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

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

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

#rockbox log for 2021-04-05

00:00:52braewoodswhy would anyone want to send so much automated traffic against
00:00:56braewoodswe have nothing really
00:04:57***Saving seen data "./dancer.seen"
00:09:58_bilgusits probably that ransom message someone forwarded
00:10:46_bilguseh n bitcoins or we will kill your firstborns in the night or wait was that it
00:11:29_bilgussomething along those lines and we wondered what they thought we had to give em
00:12:08_bilgus3 rubber bands and pocket lint better get macguiver to get something from that
00:34:24 Join beencubed [0] (~beencubed@
01:05:19 Quit ac_laptop (Ping timeout: 260 seconds)
01:20:02 Quit massiveH (Quit: Leaving)
02:04:58***Saving seen data "./dancer.seen"
03:01:30 Quit ufdm (Remote host closed the connection)
03:01:49 Join ufdm [0] (
03:23:56 Quit pixelma (Quit: .)
03:23:56 Quit amiconn (Quit: - Chat comfortably. Anywhere.)
03:24:55 Join amiconn [0] (jens@rockbox/developer/amiconn)
03:24:56 Join pixelma [0] (marianne@rockbox/staff/pixelma)
04:04:59***Saving seen data "./dancer.seen"
04:41:20 Quit ZincAlloy (Quit: Leaving.)
04:53:31 Join PimpiN8 [0] (
04:55:05 Quit PimpiN8 (Read error: Connection reset by peer)
05:08:15 Quit Stanley00 (Remote host closed the connection)
05:20:16 Join cela [0] (
05:21:01cela_bilgus tested e09df1ce5bM-210405 posted on the forum still has the same playback issue
05:21:56celaWodz might still drop by, he created the RK27xx port not sure if he still has a working device though
05:35:50celaJust retested the last commit that worked commit  [83e8e35a5888e32b131d628b7516d0de1d73d760]  , playback works fine
05:36:22 Quit cela (Quit: Connection closed)
05:39:22 Join MrZeus [0] (~MrZeus@2a02:c7f:a0aa:4400:7949:2774:350e:9e8b)
06:05:03***Saving seen data "./dancer.seen"
06:07:52 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
06:08:42 Join lebellium [0] (
06:14:48 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
06:15:56speachybraewoods: it's a host on the internet. that's all that's needed.
06:25:58 Quit Stanley00 ()
06:30:07 Quit pixelma (Ping timeout: 260 seconds)
06:30:34 Quit amiconn (Ping timeout: 276 seconds)
06:30:38 Join pixelma [0] (marianne@rockbox/staff/pixelma)
06:30:39 Join amiconn_ [0] (jens@rockbox/developer/amiconn)
06:30:39 Nick amiconn_ is now known as amiconn (jens@rockbox/developer/amiconn)
07:09:34fs-bluebotBuild Server message: New build round started. Revision 1aed109fa8, 298 builds, 10 clients.
07:30:04fs-bluebotBuild Server message: Build round completed after 1230 seconds.
07:30:08fs-bluebotBuild Server message: Revision 1aed109fa8 result: 1 errors 0 warnings
08:05:05***Saving seen data "./dancer.seen"
08:28:43speachycela, wait, current build worked? that really sounds like some sort of odd alignment quirk. But at least you narrowed it down to the codecs
08:37:32 Join ZincAlloy [0] (
08:53:40 Join cela [0] (
08:54:51celaspeachy no I just retested the last commit that worked back in 2017 :-)
08:55:06speachyaaah, okay, that was unclear
08:55:36speachybut at least straight-up PCM works. can you try other codecs or plugins that make sound?
08:55:50celawodz is the man to get hold of, i think he does drop by from time to time.
08:56:04speachy(eg opus or flac, and the module/tracker plugin..)
08:56:20speachyhaven't seen him here for some time, alas
08:56:35speachyproblem is that nobody here actually has any rk27xx hardware. :/
08:57:41_bilguscela did you compile that test build on the current toolchain by chance?
08:58:23celaflac is the same, button sounds work. if all else fails it could be retired like Archos.
08:58:42speachydo plugins work at all?
08:58:49celatoolchain not sure i'd have to check
08:59:06celano plugins are enabled
08:59:57_bilgusok let me give you this working build compiler here and see if it still works for you
09:00:22celaport was part complete, no usb transfer, charging only.
09:03:02speachythere are at least 8 rk27xx targets in the tree, including 3 that I ported over from the xvortex sources
09:04:39speachywe only provide dev/nightly builds for those three
09:04:57speachywait, and the two hifimans too
09:05:18celawow that many, hifiman would be best to test with, mine is just a no brand and they are not very consistent hardware wise.
09:05:50speachyconsistent or not, codec functionality is pure sw and would presumably be the same on all of 'em.
09:06:21celaso they could all be broken
09:06:46speachywell, can use some of the rockbox slush fund to pick up an old HM-601
09:07:55_bilgusthats that working commit compiled by me
09:09:16speachyoh, incidently, the xduoo x2 is an atj213x device. don't know about the new x2s yet.
09:17:12speachywhoops, make that AK2117C −− uses the same firmware image format as the ATJ213x
09:17:35speachytihis thig has an 8051-derived core.
09:18:33speachyinsane on-chip memory segmentation, and hw accelerators.
09:19:28celabilgus  83e8e35a5_rk27xx_rockbox-full you complied works fine
09:19:52celago to go
09:20:03cela*got to go
09:20:07speachythis makes the rknano family look like a supercomputer in comparison. sheesh.
09:20:18 Quit cela (Quit: Connection closed)
09:21:44_bilguscela thank you thats enough for me to go on
09:22:06 Quit ZincAlloy (Quit: Leaving.)
09:31:01speachy_bilgus: think it's worth getting that hifiman 601?
09:32:05_bilguswhat was the $$ on that one?
09:32:24speachy$91 shipped.
09:32:34_bilgusyeah thats not terrible
09:32:47speachythe DX50/DX90 seems to go for $150 or so
09:32:58_bilgusthats a bit steep for used
09:33:57speachyok, should be here by next week they claim.
09:44:00braewoodsit might be an opportunity to create specific ports out of the generic one
09:44:53_bilgusI wish they listed the embedded processor in the ad copy
09:45:07 Join ZincAlloy [0] (
09:53:14speachyeh, largely moot these days unless someone finds a shipping container packed with a bunch
09:57:04fs-bluebotBuild Server message: New build round started. Revision 74ae18cc8a, 298 builds, 10 clients.
10:05:09***Saving seen data "./dancer.seen"
10:11:46fs-bluebotBuild Server message: Build round completed after 882 seconds.
10:11:50fs-bluebotBuild Server message: Revision 74ae18cc8a result: All green
10:13:34_bilgushow many devices do we have with remotes like 2?
10:14:26speachyat least 3
10:14:34_bilgusmy key remap idea isn't going to work with remotes in its current form
10:15:48_bilguseh I guess its only going to be needed for the remotes but plugins I'll leave as is since there is already a way to supply button mapping for them
10:16:23_bilgusI do want to make the keymap plugin able to make them per plugin as well
10:17:14_bilgusso you could have doom_kmf and it would load when the plugin is opened
10:18:04_bilgusand since we can now call plugins from plugins the menu can be integrated in the plugin if the keymapper plugin exists
10:34:07_bilgusspeachy <this is the diff between the previous working and the enum broken commit ln 954 looks interesting
10:35:51speachyline 954 of the actual diff, or 954 in the src/dest dump?
10:36:37_bilgusthe diff line
10:36:41_bilgusnot source
10:37:59speachy'mov r0,r0,asl#24' being the difference?
10:38:12_bilgusasr is shift right
10:38:53speachyshift left, shift right
10:40:57_bilgusstill not through the whole thing but so far beyond alignment that the only real diff
10:41:28_bilgusI'm spiining up the aligned one I sent yesterday to cut down on some of the noise
10:42:58speachyso it shifts back and forth. effectively zeroing out the upper 24 bits
10:43:15_bilgusin the working one
10:43:30speachyoh, thought it was the other way around
10:43:59speachyfor an enum the max value is supposedly known
10:44:28_bilgusmmhmm so itd make sense to wipe out the upper
10:46:29_bilguseh the aligned one is just different in other places i'll finish out this one
10:52:07speachyexcept ACTION_HALT is -1. :D
10:54:09speachywe have like 4 defined values only
11:16:40_bilgusok so what I don't see is how this would only affect this target
11:17:36speachyyeah, it's hardly our only armv5 platform
11:19:13_bilgusline 3919 wehn i goes into the codec entry
11:19:26_bilgusopus is the first one but it hasa similar change
11:19:32_bilgusfor others
11:19:57_bilguscmn not sure what that one is
11:20:28_bilguscompare negative
11:24:17_bilgusmaybe not I only see opus speex and am3? whatever that is
11:24:34_bilgusdid cela say some codec worked?
11:26:05speachyhe tried flac too
11:26:10speachyso I presume mp3 + flac
11:27:31_bilgusyeah and I don't see either in here
11:27:48_bilgusopus aac wma wma pro mpa mpc
11:28:19_bilgusso I assume the codecs that are getting compiled here must have a different way of enry than the others
11:30:02_bilgusor mp3 and flac are broken
11:30:39_bilgusnope found flac
11:30:57_bilgusno mp3 though maybe its a different name
11:31:52speachyUSSC ruled in favor of Google over Oracle.
11:40:55 Quit pamaury (Ping timeout: 260 seconds)
11:53:28_bilgusso the codec enum is still an int
11:54:30_bilgusdoes that mean it should be cast to long in all these places by the lack of warning I'd say no? but maybe we can put a -1L or something
12:00:43 Quit ZincAlloy (Quit: Leaving.)
12:04:42_bilgusthe long min and long max take care of that I guess
12:05:10***Saving seen data "./dancer.seen"
12:16:58_bilgusIts been about a week since we dropped thiose fixes for the ipod I guess we can't say we fixed anything but I guess its not broken worse either
12:17:16braewoods_bilgus: enums will expand to long if any of the constants are long
12:17:18speachynobody's complained yet..
12:17:26braewoodsi learned that some time back
12:17:51braewoodsso in theory
12:17:59braewoodsif the first one is initialized as 0L
12:18:05braewoodsthe rest should be long as well
12:18:44braewoodshere's some random C trivia...
12:19:01_bilgusYEah it sets them all to the biggest alignment
12:19:06braewoodsgiven a function call, can you determine what order the arguments will be evaluated in?
12:19:39_bilgusthats up to the compiler IIRC
12:19:45braewoodsyea, it's implementation defined
12:19:50braewoodsso the answer is no
12:20:02braewoodsit's why you need to be careful with side-effects in this context
12:20:38braewoodsyou can use them but you shouldn't use a set of side-effects where the order influences the end result
12:21:16braewoodsi usually don't use side-effects in function call arguments
12:21:21braewoodsseems bad practice
12:21:36braewoodscan introduce bugs if you're not thinking carefully
12:21:47_bilgusa lot of the low level is like tha though and the compiler is generally pretty good at abstracting thoe details away but yes you are probably highlighting the reason only problem is without a device waiting 10 minutes - 1 day between builds gets annoying fast
12:21:50braewoodsi'd rather avoid tricky parts of C honestly since they tend to introduce bugs
12:22:48_bilgusas much as I hate to add any limits I think maybe we need to add a NO ASM clause going forward unless really compelling
12:23:11_bilgusand then that there be left in the original c code with a NO_ASM switc or something
12:23:11braewoodsyou mean asm for new code?
12:23:47braewoodsthe only time ASM seems appropriate is when you absolutely need stuff to happen in a specific way at that level
12:23:56braewoodsmaybe bootloader code or so
12:23:56_bilgusyeah the codecs are always going to be needed
12:24:10_bilgusASM in them just because they are way optimized
12:24:54braewoodsi generally avoid ASM since it hinders the ability to benefit from compiler advances
12:25:01braewoodsbut bootstraping ASM is hardly performance critical
12:25:42braewoodsyou usually only run it once per boot or so
12:25:48braewoodsso reliability is more important
12:26:12_bilgusjust having the path easily back to original c code would be fine then we could disable them all and start reenabling them as testing proves them
12:26:58_bilgusbut then that also add the you need to update 'dead' c code
12:27:30_bilgusstill better than searching the SVN and uopdating dead c code
12:27:59_bilgusreally dead then because it wasn't even near the ASM the last time someone touched it
12:33:39_bilgusit'll probably end up coming down to the memory layout is slightly different hiding a stackovf or something
12:33:54_bilgusor unhiding it
12:34:07braewoodswe could start by removing code specific to ports that are half-baked that no one has contributed to in ages
12:34:35_bilgusthey are kinda self contained though don't really bleed over into the others
12:35:18_bilgusnot that it matters we can just put them in their own branch and excise them
12:35:42_bilguseasy enough to bring em back in if someone wants to take them up
12:36:57_bilgusfor now..
12:37:31_bilgusok I've come up with a way to allow remotes and plugins to be remapped from the core rempa file
12:38:00_bilgusthat should allow you to override everything yet still load custom keymaps in plugins
12:39:04speachyyeah, there's very little target-specific stuff outside of its specific subdir. just the occasional special-case here or there.
12:39:33speachy(unlike the hwcodec & archos stuff.. which had tentacles everywhere, due to history)
12:50:37_bilgusI'm so glad thats gone
12:50:45_bilgusonly took 10 years
12:53:12_bilgusI'm thinking rather than oring a shifted context with the action I'll just have it search for the context as a 3 int entry and then consider everything till a stop sentinel valid for that context
12:53:28_bilgusbasically how it does it atm sorta
12:55:02_bilgushmm that will make the plugin for generating them slightly more complicated
13:04:21 Join ac_laptop [0] (~ac_laptop@
13:04:22 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:5416:e4ab:d32e:1fe8)
13:17:44 Quit ac_laptop (Ping timeout: 246 seconds)
13:26:26 Join ac_laptop [0] (~ac_laptop@
13:31:14 Nick dweeber` is now known as dweeber (
13:33:03 Quit speachy (Ping timeout: 268 seconds)
13:34:58 Quit emacsomancer (Ping timeout: 240 seconds)
13:37:27 Join speachy [0] (~speachy@
13:39:30 Join emacsomancer [0] (
13:41:28 Quit ZincAlloy (Quit: Leaving.)
13:47:41 Quit emacsomancer (Ping timeout: 240 seconds)
13:51:43 Join emacsomancer [0] (
13:56:55 Quit yosafbridge (Quit: Leaving)
14:05:12***Saving seen data "./dancer.seen"
14:06:57 Quit paulk-leonov (Read error: Connection reset by peer)
14:08:38 Join yosafbridge [0] (
14:15:37 Join paulk-leonov [0] (
14:16:08 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
14:32:57 Join ZincAlloy [0] (
15:12:48 Join PimpiN8 [0] (~PimpiN8@2001:1c04:3308:a500:686f:b150:a5a4:9c9)
16:05:15***Saving seen data "./dancer.seen"
16:31:24 Join dweeber_ [0] (
16:33:47 Quit dweeber (Ping timeout: 240 seconds)
16:36:19 Quit pamaury (Ping timeout: 260 seconds)
16:37:26 Quit ZincAlloy (Quit: Leaving.)
16:38:20 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
16:47:47 Quit Moarc (Read error: Connection reset by peer)
16:48:08 Join Moarc [0] (
16:55:49 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:e928:ee23:96e9:b094)
16:55:58braewoodsi wonder if this will ever launch
16:57:07braewoodsit won't be price competitive with pinebook pro but eh
16:57:12braewoodsmight be interesting none the less
16:57:16 Nick dweeber_ is now known as dweeber (
17:21:59 Join KalBot [0] (
17:23:19 Quit ZincAlloy (Quit: Leaving.)
18:05:17***Saving seen data "./dancer.seen"
18:06:50 Quit lebellium (Quit: Leaving)
18:20:10 Quit PimpiN8 (Quit: My MacBook has gone to sleep. ZZZzzz…)
19:04:32 Join tomato1 [0] (t0mato@gateway/vpn/mullvad/tomato)
19:06:39 Quit tomato (Ping timeout: 268 seconds)
19:06:40 Nick tomato1 is now known as tomato (t0mato@gateway/vpn/mullvad/tomato)
19:11:07 Quit pamaury (Ping timeout: 240 seconds)
19:34:33 Quit ac_laptop (Quit: WeeChat 3.0)
19:34:53 Join ac_laptop [0] (~ac_laptop@
19:44:47 Join dconrad [0] (d026e411@
19:49:59_bilgus#3274 ok so I've reworked this a bit I think it should be a pretty efficient way to do this basically the beginning consists of a LUT with contexts the entries carry an index into the array of the first remapped action for that context with a sentinel at the end
19:51:00_bilgusa caveat is that you can't remap fall through contexts
19:51:18_bilgusso no system wide remaps I suppose
19:52:06_bilgusit could be changed by running through the table again after but I don't really see the need for the extra overhead
19:58:08 Quit dconrad (Quit: Ping timeout (120 seconds))
20:05:20***Saving seen data "./dancer.seen"
20:21:33 Quit MrZeus (Ping timeout: 246 seconds)
20:43:05 Quit akaWolf (Ping timeout: 252 seconds)
21:24:07 Join dconrad [0] (d026e411@
21:40:32 Part dconrad
21:41:34 Join dconrad [0] (~dconrad@
21:51:59 Quit dconrad (Quit: Leaving)
21:53:13 Join dconrad [0] (~dconrad@
21:58:35 Quit dconrad (Quit: Leaving)
22:01:10 Join dconrad [0] (~dconrad@
22:05:21***Saving seen data "./dancer.seen"
22:18:23 Quit ac_laptop (Ping timeout: 265 seconds)
22:41:11 Join f1reflyylmao [0] (~f1refly@2a01:c22:9099:9f00:f73a:aed3:9302:975e)
22:42:15 Quit f1refly (Ping timeout: 250 seconds)
22:42:15 Nick f1reflyylmao is now known as f1refly (~f1refly@2a01:c22:9099:9f00:f73a:aed3:9302:975e)
23:10:17 Join ac_laptop [0] (~ac_laptop@
23:16:36braewoodshow annoying.
23:16:49braewoodsi can't disassemble the gigabeat S because of a stripped external screw.
23:57:29 Quit JanC (Remote host closed the connection)
23:57:53 Join JanC [0] (~janc@lugwv/member/JanC)

Previous day | Next day