheEaterOfSoulsMaybe one day I'll reject the web entirely and just do that :D
15:03:26TheEaterOfSoulsYes, complexity and rushed garbage are in now
15:03:44speachykids these days, I swear..
15:03:52TheEaterOfSoulsBut we do have a very good open-source screen reader on Windows that gets attention, and UX is pleasant
15:03:55*speachy waves his cane and shouts at a cloud.
15:04:12mendel_munkisspeachy: I take offense at that.
15:04:15TheEaterOfSoulsApologies, I'm rambling
15:04:36*TheEaterOfSouls waves white cane and accidentally hits someone
15:04:51speachymendel_munkis, no offense intended.
15:05:13mendel_munkisdon't worry. my opinion of my generation isn't very high either.
15:05:44TheEaterOfSoulsI don't think it's a generational problem
15:06:23TheEaterOfSoulsI think big companies just scoop up most of the talent and...people are into this terrible stuff for some reason
15:06:26TheEaterOfSoulsThe scramble
15:06:53Strife89Basically, capitalism engulfed the tech world
15:06:53TheEaterOfSoulsWho cares about quality? We need features features features!
15:07:05TheEaterOfSoulsAnd don't forget DRM!
15:07:09Strife89Big profits, shun quality
15:07:42TheEaterOfSoulsMemory? Resources? What are those?
15:07:42speachyI remember thinking that, during the tech bubble that popped in 2000, that it was odd that "tech news" actually meant "financial market news"
15:07:43Strife89Especially the "give us money, every month, forever"-enabling DRM
15:08:24TheEaterOfSoulsOur rockbox players will outlast them all
15:08:28speachyI can't fault folks for chasing things they perceive as making them money.
15:08:28 Join jdarnley [0] (
15:08:43speachyit's quite rational
15:08:51TheEaterOfSoulsYeah, gotta play the game, and all that
15:09:03Strife89I mean, yeah, it's perfectly reasonable to want to make a living off of whatever you do
15:09:07speachywhere I object is when they forget they depend on us mountain trolls.
15:09:29TheEaterOfSoulsIMO it's not an end in itself, though. There are better things to pursue than money
15:09:42speachythe ideals of Free Software enabled this DRM-infested, adware-laden world.
15:09:52Strife89TheEaterOfSouls: Yep. You can't take it with you.
15:10:26 Quit J_Darnley (Ping timeout: 258 seconds)
15:10:34TheEaterOfSoulsBut maybe I'm just trying to make myself feel better. I'm not career-oriented, so will probably be poor forever
15:16:58 Join MrZeus [0] (~MrZeus@2a02:c7f:70d0:6a00:f41f:55f2:65ef:d305)
15:19:04Strife89lol, eefi is so slow that it built exactly ONE item in the last round :D
15:19:49Strife89It underperformed even gevaerts' Atom machine
15:20:03Strife89The laptop with WSL
15:20:35speachydon't forget ccache
15:20:38mendel_munkisstrife89: you are making me feel really bad that I haven't managed to get linux running on my planned build machine.
15:21:01Strife89mendel_munkis: Sorry about that
15:23:16TheEaterOfSoulsI'm really thinking of going the VirtualBox route again. VMs are slower to boot than WSL2, but it's tolerable
15:23:17Strife89TheEaterOfSouls: Also, "eefi" is basically the romaji version of the Japanese name for "Espeon"
15:24:11TheEaterOfSoulsI'm not imaginative with my hostnames
15:24:22TheEaterOfSoulsCurrently on t460s. Guess what that is?
15:24:35Strife89souls-t460s? :)
15:25:24Strife89I just wanted something more memorable than a model number (because I easily forget those).
15:25:54TheEaterOfSoulsMakes sense
15:25:56Strife89And I once saw someone's list of their computers, listed by hostname, and all of them were Japanese Pokemon names.
15:26:06Strife89So, as a Pokemon fan, I copied that
15:26:29TheEaterOfSoulsMy VPS has an arbitrary name. Nothing else, though
15:27:38mendel_munkisI currently have menashe, chizkiyahu, shlomo, and am trying to get a bootable uzziya.
15:28:07Strife89Those sound cool. What's your naming scheme?
15:28:34mendel_munkisIsraeli kings.
15:28:45mendel_munkis(it goes well with my domain name.)
15:29:37mendel_munkisTheEaterOfSouls: sorry if your reader couldn't handle those.
15:31:05TheEaterOfSoulsThe only things it doesn't handle are non-Latin languages
15:46:27speachyin a former life we had our database servers named "chomp, bachomp, chewy, bachewy"
15:59:48TheEaterOfSoulsMaybe I'll just stick with WSL for now
16:04:35 Quit jdarnley (Ping timeout: 240 seconds)
16:06:01 Join J_Darnley [0] (
16:07:35 Quit miner49er (Ping timeout: 245 seconds)
16:17:25***Saving seen data "./dancer.seen"
16:23:10 Join livvy [0] (~livvy@gateway/tor-sasl/livvy)
16:29:32 Quit J_Darnley (Ping timeout: 256 seconds)
16:29:51 Quit MrZeus (Read error: No route to host)
16:31:18 Join MrZeus [0] (
16:32:56 Join J_Darnley [0] (
16:37:47 Quit ac_laptop (Ping timeout: 240 seconds)
16:43:20 Join ac_laptop [0] (~ac_laptop@
17:02:23 Quit livvy (Ping timeout: 240 seconds)
17:05:21 Quit ac_laptop (Ping timeout: 265 seconds)
17:12:10 Quit lebellium (Quit: Leaving)
17:13:03 Join ac_laptop [0] (~ac_laptop@
17:23:05__builtinheh, lambda-builtin is crawling slow, but at least the UL speed is crazy fast :)
17:30:14speachyI'm rather disappointed lunchbox-speachy doesn't have a much faster upload speed. probably points at some sort of bottleneck in the upload script.
17:30:46speachy(it's plugged into the same GigE switch as the buildmaster..)
17:39:33__Bilgus_why the flip flop on removing HWCODEC?
17:42:39 Quit __Bilgus_ (Remote host closed the connection)
17:43:36speachyit still works, I guess
17:44:54speachyspeaking personally it's not in the way of what I'm trying to deal with
17:45:27__builtinyeah, I'm not sure if we really have a pressing need to remove it
17:45:57speachyI don't want a half-assed measure like simply disabling the build; if it's going to go away, it needs to be *gone*.
17:46:35__builtinthe question we need to answer is whether people still care about HWCODEC
17:47:19speachythere's probably still a couple of 'em out there. might be able to drill into the server logs to see if anything vaguely human-like is downloading 'em
17:47:28__builtinespecially whether any devs still care
17:48:07speachyjust the "don't break stuff that's working" principle
17:49:11__builtinand that too, of course
17:50:31speachythere's no maintainence advantage to removing 'em unless we strip out all the code too.
17:53:12speachyoh, it's not just hwcodec, it's non-bitmap displays too. :)
17:54:41speachythere's also still a _ton_ of documentation that considers the archos players the primary targets
18:15:52 Quit blackyus17 (Quit: AFK)
18:16:58 Join blackyus17 [0] (
18:17:26***Saving seen data "./dancer.seen"
18:21:14 Join kugel_ [0] (
18:21:14 Quit kugel_ (Changing host)
18:21:14 Join kugel_ [0] (~kugel@rockbox/developer/kugel)
18:22:14 Quit kugel (Ping timeout: 260 seconds)
18:37:13 Quit sakax (Quit: Leaving)
18:40:26 Quit J_Darnley (Ping timeout: 256 seconds)
18:40:55 Quit ZincAlloy (Quit: Leaving.)
18:41:50 Join J_Darnley [0] (
18:54:35 Quit kugel_ (Ping timeout: 240 seconds)
18:58:33 Join __Bilgus_ [0] (41ba23be@
19:00:22__Bilgus_the idea IIUC was to branch off or archive at least HWCODEC and then remove all ifdefs & code on a grep basis then finally remove vestiges left over as we run into them
19:13:10 Join kugel [0] (
19:13:10 Quit kugel (Changing host)
19:13:10 Join kugel [0] (~kugel@rockbox/developer/kugel)
19:38:53 Quit petur (Quit: Leaving)
19:41:37speachyok, here's a start: g#2545
19:41:40fs-bluebotGerrit review #2545 at : WIP: Remove sh support and all archos targets by Solomon Peachy
19:42:42speachyover 30K lines deleted. Still haven't excised the keypad mappings in the plugins or large parts of the ocumentation
19:43:17 Join reductum [0] (
19:43:45speachynearly all HWCODEC and LCD_CHARCELL stuff is still present
19:44:23 Quit reductum (Client Quit)
19:57:52 Quit MrZeus (Ping timeout: 272 seconds)
20:17:28***Saving seen data "./dancer.seen"
20:20:27speachyup to over 36K lines now.
20:20:45speachygit status
20:31:23speachyokay, that's the first passs.
20:45:46 Join livvy [0] (~livvy@gateway/tor-sasl/livvy)
21:32:49speachyHAVE_LCD_CHARCELL is another ~4.8K LoC.
21:34:59 Join aenertia [0] (~aenertia@
21:35:09aenertiaHi all.
21:35:34aenertiaAm trying to figure out if there is a usable target for the action-semi based mp4/mp3 players that seem prolific
21:35:53aenertiaform factor wise they appear very similar to the AGPtek playres
21:36:03aenertiabut generally have 8gb of EMMC
21:36:42aenertiaThey appear as : Bus 001 Device 009: ID 10d6:1101 Actions Semiconductor Co., Ltd D-Wave 2GB MP4 Player / AK1025 MP3/MP4 Player
21:37:14aenertiaThe Tool fear innoculum Limited Edition uses a similar SoC
21:39:57speachyI don't think anyone has looked into it. it's a reasonably capable SoC, even without using the decoder engines
21:46:22speachyassuming most of the hardware out there is using that reference schematic, the only real question seems to be how to get one's own code in there.
21:47:48speachygot a link to one?
21:54:26aenertiaThis is the one I bought:
21:54:44aenertiaIts branded as : Geruida on the Screen
21:55:08aenertiaalthough it powers on with a graphic that says Finesound by Akor
21:56:23aenertiaInfo screen shows: US212C Software Version; and has as support website
21:57:01aenertiaVersion V1.101.10
21:57:48aenertiaI want to customize menu layouts and replace BMPs/start-up screens for use as a Mixtape/Gift
21:57:56 Join bonfire [0] (
21:58:00speachygood luck with that..
21:58:53aenertiaIts been a long time since I looked at Rockbox, 2006/7 I had archos20 and iriver; but they are pricey as single shot mixtape gifts ;)
21:59:29aenertiaI have been doing a lot of esp32 development over the last 2 years
21:59:39aenertiaI am wondering if an esp32 port might be possible.
22:00:31aenertiaLack of USB host on the original esp32 is a problem, but the ESP32s includes hardware USB host now, so even without external DAC you can pretty much have a fully usable player
22:04:31aenertiapretty sure this is the same unit :
22:06:10aenertiaexcept with a csr bluetooth chip in addition
22:06:18speachythat ruizu-x2 is based on a different SoC, rockchip nano-b.
22:07:51aenertiaLooks excactly the same heh, with the exception of a 6th lock/return button
22:08:18aenertiadown to the shitty plastic slidre power switch on the bottom and the holographic logo and font
22:09:55aenertiapossibly its the d51...
22:10:29aenertiaalthough that looks like it has a couple of extra buttons on the sides
22:17:30***No seen item changed, no save performed.
22:18:58speachythe nano-b isn't really up for rockbox either, but the nano-c could probably manage, barely
22:19:23aenertiaI just shimied it open
22:19:39aenertiaGive me a few mins Ill take some pictures
22:19:48aenertiaFigure out what exactly the chips are
22:21:49speachyI have a really cheapie no-name dap here based on another Actions AK2117
22:27:20speachythe atj2127 will never see a rockbox port. it has something like 88KB of RAM total.
22:33:42speachyyeah, it's a lost cause for rockbox.
22:34:11Strife89Say, did Rockbox ever get ported to any swcodec players with alkaline batteries?
22:34:39speachyI guess you could have put alkalines in the original Archos models..
22:35:35speachythere were very few players that could run on standard batteries. I recall an early Creative model that ran on one AA (or AAA?) cell.
22:35:47speachynot for long, mind you.
22:36:23Strife89Just mulling over the longevity aspect
22:37:20speachyLIon batteries are really much better than everything else; much better power density, higher current capacity, more form-factor options..
22:37:34aenertiathere are two SOIC8 on the back of this board... /me wonders if I SPI dump them; is there any tools for firmware analysis for this platform. There are a couple of hits on Github that appear abandonded
22:37:46Strife89Yeah but we depend on suppliers to make them for each device
22:38:02Strife89And you definitely can't grab a new one from a nearby store
22:38:14aenertia18650s are easily available
22:38:18aenertiafrom any vape shop
22:38:27speachyyeah, they're generally pretty bulky
22:38:44speachybut they do need smart chargers lest things go all explode-y
22:39:48Strife89I have a bunch of Palm OS devices, see. Most of them use rechargables, and a lot of those are dead. There are so many different ones to replace that I've boxed a few of them to take care of later.
22:39:52aenertiaback to previous comment around esp32
22:40:09aenertiaa lot if not most of the dev boards have lipo charge cicuits built into them
22:40:49Strife89Meanwhile, I have two Visors and a couple of 3-Com-era models that run on two AAAs and have no problem keeping on with rechargable or alkaline ones
22:41:25speachyfor an off-the-shelf microcontroller, we'd need one with enough pins to handle external DRAM, boot flash, external mmc/sd, display, i2s (or builtin codec), and of course gpios or equivalent for buttons.
22:41:47speachythough we can probably get away with using internal flash for base firmware
22:42:37speachyit would also need to have the oomph comparable to at least a 100MHz ARM7 to be minimally viable
22:42:54speachyoh, high speed USB2 (therefore ULPI)
22:43:18aenertiaesp32s are 320Mhz
22:43:20speachyby the time you add that together, plus the external DRAM, the BoM cost is already scary even before one considers physical cases
22:43:28aenertiadual core and have offloads
22:43:35speachywe'd be better off building a case around a RPi Zero. :)
22:43:51aenertia320k usable RAM ; and the WROVER-D modules have an additiona 8Mbit of PSRam
22:44:01speachyoffloads are pretty useless for rockbox.
22:44:06aenertiafor 4$ they wipe the floor with similar cortex m0 boards
22:44:13aenertiaoh and they have full bluetooth and wifi stack
22:44:18Strife89I for one would be fine with a RPi0 player
22:44:27aenertiaeugh gross
22:44:37aenertiaYou then have a full OS you have to manage
22:45:00speachynot really −− the full OS is someone else's problem.
22:45:19aenertiaIve completely gone off PI for IOT/Home Automation stuff because they break in all sorts of interesting ways I dont want to have to deal with for permanent install single purposes stuff
22:45:21speachywe can always go bare-metal too
22:45:53speachyI mean, porting rockbox to a new hardware platform (and cpu platform too, in the esp32 case) is qutie a lot of work
22:46:06speachyand far be it for me to tell folks what to do with their spare time
22:46:26aenertiais there any stm32 target?
22:46:29speachybut something like that won't be generally useful to other folks that want to put music in their pockets.
22:46:42speachynope, though I'm toying with writing one
22:47:16aenertiaGenerally from AVR code to espresif or Stm32 ... it tends to be a couple of defines and pin changes to port
22:47:38speachy320K IRAM is usable, but only 8Mbit of PSram would make it the lowest-end port, memory-wise. even worse than the archos targets.
22:48:04aenertianot 8mbit
22:48:21speachyah, ok. you said 8Mb :)
22:49:09speachyI consider 8MB to be the minimum viability point with a full RB feature set (color screen, voice)
22:49:17aenertiaThere are heap of Dev boards which have I2C oled, 18650 , rocker switch and TF card slots
22:49:32aenertiaA couple with touch tft etc
22:49:47speachyby the time you find one with all the bells and whistles the price isn't so cheap any more.
22:50:12aenertiaPins are all software definable bar for a couple of special pins for hall effect sensor and capacitive adc samples
22:50:23aenertia10$ gets you a WROVER-D cam kit
22:50:31aenertiaI have 5 sitting on my desk
22:50:45aenertiaAnd that is with a 1600x1200 webcam you can throw away
22:51:42speachyfully open toolchain and libraries?
22:52:02aenertiawith the exception of the wifi stack
22:52:20aenertiai pretty much dont touch the avr/arduino stuff.
22:52:34aenertiaor the stm32 - because the esp32 is so much nicer to work with due to the wifi
22:53:03aenertiai have a couple of stmf7 discovery boards, that are gathering dust next to the h3 SBC Pi clones
22:54:38speachyI was thinking of going with one of the stm32h7 boards
22:55:30speachyI have a fond spot for the stm32 family, having done a ton of work with 'em in a former life.
22:56:31aenertiaYeah - they are nice ; especially if you are doing high precision analog/pwm
22:56:37speachybut even then it would still be a time-consuming vanity project
22:57:04speachythe h7 is vast overkill cpu-and-peripheral-wise, but I like it's large amount of IRAM.
22:57:13aenertiaJust - when the Wifi SoC has more oomph than the target, kinda makes it pointless
22:57:35speachysuch is the reality of modern wifi.
22:58:11aenertiaDefinately grab a wrover-d board off aliexpress and have a tinker. For the price of a cup of coffee
22:58:24speachyI cut my teeth making wifi modules based around the stm32f1 and their cw12xx wifi chipset.
23:00:15speachyif I'm being honest and wearing a "rockbox coder" hat, my time would be better spent getting rockbox ported to commercially-available hardware.
23:00:33speachymuch as the idea of rolling my own appeals to me.
23:00:57*speachy looks over at a closet full of half-finished microcontroller projects
23:11:21speachyah, the PSRAM is SPI-attached. that's going to incur a substantial performance penalty.
23:14:18speachyand only 4MB is actually usable at a time. hmm.
23:21:43 Quit livvy (Ping timeout: 240 seconds)
23:34:20Strife89I have no patience
23:38:00Strife89(Still waiting for the replacement battery so I can actually *use* the mini.)
23:40:54 Quit [7] (Disconnected by services)
23:41:00 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)

Previous day | Next day