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 2014-02-16

00:05:51***Saving seen data "./dancer.seen"
00:17:10[Saint]nonMOOSE: It might be a good idea to familiarize yourself with our fine manual.
00:17:56[Saint]Though, although Rockbox has changed a lot in the past 4~5 years, at least two of the things you stated never happened, ever, 5 years ago or otherwise.
00:18:31[Saint]...and, soap already said that. Yay. I'm blind today.
00:26:48 Quit ender` (Quit: Just because I have a short attention span doesn't mean I)
00:37:25 Join Water255 [0] (
00:41:40Water255Is it okay if I ask, here, how much a coder would charge me to modify the source code to do something I want?
00:42:45wodzI guess you can ask what has to be done not how much would it cost
00:50:13 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
00:54:20[Saint]I would /hope/, that the active developers wouldn't be moved by money, though.
00:54:40[Saint]I would much rather assist someone to do what they wanted than take a paycheck for it.
00:55:01gevaertsWell, it depends on a lot of things
00:56:18[Saint]In an ideal world I would like to see any funds claimed diverted back to Rockbox. But, that's me.
00:56:40[Saint]Of course, we still don;t know that what you want to do is even possible. :)
00:58:01gevaertsSuppose someone wants something done that's a *significant* (e.g. several months full time) amount of work, and not generally useful, you'd basically need someone to take this up as a job
00:58:17gevaertsI'm guessing that's not the case here though :)
00:58:30[Saint]Right, but it irks me that that would be on the back of thousands of hours of unpaid work.
00:58:40[Saint]ANd none of those developers get paid.
00:59:12[Saint]I know there's technically nothing wrong with it, but, it doesn't sit right with me.
00:59:24gevaertsYou should have worked on a CC NC project then :)
00:59:31[Saint]Heh. :)
00:59:34bertrikAren't most requests either keymap changes (because a button broke, or because someone's finger hurts somehow) or changes to remove features (e.g. radio)?
01:00:07[Saint]I guess we need to wait with baited breath and see what Water255 actually wants.
01:01:12Water255Wow, sorry guys, I just realized i was getting replies... reading.... oh, btw this is what I want: How much money would a anyone want to modify the Fuze+ source so that it does this: When device is in hold mode, the mechanical 'volume up' key works like the 'play / pause' key, and mechanical 'volume down' key works like the rewind key?
01:01:40gevaertsThat's just *wrong*!
01:01:43[Saint]You could do that yourself *incredibly* trivially.
01:01:50[Saint]And, yes. It is wrong. :)
01:02:30saratogayeah just edit the keymap file, then follow the instructions to compile
01:02:49saratogaoh wait while the screen is on hold
01:02:53saratoganot sure how that works actually
01:03:28[Saint]I imagine just the same as software hold works.
01:03:38[Saint]*unless* the hold key actually disables the buttons.
01:03:42[Saint]Then you're screwed.
01:03:54Water255they wake the screen while in hold
01:03:54saratogaIIRC on software hold devices button presses are dropped, so probably you have to edit the software hold code to not do that
01:10:29pamauryediting the code to do that shouldn't be very hard but the "to rewind" part kind of irk me
01:10:58Water255okay - 'skip back' button?
01:12:05Water255I have a build environment set up.
01:12:24pamauryno I mean, you could easily make it rewind if you want, I just don't understand why you would want to do that :)
01:13:26Water255i listen to speech a lot while working. Sometimes I'll miss a few seconds, and I need to skip back.
01:14:16pamauryok, make sense
01:14:29[Saint]We have a function for that, do we not?
01:14:34Water255I'm sure making the buttons do this would be pretty easy, but "only while on hold" is where I wouldn't know where to start
01:14:34[Saint]rewind N seconds?
01:14:47*[Saint] is pretty sure we can do that.
01:15:14Water255So... RB contribution?
01:15:26pamaury[Saint]: we haver a rewind on pause function I think
01:15:55[Saint]Ohhhh, sorry. I'm thinking of skip length.
01:16:00[Saint]Yeah, not quite the same.
01:16:14Water255I just want to button to do <whatever> i have the real rewind button set to do in the settings.
01:17:29pamaurythat sounds pretty doable, at least in an ugly way ^^
01:17:40Water255ugly is fine
01:19:30wodzpamaury: Any recommendation for string hashing function? I did google survey but I can't really judge anything except expected computational effort
01:19:40 Quit akaWolf (Ping timeout: 252 seconds)
01:20:27Water255And for the mechanical volume-up button to do whatever the 'play / pause' button does (i.e. long-hold == stop and create bookmark).
01:21:55pamaurywodz: have a look at this:
01:22:02pamaurylooks very serious
01:23:28JdGordon[Saint]: shutdown in menu needs a new LANG id.... i tried it last night and with an old lang file it messed things up
01:24:26[Saint]If the user installed the new build, it would come with a new lang file, so I don't understand the problem.
01:24:54[Saint]A partial install is a good way to break stuff, yes.
01:26:00[Saint]Hmmm...though I guess that only works right off the bat for english.lang
01:26:26 Quit ZincAlloy (Quit: Leaving.)
01:28:10[Saint]If the lang file and binary are from the same build, it should "Just Work", I believe.
01:29:03[Saint]I haven't seen any issue with English, but I don't test any other languages. Nor do I deliberately test with old lang files..because, why?
01:30:22pamauryWater255: in any case you're first step is to setup a build environment
01:30:45[Saint][13:12:07] <Water255> I have a build environment set up.
01:30:46Water255Already set up.
01:33:01pamaurysorry, missed that message ^^ I don't remember the code enough to propose a patch right now but I can have a look for tomorrow maybe
01:33:20JdGordon[Saint]: i agree... but yeah, we try not to break old lang files remember? :)
01:34:20[Saint]JdGordon: It would seem to me that this would be working around a broken installation.
01:34:29[Saint]I'm not sure why we should do that, personally.
01:34:38Water255pamaury: That would be great. Thank you.
01:34:46[Saint]If you replace just the binary - you're asking for trouble.
01:35:02[Saint]ZZFull install? Works. All is well. Everyone happy.
01:35:54[Saint]And even if we did give it a new LANG id, if the user had an old lang file, it *still* wouldn;t work.
01:36:05[Saint]..because neither the old, nor the new, ID would be there.
01:36:07wodzpamaury: it looks like fnv1 or fnv1a are good candidates.
01:37:05[Saint]I see this as a non-issue.
01:37:20[Saint]Even if I change the lang id, an old lang file is still going to shit it up.
01:37:27[Saint]As it should.
01:37:47 Quit wodz (Quit: Leaving)
01:37:55pamaurywodz: my guess is that you should try on actual data see if you get collisions
01:38:53[Saint]So, yeah: TL;DR - Update your bloody lang file with the build. ;)
01:55:51 Quit lebellium (Quit: ChatZilla [Firefox 28.0/20140210161136])
01:56:25 Join Zarggg [0] (
02:01:50 Join Strife89 [0] (
02:05:52***Saving seen data "./dancer.seen"
02:33:30 Join Water255_ [0] (
02:34:03 Quit Water255 (Ping timeout: 252 seconds)
02:34:10 Nick Water255_ is now known as Water255 (
02:43:39 Quit bertrik (Remote host closed the connection)
02:46:18JdGordon[Saint]: this is what we do... this is what we always do. if the string isnt in the lang file the user will get the inbuilt english string
02:47:01[Saint]But the string _should_ be in the lang file.
02:47:11[Saint]If it isn't, by definition, its a broken install.
02:47:28[Saint]You can't expect a mismatched install to work. Nor should you.
02:47:52JdGordonenglish.lang is compiled into the binary
02:49:15[Saint]Right. ANd it should bloody well "Just Work" if you're using English and/or a current *lang file.
02:49:30[Saint]I really don't understand the problem.
02:49:44[Saint]If you have old lang files floating around, fix it.
02:51:06[Saint]I've had two people tell me that the system is designed to cope with this and adding a voice string to a previously unvoiced lang ID is perfectly fine.
02:51:18 Quit dfkt (Remote host closed the connection)
02:51:23[Saint]...and I've seen no issue personally.
02:52:45[Saint]That being said, if there's some problem I'm just not seeing, I believe you know how gerrit works.
02:55:53[Saint]I feel I should point out that all I did was bring your origial version that did the exact same thing up to date. ;)
02:56:30JdGordoni know...
03:07:53 Quit Gallomimia (Ping timeout: 260 seconds)
03:32:11 Nick DormantBrain is now known as SuperBrainAK (~andy@2001:470:8:a61::5f92:59a1)
03:40:06 Quit Water255 (Quit: ChatZilla [Firefox 25.0/20131025150754])
03:40:14 Quit pamaury (Ping timeout: 245 seconds)
04:05:56***Saving seen data "./dancer.seen"
04:11:29 Join ungali [0] (~ungali@unaffiliated/ungali)
04:14:02 Quit endee_ (Ping timeout: 264 seconds)
04:16:47 Join endee_ [0] (sid15467@gateway/web/
04:23:13 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:23:13 Quit amiconn (Disconnected by services)
04:23:13 Quit pixelma (Disconnected by services)
04:23:13 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:23:15 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:23:15 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:40:14 Join bzed_ [0] (
04:41:31 Quit bzed (Ping timeout: 245 seconds)
04:41:38 Nick bzed_ is now known as bzed (
05:05:41 Quit Zarggg (Ping timeout: 245 seconds)
05:11:48 Quit [7] (Disconnected by services)
05:12:01 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
05:34:27 Quit Strife89 (Quit: Leaving)
05:34:40 Join Strife89 [0] (
06:05:57***Saving seen data "./dancer.seen"
06:38:48 Quit alexbobp (Ping timeout: 245 seconds)
06:39:52 Join alexbobp [0] (
07:27:21 Nick SuperBrainAK is now known as DormantBrain (~andy@2001:470:8:a61::5f92:59a1)
07:41:20 Join Zarggg [0] (
08:04:39 Quit ps-auxw (Ping timeout: 260 seconds)
08:06:01***Saving seen data "./dancer.seen"
08:11:26 Join akaWolf [0] (~akaWolf@unaffiliated/akawolf)
09:00:58 Join draft [0] (
09:04:01drafthas anyone installed Rockbox on iPod 4G Nano?
09:33:37 Join ender` [0] (
09:51:55 Join lebellium [0] (
10:06:02***No seen item changed, no save performed.
10:23:45 Join guestnano2g [0] (b3b0c334@gateway/web/freenode/ip.
10:38:30 Part ungali
10:40:53guestnano2ghello! I'm developing a theme for ipod nano2g (and publish in Rockbox theme site :D ),I wanted to know if you still have people trying to fix USB for ipod nano2g,is too slow (and dangerous) rebooting in apple firmware constantly.
10:49:51pixelma[Saint]: reading last night's log, I thought JdGordon was talking about older voice files which aren't updated automatically with the build. There I'm not sure if kugel's earlier comment about IDs was correct. IIRC what happens if you don't do this correctly is that voice users get different things spoken for all strings below the changed one in the source. When done right thebn at least existing things are still correct and only the new one won't
10:49:51pixelmabe spoken with an old voice file.
10:54:58 Join wodz [0] (
10:55:28wodzholly crap! if this is right,48014.msg227064/topicseen.html#msg227064 , clip sport is based on atj2127 SoC
10:57:25wodzThat would be really pity as it is truly weird SoC
10:59:14coppernot loading for me
10:59:50copper"waiting for…"
11:00:44wodzbasically if this IS atj2127 rockbox for this little player will nott happen
11:01:13coppernot well documented enough?
11:01:52lebelliumnot loading either
11:02:11wodzThat is only part of the story. This SoC is mips4k core + DSP with VERY limited memory splited into a few banks
11:02:25copperso, not ARM
11:02:37wodzit is not like general purpose processor
11:02:57copperwell, that sucks
11:02:58wodzthe memory layout is truly awkward
11:03:02copperI was almost tempted
11:03:43wodzThat is based on what I know about atj2127. Maybe someone proves me wrong
11:04:31copperthat's weird
11:04:46copperdoesn't that mean more work for SanDisk to develop the firmware?
11:04:48wodzthe mem is 88kB total IIRC
11:04:55lebelliumwodz: I don't see where it is written in the forum thread
11:05:10wodzcopper: they got sdk from actions to start with
11:05:18wodzlebellium: look at the pictures
11:05:27wodzfrom the last post
11:05:41copperthere are none
11:05:44lebelliumthere is no picture in the last post
11:06:11wodzthere are links
11:06:23copperone link
11:06:28copperno picture there either
11:06:49copperI'm not registered though
11:07:17lebelliumyou talk about ABI
11:07:23lebelliumwhy linking the rockbox forum then
11:07:42wodzno, I read this from our forum
11:07:56lebelliumokay I see the CPU now
11:07:56[Saint]I almost want to believe that silkscreen is trolling...
11:07:58copperthat's on ABI
11:08:18[Saint]That's a very odd choice.
11:08:30[Saint]Even if you are handy with the schematics.
11:08:44copperwhy is it 4 times the same pic
11:09:15copperneeds moar RMAA
11:20:28copperwodz: 88 KiB of total RAM?
11:20:39copperthat sounds insane
11:22:28pixelmahmm, now thinking about voice files... more things come to mind. I wonder what (if anything) kugel changed lately for voice files not fitting in RAM on harddisk based Archoses. Earlier it was recommended to squeeze the size of the voice file (different quality, stronger compression) until it fits. Chunked loading as done for the flash based Ondio has disadvantages on harddisk - you would need to wait for the disk to spin up which also drains the
11:22:28pixelmabatteries way more
11:22:55copperwodz: that's not even enough for a 128x128x16 bitmap
11:23:34coppere.g. a backdrop
11:27:05 Join bertrik [0] (
11:27:05 Quit bertrik (Changing host)
11:27:05 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
11:31:21copper"4000 folder/file database limit"
11:31:26 Join rela [0] (~x@pdpc/supporter/active/rela)
11:34:03guestnano2gSIL Open Font License is acceptable for upload in rockbox theme site?
11:37:23 Join Rower [0] (
11:37:30[Saint]guestnano2g: seems so.
11:39:42guestnano2goh thanks for the reply :) have a nice day!
11:42:01 Part guestnano2g
11:48:31copperwodz: actually, 88 KiB would not even be enough for 76x76x16 album art, as pictured on the SanDisk renderings
12:06:04***Saving seen data "./dancer.seen"
12:32:23relais the clip sport new?
12:33:52relahow are the volume buttons on this one? i thougt the ones on the zip were to small and with bad feedback
12:35:40 Join y4n [0] (~y4n@unaffiliated/y4ndexx)
12:36:51 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
12:44:28pamaurywodz: this is pretty bad news if it's the case
12:44:47pamaurymaybe the atj is just a companion chip and not the main one ?
12:48:21pamaurythis thing is just horrible to program, we will never be able to port rockbox to it if that's really it
12:49:19relawat? no rockbox for the sport?
12:50:12[Saint]This stage it seems unlikely, indeed.
12:50:14wodzpamaury: yeah. Fitst we need confirmation about the hardware
12:50:23pamaurywodz: yeah we need to wait
12:50:41copperdo you guys think there's another chip on the other side of the PCB?
12:50:58[Saint]Its possible. SUre.
12:51:09pamaurypossible but I'm pessimistic
12:51:22[Saint]Reasonanly unlikely, with button placement and LCD. But /possible/.
12:51:31pamaurybased on this:
12:53:18[Saint]You kinda have to wonder why.
12:53:21pamauryI will ask the guy to disassemble it more
12:53:38relawell... i will cancel my order right away and wait :)
12:54:17[Saint]I really thought there was a chance this would be an imx target.
12:55:05wodzwhy the didn't use atj213x, then my work on the platform would get the spin :P
12:55:25[Saint]Inconsiderate bastards.
12:55:27relaits like they just dont know how homebrew fires up the sales
12:55:37relaor they dont care
12:55:42pamauryI still don't understand how you can sell something like an ATJ2127 in 2014, it's just so wrong
12:56:00[Saint]rela: I suspect they don't care.
12:56:07relapamaury: probably they found a stack of those in the basement somewhere
12:56:28[Saint]the amount of people buying this specifically for Rockbox, even if it was compatible, is *tiny*.
12:56:39[Saint]So tiny they needn't give a shit.
12:56:44relapossible, yes
12:56:58wodzpamaury: you can if its gonna be dirty cheap
12:57:01copper"It is a low-cost, low-power, high-efficiency digital multimedia chip" ← I guess that's why?
12:57:23[Saint]There's many other non-suck contenders in that class.
12:58:19pamaurythe win is unclear to me: you have such a small amount of memory you either have to rewrite everything specifically for it OR to swap all the time, which is extremely power hungry
12:58:34copperSNR specs aren't very impressive
12:59:03relathe clip zip / + always had great audio quality. it sounds like this one will lack this great sound
12:59:38[Saint]I think the 4K file/folder limit is hilarious.
12:59:45[Saint]That's an immediate turnoff.
13:00:10[Saint]Joe Average won't care, though.
13:00:38relasaint: just merge your files to one big audio stream
13:00:43relaproblem solved
13:01:07relafast forwarding to the song you like can be tricky, but hey, thats what we did with audio tapes too
13:05:21copper4000 four minutes long songs at 128 kbps is about 15.3 GB
13:05:52copperwhich happens to be the largest storage model which will come out later
13:05:58copper(16 GB)
13:07:02copperthe MSRP is quite high for such a limited DAP
13:11:21 Join ikeboy [0] (
13:15:02relaunfortunately i am not happy with the volume rocker on the clip zip
13:15:24relaits too small, not much feedback from the buttons
13:15:35relaalmost impossible to use with gloves
13:17:35wodzI'd be very surprised if something of clip's size be comfortable to use with gloves.
13:23:20copperI guess they figured that the only remaining market for DAPs is gym / workout stuff
13:23:54copperand I wonder if they misinterpreted the failure of the Fuze+ as "it was too big"
13:24:05wodzpamaury: I create symbols lookup table with something like this So in source .c file you add EXPORT_SYMBOL(function_name); and the entry automagicaly is created. Now what to do with functions implemented in separate .S files?
13:25:04wodzhmm, maybe add ASM_EXPORT_SYMBOL equivalent
13:29:42 Quit rela (Read error: Connection reset by peer)
13:30:02pamauryto the same in assembly
13:30:08pamaurywith ".section"
13:31:32pamauryor maybe you can detect the file type, I have all assembly files are compiled with some special define
13:31:37pamaury*I think all
13:32:25 Quit [Saint] (Remote host closed the connection)
13:32:31pamaurylike __ASSEMBLER__
13:33:08 Join dfkt [0] (dfkt@unaffiliated/dfkt)
13:33:09pamaurythat's very strange: my interrupt nested code causes the stack to goes unaligned (not even on word), I can't see why
13:33:58 Join [Saint] [0] (~saint@rockbox/staff/saint)
13:40:27 Quit ikeboy (Quit: Leaving)
13:40:38wodzwhat was the switch to only preprocess the file -E ?
13:47:19wodzpamaury: this almost work except that preprocessor doesn't emit newlines and assembler barfs
13:50:17pamauryyeah -E or -M I don't recall
13:50:54pamauryhum, I don't know how to put newlines
13:54:09 Quit bertrik (Remote host closed the connection)
13:55:44 Join ZincAlloy [0] (
13:59:27ZincAlloywow, that is really bad news about the clip sport :(
14:00:59 Join bertrik [0] (
14:01:00 Quit bertrik (Changing host)
14:01:00 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
14:04:13pamauryyeah, I guess that means no port
14:04:43ZincAlloythat explains the shitty firmware sandisk came up with
14:05:16pamaurythey don't learn from mistakes
14:05:59ZincAlloyon the sandisk forum people are already stating that "rockbox could make this a good player" ^^
14:06:07***Saving seen data "./dancer.seen"
14:07:08ZincAlloybummer. a clip zip with bigger and better screen and better battery life are exactly what I want
14:27:48pamauryI guess now it's "rockbox could have made this a good player" ;)
14:28:15ZincAlloyapparently it couldn't.
14:28:53ZincAlloyI guess sandisk was just being too cheap
14:30:46pamauryor maybe wodz is incredibly motivated to do the port ?
14:31:19ZincAlloywould it be even possible to make it run reasonably well?
14:35:37pamauryif the soc is what we think it is, I think not
14:36:42ZincAlloyno point then
14:37:44pamaurythe problem is the lack of memory and the weird memory architecture
14:38:24ZincAlloyis it possible to use the flash as additional memory?
14:39:47pamauryyes but that's 1) slow 2) difficult to program with so less memory 3) bad for the flash life
14:39:57ZincAlloyI see
14:42:08pamauryand if I read correctly, it uses a hardware audio decoder ?!
14:42:39ZincAlloyI think so
14:47:05copperthey went full cheapo
14:47:38ZincAlloybecause they were targeting this thing to a very specific audience. which is not going to work. people are already complaining
14:48:21pamauryor maybe the thing has an even weirder DSP like other ATJ
14:48:38ZincAlloyit does say atj2127
14:49:16pamauryin the SoC, many atj have mips core + dsp
14:53:27dfktfrom a glorious Austrian SoC to some random Chinese one? i bet polka won't sound good on the zip sport
14:55:17ZincAlloyI guess they traded in quality for battery life..
14:55:33pamauryATJ is not exactly a random chinese one, it's has been here for some time
14:56:18 Join cramler [0] (
15:02:49pamaurywodz: did you solve the problem with assembly macro ?
15:05:46pamaurywodz: otherwise, you can define a gas macro and call it
15:17:30cramlerGreetings! I thought I might blarb in here, since the forums seem to go on and off right now. I recently got a "zen x-fi3" and thought I might just get my hands on the keymaps, but I am experiencing problems building rockbox correctly it seems. I have no problems building my own bootloader (v1.00.25), and building (unaltered) rockbox gives no errors, but starting it up on the Zen gives a "No .rockbox directory" and "Installation incomplete
15:17:31cramler". Tried on two machines, one with gcc 4.4.4, the other with 4.6.3. Are those known to be problematic? Any ideas?
15:19:02cramlerI also tried Amaury Poulys bootloader with same results, but his rockbox build works for me...
15:19:48pamaurycramler: so you correctly flashed the bootloader and it runs rockbox but then you get a "no .rockbox" message in a splash message ?
15:20:06cramlerpamary: Yes
15:20:29pamaurystrange indeed, give me 10 minutes and I'll check on mine
15:21:39cramlerAs I said, using your build works fine, but mine wont do.
15:21:56pamauryoh I see
15:22:12pamauryby my build you mean the daily builds ?
15:23:56cramlerIt compiles my custom bootloader correctly, but the actual rockbox builds that I compile dont work.
15:24:10pamauryinteresting, are you using our cross compiler or a custom one ?
15:24:36cramlerI exactly followed your tutorial.
15:25:12cramlerSame Problem on two different host computers.
15:25:53cramlerRunning Ubuntu 12.04, otherwise vanilla.
15:27:50cramlerSorry, if confusing... regular updates (12.04.3), but no custom software, aside of the toolchain.
15:28:01 Join ps-auxw [0] (~arneb@2001:470:c807:0:1532:4e5f:2ad3:4123)
15:29:52pamaurycramler: have you checked the file system for errors ?
15:30:49pamauryI'm doing a custom build, let's see
15:36:32pamaurycramler: I'm sorry, lunch time, I'll be back in 15 minutes to try
15:36:57cramlerthank you
15:54:56pamauryok, so my custom build works here. Did you correctly create a zip at the end of the build and unzip it at the root of the device
16:03:40pamaurythat must be a build problem but I don't see why. Can you run this in a console: "arm-elf-eabi-gcc -v" ?
16:05:18cramlerOutput as follows:
16:05:20cramlerUsing built-in specs.
16:05:22cramlerTarget: arm-elf-eabi
16:05:24cramlerConfigured with: ../gcc-4.4.4/configure −−target=arm-elf-eabi −−prefix=/usr/local −−enable-languages=c −−disable-libssp −−disable-docs MAKEINFO=missing
16:05:26cramlerThread model: single
16:05:28cramlergcc version 4.4.4 (GCC)
16:06:10***No seen item changed, no save performed.
16:07:01cramlerSince my bootloader is built correctly (functional with your builds), I am a bit lost...
16:07:25pamauryyeah, since you build the bootloader yourself, that's quite a surprise
16:07:31bertrikpamaury: at least we see a JTAG interface on the clip sport
16:08:04pamaurycramler: did you use any advanced option when compiling the main build ?
16:09:55cramlerNo. Creating rockbox with option 94 for x-fi3, then (N)ormal.
16:09:58cramlerTried both "make -j" and just "make"...
16:10:35pamauryhave you tried to remove the .rockbox directory entirely and put it back again ?
16:12:02cramlerI tried any combination I could think of, using both your/mine bootloader and builds.
16:13:49pamauryand so whatever you use, your main binary fails but not mine
16:18:05wodzpamaury: gas macro works as intended
16:18:33pamaurycramler: can you try this one:
16:19:13pamaurywodz: great to hear that
16:20:27wodzits just slightly inconsistent as in C you need to call EXPORT_SYMBOL(name) and in .S you need to call EXPORT_SYMBOL name. I think we have to live with that
16:20:56pamaurywodz: you can have a C macro whichs the gas macro
16:21:06pamaury*which calls
16:21:16wodzha, apparently
16:22:50wodzyup, that works :-)
16:23:26cramlerpamaury: The one you uploaded just now also works like a charm.
16:24:13pamaurydid you use our script to build the cross compiler ?
16:24:25 Quit Unhelpful (Ping timeout: 245 seconds)
16:24:40wodzNow what left is to write little program to patch symtab hashtable in elf. Should be easy.
16:26:55 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
16:27:40pamaurycramler: is your copy of the git repo up to date ? which is your HEAD revision ?
16:31:38cramlerI am a bit new to git, how can I check that? I cloned it two days ago.
16:31:40 Quit Unhelpful (Ping timeout: 260 seconds)
16:33:03pamaurycramler: "git log | head"
16:33:09pamauryassuming you didn't modify the source
16:34:19 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
16:34:43cramlerThank you! Output as follows:
16:34:47cramlercommit 101d10ff6c6ee6992504700452901fd779cbb26e
16:34:49cramlerAuthor: Marcin Bukat <>
16:34:51cramlerDate: Fri Feb 14 08:37:01 2014 +0100
16:34:55cramler atj213x: rework crt0.S in test code
16:34:59cramler Rework init code so relocation is safe for binaries linked
16:35:01cramler into cached unmapped KSEG0 region.
16:35:05cramler Change-Id: I705ee3f0334f8998237f249a2c3059d6dbe78a43
16:36:05wodzATJ stuff again :P
16:36:14pamaurythat's a sign ;)
16:36:44pamaurycramler: I'm confused, I have no idea of why it's failing
16:36:57pamaurydid you try to run "make clean" and restart a full build also ?
16:38:28cramlerYes, as I retried it on another machine: No change :/
16:38:30 Join saratoga_ [0] (4160ac54@gateway/web/freenode/ip.
16:38:35 Quit Unhelpful (Ping timeout: 252 seconds)
16:38:47 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
16:39:09saratoga_given that the sport can do album art and vorbis it probably has embedded dram on the package
16:39:55saratoga_like the original clip where Sandisk used a chip with 320kb of ram and then customized it with on package dram
16:41:05saratoga_but there's no way you can decode vorbis with less than about 300kb of ram for the decoder alone, not counting the os and various buffers
16:42:21pamaurysaratoga_: yeah, maybe the package markings is just "wrong", ie they got a custom chip
16:42:30pamaurybertrik says there are JTAG pins, that could be useful to check
16:43:07pamaurycramler: I have to go, but I'll be back tonight. In the mean time, I suggest you try variations of the build, like Advanced (A) and enable things like Log (L) and maybe Debug (D), see if that changes things
16:43:26pamauryIn last resort, upload your build so I can have a look at it
16:44:23cramlerpamaury: I will do that. Thank you so far!
16:44:24pamauryI think ATJ chips have a recovery mode, someone should try to reach it
16:44:50saratoga_Sandisk doesn't seem to want to be clever about software optimization either so my guess is they use the same 8 MB the zip had
16:45:56pamaurythe chip looks awfully small though, can you fit 8MB of ram in it ?
16:46:01ZincAlloyhow would that recovery mode be reached?
16:46:09bertriksomeone ought to record something on our wiki about this
16:50:57 Quit alucryd (Quit: Alucryd out.)
16:51:00 Join rela [0] (~x@pdpc/supporter/active/rela)
16:54:37saratoga_the sil screening on the PCB says it was outsourced to these guys
16:57:35relacopper: i dont mean winter gloves, more like thin mountainbiking-gloves
16:57:56wodzatj recovery mode can be reached by shorting nand datalines during boot
16:58:59ZincAlloyso not something we could ask a random user to do
17:03:41 Join kuldeepdhaka [0] (~kuldeepdh@unaffiliated/kuldeepdhaka)
17:08:17 Quit rela (Read error: Connection reset by peer)
17:13:21 Join alucryd [0] (
17:13:21 Quit alucryd (Changing host)
17:13:21 Join alucryd [0] (~alucryd@archlinux/trusteduser/alucryd)
17:17:42 Quit saratoga_ (Ping timeout: 245 seconds)
17:23:33 Quit alucryd (Ping timeout: 246 seconds)
17:25:30 Join alucryd [0] (
17:25:34 Quit alucryd (Changing host)
17:25:34 Join alucryd [0] (~alucryd@archlinux/trusteduser/alucryd)
17:33:24lebelliumno chip on the other side of the Clip Sport :(
17:34:45ZincAlloyand a 220mAh battery
17:35:34bertrikpretty good to get 25 hours out of that
17:36:02ZincAlloythat SoC must be rather efficient
17:40:33lebelliumI get 25,5hrs on Samsung YP-U7 with color screen and a 120mAh battery :)
17:46:27lebelliumwhat's the minimum amount of memory needed for Rockbox?
17:49:39wodzwell thats on arm
17:49:48 Join tertu3 [0] (~tertu@
17:49:48 Nick tertu3 is now known as tertu (~tertu@
17:50:12 Quit saratoga (Ping timeout: 245 seconds)
17:55:20lebelliumwodz: do we talk about 88KB for Clip Sport vs 2MB for ARM or am I wrong?
17:57:50wodzyou asked how much ram is needed for rockbox. 2MB is the bare minimum. Old archoses have this BUT the decoding is in hw. AFAIK there are some arm targets also that limited but decoding in software.
17:58:12lebelliumyes sure I understood tat
17:58:27lebelliumbut I was asking if 88KB for Clip Sport is the RAM or something else?
18:02:02 Quit TheSeven (Disconnected by services)
18:02:16 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
18:03:52 Join mc2739_ [0] (~mc2739@rockbox/developer/mc2739)
18:03:56 Join jmspeex_ [0] (
18:03:58 Quit mc2739 (Disconnected by services)
18:04:03 Join markun_ [0] (~markun@rockbox/developer/markun)
18:06:13***Saving seen data "./dancer.seen"
18:08:13 Quit wodz (*.net *.split)
18:08:14 Quit prof_wolfff (*.net *.split)
18:08:14 Quit markun (*.net *.split)
18:08:14 Quit jmspeex (*.net *.split)
18:08:15 Quit tchan (*.net *.split)
18:12:33 Join saratoga [0] (123e11e0@gateway/web/freenode/ip.
18:13:30saratogaits probably SRAM
18:13:40saratogalike the 320KB on AMS
18:13:41*whiskers75 hugs saratoga
18:14:16saratogai don't think it was possible to put on die (just on package) DRAM in the era when the fabs for these chips were made
18:15:05 Join prof_wolfff [0] (
18:15:21saratogahaha the mips people are actually excited about this chip:
18:15:23 Join wodz [0] (
18:15:27saratoga"look, someone cares about us still"
18:16:20 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
18:18:30 Quit uwe_ (Ping timeout: 246 seconds)
18:18:32 Quit kuldeepdhaka (Ping timeout: 260 seconds)
18:20:32 Join uwe_ [0] (
18:33:21wodzsaratoga: mips by volume is still impressive. It is doing much better then say power architecture. It is just that arm is doing better :-)
18:48:26 Nick jmspeex_ is now known as jmspeex (
18:51:03 Quit Strife89 (Ping timeout: 246 seconds)
18:54:55saratogabetter than power is like being better than coldfire :)
18:55:02saratoganot much of an achievement
19:41:19 Quit mc2739_ (Ping timeout: 252 seconds)
19:41:23toehser1 has 88Kb as a ROM number, not the RAM number... or am I misreading?
19:43:09 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
19:46:15wodzOn-chip RAM0~RAM5 (RAM0:8KB, RAM1~RAM5:16KB) is mapped to MCU program memory space, On-chip Multi-Use RAM1(20.25KB), RAM2(4KB), RAM3(6.75KB) and PCM RAM(8KB) that can be switched to be MCU data memory space or Audio codec Memory
19:48:42wodzthis gives 88kB program memory and 39kB as ram
19:49:23wodzI can't imagine who came up with such idea.
19:53:50copperhow is that even possible
19:53:57copperhow could they display album art?
19:54:07copperit must have more RAM somehow, somewhere
19:55:52 Quit akaWolf (Remote host closed the connection)
19:56:52 Join akaWolf [0] (~akaWolf@unaffiliated/akawolf)
19:57:27cramlerpamaury: In case you read this
19:57:29cramlerMy weird build can be found here:
19:57:35cramlerThere is a difference of 10256 bytes between our rockbox.creative files, but I dont know why...
20:06:17***Saving seen data "./dancer.seen"
20:18:49pamaurysaratoga: how do you know atj2127 has a dram controller ?
20:19:57 Quit cramler (Ping timeout: 246 seconds)
20:20:45 Join bertrik_ [0] (~quassel@rockbox/developer/bertrik)
20:21:05 Join cramler [0] (
20:21:24 Quit bertrik (Ping timeout: 245 seconds)
20:30:58pamaurycramler: indeed your build fails on my device
20:50:42 Quit ZincAlloy (Read error: No route to host)
20:51:19 Join ZincAlloy [0] (
20:51:30lebelliumpamaury: bump to get something to try on my X-Fi Style when you have some time :D
20:52:30pamaurylebellium: I'm about to commit some stuff for the X-Fi Style, the port is pretty complete now, with speaker support
20:52:41pamauryif you are here tonight I can upload files for you to try
20:52:41lebelliumreally? That's a good surprise :)
20:53:02lebelliumwill I have to format the device?
20:53:29lebelliumthere isn't the specific filesystem issue on this model?
20:53:47pamauryno, fortunately
20:53:48pamauryat the moment you have to manually enable the speaker in the options (ie it's not automatic when you don't have headphones)
20:54:40tertumaybe this isn't super relevant but maybe the archos devices could get "retired" status or something
20:55:18tertuso they'd keep being built if they still worked but if they broke they'd get dropped
20:56:06pamaurytertu: our plan is to declare the 3.13 release to be the last one, create a branch for them, backport a few things maybe
20:56:16pamauryin the main branch, we will likely drop support for them
20:56:25tertuthat seems reasonable
20:56:28whiskers75is rockbox getting anywhere?
20:56:35whiskers75it seems pretty dead
20:56:36tertuwhat do you mean
20:56:40tertuit's moving
20:56:51tertuit runs on devices that still get built
20:56:58whiskers75there hasn't been much activity
20:57:19lebelliumthat doesn't mean it's dead
20:57:30lebelliumsee how many new targets we have thanks to pamaury :)
20:57:35tertuto some extent
20:57:38whiskers75good point ;P
20:57:55wodzpamaury: you mean 3.14, right?
20:58:00lebelliumfor sure there are fewer active developers but pamaury is doing the job for 10 people lol
20:58:06pamaurywe are less and less active, that is true but mostly because there are less and less (interesting) targets to port to
20:58:18tertuthis is sort of a stupid question but
20:58:20whiskers75DAPs are dying
20:58:27tertui'm sad about that
20:58:33tertuDAPs are nice
20:58:54whiskers75I use my sansa less now because of spotify, rdio, google, etc
20:58:59tertuand i don't think enough people realize how nice they are
20:58:59pamaurycramler: it appears you build env might just work and the revision you chose is buggy :-/
21:00:06pamaurycramler: I will commit some stuff pretty soon, you can retry after that
21:00:10whiskers75tertu: i know :(
21:03:56cramlerpamaury: Probably you're right...
21:03:59cramlerI will nonetheless try on debian with higher gcc version (downloading debian right now, building gcc on ubuntu would take some hours...)
21:04:01cramlerThank you for your time :)
21:06:25wodzcramler: to be clear we support *only* the version of toolchain provided by script
21:06:41pamaurywodz: he uses the right version, everything is fine
21:07:05pamaurythat's why I was very confused, so I tried the same build with the same version and it fails too so I think that revision is just buggy
21:07:14wodzpamaury: I know, but now he says he want try newer version of gcc which is asking for troubles
21:09:32cramlerwodz: Understood that, thank you.
21:09:54pamauryah yeah indeed, especially since some recent gcc are known to fail
21:10:37pamaurycramler: since you are the X-Fi3 user and you probably used the OF, do you know if it can detect when the headphones are unplugged ?
21:13:00saratogathe page says external memory support, although maybe they mean FLASH
21:14:51wodzsaratoga: they list nand, mmc and spi nor flash
21:15:54*whiskers75 boots rockbox
21:16:03whiskers75Thank you rockbox devs. <3
21:16:08 Quit bertrik_ (Remote host closed the connection)
21:21:58 Join ikeboy [0] (
21:22:27cramlerpamaury: It can.
21:22:30cramler When playback is started with speakers enabled, and headphones are plugged in, it will switch to the headphones.
21:22:32cramlerWhen the headphones are unplugged, it will switch to the speaker again, but pause playback.
21:23:30pamauryhum ok, strangely I don't find any GPIO to detect this, I need to investigate this
21:24:04whiskers75THE GPIO ADVENTURES
21:24:50 Quit y4n (Quit: Today is the perfect day for a perfect day.)
21:32:01ikeboyHow can I edit an existing plugin on the fuze+ to always boost CPU?
21:32:41ikeboyIt's the textviewer, it's about 6 times as fast loading which would really help for bookmarks near the end of long books
21:33:47pamauryikeboy: in theory just add trigger_cpu_boost() and cancel_cpu_boost() somewhere
21:34:27ikeboyWould you be okay with boosting by default in the textviewer?
21:37:24pamauryboosting consumes a lot of power so probably not, clever boosting would be much better
21:37:44saratogaheadphone detect is often done by the dac, which then interrupts or at least raises a flag when state changes
21:38:45ikeboyHow about boosting right before seeking to a bookmark and unboosting after?
21:39:20saratogayou should boost when its needed and unboost as soon as possible
21:39:28saratogaif it improves performance in plugins its likely ok
21:40:01pamaurysaratoga: on most imx devices it's done a GPIO, the dac has some "short detect" method which could be used but it didn't work on all daps I tried, maybe that's how it work on the X-Fi3
21:40:11saratogaah ok
22:03:33 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
22:06:18***Saving seen data "./dancer.seen"
22:11:55fs-bluebotBuild Server message: New build round started. Revision cefaa46, 250 builds, 33 clients.
22:15:09fs-bluebotBuild Server message: Build round completed after 195 seconds.
22:16:06pamaurygevaerts: do I have to do something for the X-Fi Style to be built ? I add it to the buildserver/builds file in www/ but it wasn't built in the last round
22:17:36gevaertspamaury: no, but looking at your change, you didn't give those three builds unique identifiers...
22:17:44gevaertsi.e. the third field
22:18:01gevaertsI don't know what the effects of that will be
22:18:42pamauryoh damn, I knew I forgot something but couldn't remember what
22:19:29ikeboypamaury: can you test g#758 for me?
22:19:31fs-bluebotGerrit review #758 at : Boost CPU before loading operations in textviewer by Avi Eisenberg
22:19:43pamaurygevaerts: hum wait, my identifiers are unique I think
22:20:09gevaerts"cat builds |cut -f 3 -d :|sort|uniq -c|sort -n" says not
22:20:12pamaurygevaerts: ah no you're right
22:21:01pamauryfixed, I would have expected it to build at least one of them
22:21:26gevaertsMaybe it did
22:21:31*gevaerts also laughs
22:21:36gevaertsNot sure why though
22:21:52*whiskers75 has had many a trouble with building Rockbox.
22:22:02whiskers75Especially on Travis CI :p
22:22:26gevaertsYou mean you've hit many stupid travis ci limitations?
22:22:32gevaertsNot really our fault...
22:22:58whiskers75I know
22:23:04whiskers75I'm not blaming you.
22:23:12whiskers75I remember...
22:23:21whiskers75Build #215
22:23:27whiskers75and it was still broken
22:26:32pamaurysaratoga: does "headphone common short" and "l/r short" detect rings a bell ?
22:28:42saratogado the l/r short when you plug in a pair of headphones?
22:30:52pamauryI don't know, I'm trying to understand what the DAC provides, the doc says it can detect "common mode" short and "l/r short". I tried with common and got nothing, I'm going to try with l/r
22:31:05saratogano i guess they would not
22:31:12saratogathat sounds like some kind of fault detect maybe?
22:32:27pamaurythen I wonder how the OF detects the HP
22:32:59lebelliumthe Haut Parleur? :D
22:33:39pamauryHead Phones ;p
22:34:52pamaurymaybe it relies on the bluetooth chip to check that
22:36:04ikeboyAnyone want to test g#758?
22:36:05fs-bluebotGerrit review #758 at : Boost CPU before loading operations in textviewer by Avi Eisenberg
22:37:05pamauryikeboy: I'm quite busy but I'll try to test it tomorrow
22:39:10saratogathat should be safe i think
22:39:21saratogai assume boost just does nothing on application builds?
22:39:37lebelliumWhy do text files need a CPU boost? Oo
22:40:16ikeboyWhen you load a bookmark it takes about 6 times faster with a boost than without on fuze+
22:40:55ikeboyIt was annoying me that to continue reading I'd have to wait a minute to load
22:40:59lebelliumI wouldn't have thought that is so demanding
22:41:23ikeboy1 min vs 10 seconds is significant
22:42:47pamauryhum maybe it relies on the video dac instead
22:46:03pamauryindeed the video dac jack detect seems to work
22:48:59pamauryhowever I'm not sure the OF uses this
22:50:47 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
22:52:28 Quit fs-bluebot (Ping timeout: 260 seconds)
22:54:02 Quit bluebrother^ (Ping timeout: 272 seconds)
22:54:13 Join fs-bluebot [0] (
22:57:12cramlerpamaury: Can I help somehow?
22:59:20pamaurysaratoga: haha, apparently they do it using the adc
22:59:47pamaurythey wired a channel to measure <something> which is high when inserted and low when it's not
22:59:59pamaurycramler: I committed a few things, you can retry a build
23:01:47wodzpamaury: this is not that uncommon idea
23:04:06pamaurybut why a LRADC ? can't you just wire something a GPIO ? that's simpler no ?
23:07:44cramlerMaybe they mean to make it complicated? Nintendo uses to release new wii firmware regularly just to keep homebrew out.
23:11:46 Quit ikeboy (Quit: Leaving)
23:14:59 Quit cramler (Ping timeout: 265 seconds)
23:15:53lebelliumpamaury: I'll go to bed now. I'll try it on the X-Fi Style tomorrow if you had time to write the install instructions by then
23:17:34 Quit wodz (Quit: Leaving)
23:19:48 Quit lebellium (Quit: ChatZilla [Firefox 28.0/20140213172947])
23:24:55 Quit tertu (Disconnected by services)
23:24:55 Join ter2 [0] (~tertu@
23:40:45 Quit tweek_ (Read error: Operation timed out)
23:45:43 Quit ender` (Quit: I like long walks, especially when they are taken by people who annoy me. -- Noel Coward)
23:54:47 Quit Rower (Quit: Hmmm...)

Previous day | Next day