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).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2021-06-12

00:09:18 Join LambdaCalculus37 [0] (~robert@
00:11:20LambdaCalculus37Anyone considering removing some other stillborn/dead code bits from the master branch?
00:12:05LambdaCalculus37There's a stub of a port for the Tatung Elio TPJ-1022 in the codebase, but that port hasn't been worked on in ages and doesn't do very much at all.
00:12:15LambdaCalculus37Perhaps it can be removed?
00:15:08braewoodsLambdaCalculus37: you fixed the H300?
00:19:24LambdaCalculus37Yup, it's working now.
00:19:46braewoodsi've moved onto the gigabeats, another port that needs some polishing so it can move to stable
00:20:21braewoodsmostly support in rbutil, it has unique requirements this port
00:20:43LambdaCalculus37Way back when, I was attempting to get the Gigabeat T into the Rockbox family, since it's the same chipset as the Gigabeat S.
00:20:49braewoodsit's odd how the port is nearly complete but no one finished the last few bits
00:21:10braewoodsi saw that. but the gigabeat T is pretty obsolete now. not much point trying to finish that now.
00:21:18braewoodsand rarer than the gigabeat S.
00:21:34braewoodsi found exactly one for sale.
00:21:40LambdaCalculus37True. I do have one, but its battery is completely done for.
00:21:49braewoodsyou might be able to replace the battery.
00:22:00LambdaCalculus37Yeah, it's the same battery as the other Gigabeats.
00:22:01braewoodsi found a supplier of gigabeat S batteries.
00:22:19braewoodsnot quite, the gigabeat F and S have slightly different batteries
00:22:23braewoodsthe S requires thinner battery
00:22:33braewoodsi learned that from trying to mess around with it
00:23:08braewoodsthe S is also limited in terms of storage capacity but it's still a solid all around port
00:23:08LambdaCalculus37I haven't had a working Gigabeat in ages.
00:23:22braewoodsI do, i bought one from bluebrother and some others off ebay
00:23:30LambdaCalculus37I only have the iPod Video and the H320, and even there, those weren't my originals.
00:24:00braewoodsthere's several gigabeat S on ebay right now but all way overpriced
00:24:23LambdaCalculus37Even common players like the iPod Classic are stupid priced!
00:24:25braewoodsi got a few white ones at a lower price by waiting
00:24:42braewoodsi think i can use the two i got to get a better fixed up unit
00:24:57LambdaCalculus37And I would like to see how far along the iPod Classic port has come in the last 10 years!
00:24:59braewoodsone major problem with the gigabeat S
00:25:18braewoodsit is weird how the ribbon cable works so it's incompatible with CF adapters
00:25:22braewoodsthe common ones anyway
00:25:35braewoodsyou need an actual drive replacement
00:25:40braewoodsadapters don't work
00:25:50braewoodsi installed an 128GB SSD in one
00:25:52braewoodsthat does work well
00:26:37braewoodsbilgus was able to fix some old reliability bugs with the iriver H10 i sent him
00:26:51braewoodsit was causing boot hangs on my H10
00:27:08braewoodsapparently it was some kind of race condition from ages past that may have impacted other PP ports
00:27:13braewoodsbut mainly appeared on the H10
00:27:15braewoodsfor some reason
00:27:19LambdaCalculus37I think I have an H10 20GB here that has a dead battery.
00:27:32braewoodsi know what battery you can use for that one
00:27:40braewoodsit's the same one used in the H300 and H100 series
00:27:53braewoodst oreplace
00:28:16braewoodsbut the first time you do it you have to pry off the battery cover
00:28:27braewoodsit's held in by strong adhesive
00:28:41LambdaCalculus37Got tools for adhesive removal.
00:28:53braewoodsi used some electrical tape to help keep the battery in place after that
00:29:02braewoodsmake-shift double sided tape
00:29:46LambdaCalculus37I don't know if I would keep the H10 though.
00:29:49braewoodsit works but the cable is a bit long so you need to find out a way to make it sit well when you reassemble it
00:30:26braewoodsi do know the H10 works with CF mods but it doesn't run as fast as other ports do
00:30:57braewoodsthe write speed is pretty slow with the mod; like 5-6 MB/s
00:31:00LambdaCalculus37Been trying to lighten the load a little bit, plus I'm not an active developer anymore.
00:31:21LambdaCalculus37I only really had all the players to help do testing and such.
00:32:32LambdaCalculus37I still use Rockbox, and I might help out here and there still, but I don't think I'll be doing coding for the project again.
00:32:50LambdaCalculus37Nothing against the project... just kinda moved on, ya know?
00:33:46braewoodsi could always use more hardware; once i finish with this port i'm probably going to move onto USB drivers to expand the feature set available to USB software stacks
00:34:21braewoodsone thing i'd like to do is create a remote control thing over USB
00:34:24LambdaCalculus37Hey, if you'd like some of the players I have here, next time I do cleaning, I'll pull them out and set them aside for you.
00:34:26braewoodsnot quite sure how yet
00:34:37braewoodsLambdaCalculus37: let me know what you find
00:34:51braewoodsit would help with testing new drivers
00:35:04braewoodsi think the main area we can still innovate is USB
00:35:08braewoodsfor older ports anyway
00:35:25braewoodswe lack true networking so this is the next best thing
00:35:39LambdaCalculus37I can tell you what I do have right now: an H10 20GB that needs a battery and cables, a GoGear SA9200, a GoGear HDD1630, an iriver H140 that has no battery or HDD.
00:36:07braewoodsSA9200... those are pretty rare to find today.
00:36:20LambdaCalculus37I don't know where the GoGear cable is, though.
00:36:30braewoodsi have some already
00:36:38braewoodsplus i can buy gomadic if i have to
00:36:51LambdaCalculus37I also have a Sansa c240 that unfortunately had a swollen battery so I had to remove it. Can't find a replacement anywhere.
00:37:28braewoodsis that the one with a battery cover?
00:38:41braewoodsfound one
00:38:48LambdaCalculus37Ahh! Perfect!
00:39:06braewoodsiirc, the C200 series all use the same battery
00:39:39LambdaCalculus37They do.
00:40:00braewoodsdespite their limitations it does have one major advantage over the others
00:40:05braewoodseasily removeable battery
00:40:08braewoodsthe others are all soldered
00:40:34braewoodsi have to say the gigabeat S impressed me
00:40:46braewoodsit was designed with ease of repair in mind
00:40:56braewoodsno weird crap that is prone to cause damage during disassembly
00:43:25LambdaCalculus37That, and also the power it had under the hood. IIRC the Gigabeat S was the first device we supported that could just about do realtime playback of -c5000 APE files.
00:43:58LambdaCalculus37Plus good video frame rates in mpegplayre.
00:44:10braewoodskinda moot now with how codecs have advanced
00:44:29braewoodsbasically need hardware accelerations for playback to be possible on mobile
00:46:11LambdaCalculus37I take that back... just about do realtime playback? Nah, it *did* realtime playback!
00:47:23braewoodsdesowin has been working on finishing the Sansa Connect port. they finally found a way to inject a bootloader consistently.
00:48:07braewoodsif nothing else having a new port would help with options.
00:49:04braewoodsconsidering how long it's been since the gigabeat S was worked on... i'm surpised how little was broken by recent uphevel since 3.15
00:49:17braewoodsthe only show stopper was a stack overflow in a thread stack.
00:49:52LambdaCalculus37At least there's no more headache of HWCODEC targets breaking stuff.
00:50:19braewoodsnot even any LCD regressions
00:50:43braewoodsbilgus reworked the LCD code
00:51:07braewoodscaused a fair number of issues for awhile; thought it might be an issue here as well
00:51:33braewoodsright now just trying to work out how to integrate with rbutil
00:52:03braewoodsif i do implement a USB debugging driver...
00:52:18braewoodsi'll probably need to write a library for accessing it from hosts.
00:52:43braewoodsso it can be used by more than just a GUI application
00:53:05LambdaCalculus37One last before I do, since you mentioned rbutil and I was curious about something.
00:53:38LambdaCalculus37What's going to happen with the macOS and 32-bit Linux builds of rbutil? I noticed that there's no build for either of 1.4.1.
00:55:23braewoodsthe 32 bit build will probably be dropped... that's the trend for Linux lately. 64 bit is the most common variant of x86 now.
00:55:34braewoodsi can't speak for MacOS.
00:56:03braewoodsi've been considering opening a repository for rbutil packages
00:56:26braewoodsbuilding rbutil isn't that difficult, at least for Linux
00:57:09braewoodsOBS would even allow us to build for other architectures of Linux
00:57:38braewoodsimagine being able to use rbutil from an ARM64 port
00:57:43braewoodslike raspberry pi
00:57:48LambdaCalculus37That would be pretty cool.
01:00:23LambdaCalculus37But as for macOS, my only guess is unless all the necessary libraries are available on Apple Silicon, and considering Apple's development tools are getting more and more convoluted...
01:00:56braewoodsi can't justify spending $1k+ for something just to have a testing platform
01:01:03braewoodswindows is cheap and easily available
01:01:07braewoodsMacOS hardware not so much
01:01:08LambdaCalculus37Might end up being a case of "if you can build it, great... if not, ehh"
01:01:28LambdaCalculus37Anyway, time to bounce... later!
01:01:33 Quit LambdaCalculus37 (Quit: Leaving)
01:29:36***Saving seen data "./dancer.seen"
02:27:46desowinbraewoods: no need for the capture more, found my e280
02:28:12desowinthe difference between the two is that Sansa Connect sends the ZLP twice, and the second ZLP results in Windows resetting the device
02:28:20 Quit TorC (Quit: Vanishing into the mist)
02:29:08desowinI just have to figure out what's going on with this ZLP, but considering Linux kernel had the bit define as CB_ZLP_GARBAGE, I guess there's some shenanigans with it
02:29:09 Join TorC [0] (~Tor@user/torc)
02:30:45braewoodsdesowin: huh.
02:37:12braewoods g#3476
02:37:14rb-bluebotGerrit review #3476 at : mknkboot/beastpatcher: implement basic firmware validation by James Buren
02:37:40braewoodswhoever cares to review please do so. first major improvement; validating the firmware for dual boot operation.
03:26:16 Join p9 [0] (~p9@
03:29:38***Saving seen data "./dancer.seen"
04:26:42 Quit TorC (Ping timeout: 264 seconds)
04:56:47 Join TorC [0] (~Tor@user/torc)
05:26:46 Quit p9 (Ping timeout: 244 seconds)
05:29:42***Saving seen data "./dancer.seen"
06:01:51 Join Guest58 [0] (~Guest58@
06:46:49 Quit Guest58 (Quit: Client closed)
06:53:32rb-bluebotBuild Server message: New build round started. Revision a90ef8195b, 297 builds, 9 clients.
06:56:38 Join p9 [0] (~p9@
07:02:59rb-bluebotBuild Server message: Build round completed after 567 seconds.
07:03:02rb-bluebotBuild Server message: Revision a90ef8195b result: All green
07:29:46***Saving seen data "./dancer.seen"
08:25:08 Join dconrad [0] (~dconrad@
08:28:38 Quit p9 (Ping timeout: 244 seconds)
09:18:14 Quit TorC (*.net *.split)
09:29:50***Saving seen data "./dancer.seen"
09:36:56 Join TorC [0] (~Tor@user/torc)
09:54:56rb-bluebotBuild Server message: New build round started. Revision 663d846cf3, 297 builds, 9 clients.
09:55:39desowinhopefully this solves it, I tried messing around with the CPPI but well, without datasheet it is just a guesswork
10:04:11rb-bluebotBuild Server message: Build round completed after 555 seconds.
10:04:15rb-bluebotBuild Server message: Revision 663d846cf3 result: All green
10:10:26 Quit dconrad (Remote host closed the connection)
10:20:36 Join dconrad [0] (~dconrad@
10:24:53 Quit dconrad (Ping timeout: 244 seconds)
10:32:36 Join p9 [0] (~p9@
10:36:34 Quit p9 (Client Quit)
10:47:17rb-bluebotBuild Server message: New build round started. Revision f3f9d1fb95, 297 builds, 9 clients.
10:56:22rb-bluebotBuild Server message: Build round completed after 545 seconds.
10:56:25rb-bluebotBuild Server message: Revision f3f9d1fb95 result: All green
11:29:54***Saving seen data "./dancer.seen"
11:48:04 Quit pablocastellanos (Ping timeout: 244 seconds)
11:54:31 Join pablocastellanos [0] (~pidgin@user/pablocastellanos)
12:03:51 Join TheLemonMan [0] (~lemonboy@irssi/staff/TheLemonMan)
13:29:56***Saving seen data "./dancer.seen"
14:50:07 Quit TheLemonMan (Quit: "It's now safe to turn off your computer.")
15:17:12 Quit ddevault (Quit: taking a break)
15:26:00 Join amachronic [0] (~amachroni@user/amachronic)
15:30:00***Saving seen data "./dancer.seen"
15:31:45 Quit skipwich (Ping timeout: 252 seconds)
15:33:42braewoodsdesowin: what chip is it?
15:34:19braewoodsi sometimes can find stuff other people don't
15:54:18 Quit amachronic (Ping timeout: 252 seconds)
16:06:26 Join amachronic [0] (~amachroni@user/amachronic)
16:46:46amachronicso, is anyone opposed to g#3474?
16:46:48rb-bluebotGerrit review #3474 at : Enable float formatting in printf by Aidan MacDonald
16:48:03 Join dconrad [0] (~dconrad@
16:56:42braewoodsamachronic: is it disabled for bootloaders?
16:57:15braewoodsthat extra space is a luxury we can't afford for bootloaders.
16:57:37braewoodsthat would probably overload the printf used in the H120 bootloader
16:57:48braewoodsit only has 64k to work with
16:57:56braewoodsand was like 58k when i cut the last release
16:58:29braewoodsthat's all i can say about it
16:58:42braewoodsthese extra luxuries should not be included in bootloader builds
16:58:49braewoodsthey don't need floating point
16:59:23amachronicgood thinking
17:04:10amachronicit seems that block was already #ifndef BOOTLOADER so it's fine.
17:05:48dconradamachronic, I got g#3461 updated if you want to peek at it
17:05:49rb-bluebotGerrit review #3461 at : FS #13297: M3K Autolock allows one action before disabling touchpad by Dana Conrad
17:06:46dconradI didn't even think about using the local action instead of cur, just automatically used it since I was already doing cur->button haha
17:07:52dconradI tested it here, seems to work great
17:09:05amachronicseems good
17:09:47amachronicjust to clarify is the manual change just something forgotten from the last softlock thing you did?
17:09:55 Join rockboxman [0] (
17:10:14dconradin g#3436
17:10:16rb-bluebotGerrit review #3436 at : Softlock Improvements by Dana Conrad
17:11:05dconradit was added in g#3285
17:11:07rb-bluebotGerrit review #3285 at : Add ability to always have autolock on by Dana Conrad
17:11:10amachronicoh yeah it's in the commit message, didn't look
17:11:12rockboxmanDidn't know you could still get a Sansa connect brand new.
17:11:30rb-bluebotBuild Server message: New build round started. Revision c067b344e8, 297 builds, 9 clients.
17:11:47 Part rockboxman
17:20:08rb-bluebotBuild Server message: Build round completed after 517 seconds.
17:20:10rb-bluebotBuild Server message: Revision c067b344e8 result: All green
17:30:01***Saving seen data "./dancer.seen"
18:37:08 Quit dconrad (Remote host closed the connection)
18:48:22 Join dconrad [0] (~dconrad@
18:50:12 Join speachy [0] (~speachy@
18:50:12Mode"#rockbox +v speachy" by ChanServ (
18:53:23 Quit amachronic (Read error: Connection reset by peer)
18:53:25 Quit dconrad (Ping timeout: 272 seconds)
19:08:10speachybeen digging into FS #13299 −− it's a bit of a head scratcher.
19:08:11rb-bluebot Cannot play some music file. Ipod classic 120gb Rockbox. One of the music attached. (bugs, unconfirmed)
19:09:16speachythe interesting thing about this file is that it's not only variable bitrate but also variable sample rate
19:17:10speachybasically the "two consecutive headers with similar enough headers" test is failing, so we're never considering it a valid file.
19:30:04***Saving seen data "./dancer.seen"
19:49:20braewoodsspeachy: variable sample rate? i thought audio files had a fixed sample rate that was used to determine how long it is.
20:20:02speachypretty sure that's not what's causing the file to fail, but it is a curiousity
20:20:29speachysample rate is a property of a given chunk/frame, not the file as a whole.
20:20:49speachy(ie just like the bitrate, which can and does routniely vary from chunk to chunk)
20:25:26speachyI've never seen a variable sample rate file but it is technically legal. and handled just fine by $random_pc_software
20:26:16speachythere's something about this file that's breaking the header parsing; what happens is that we find the second header, then back up to find the previous one.
20:26:31speachyit's the backing-up that's going wrong
20:30:15speachy...because the frames have different sampling rates, they have different sizes.
20:34:22 Join dconrad [0] (~dconrad@
20:48:32 Join Saratoga [0] (
20:48:48SaratogaI don't think libmad even supports vbr for later 1 files
20:50:01SaratogaLayer 1 is usually cbr only
20:50:39SaratogaYou can change the frame sizes but decoders tend not to allow that
20:50:51 Quit Saratoga (Client Quit)
20:51:49speachywhatever encoded this added padding between the frames.
20:52:10speachyour metadata parser expects none.
20:54:32speachyour "find frame start" code seeks byte by byte, so we find each frame properly, but the "read the next frame to see if it's similar" code is expects no padding.
21:11:55speachyI'd appreciate a sanity check on g#3478
21:11:56rb-bluebotGerrit review #3478 at : mp3: Greatly simplify VBR frame parsing in the metadata decoder. by Solomon Peachy
21:12:25speachyhandles every file I've thrown at it so far, including that iffy one
21:12:37speachy(which won't play back, but at least we can parse out its metadata now)
21:14:03speachybecause it seems to be win-win, removed code and run-time complexity, and fixes a bug in the process.
21:30:08***Saving seen data "./dancer.seen"
21:45:55speachydesowin: have you been updating wiki's port status pages to reflect your Connect work? And is it something we can promote to "Unstable" status, generating nightly builds etc?
21:54:37mendel_munkisI just discovered that new gerrit needs JS :(
21:57:09speachy(I thought the old one did too?)
21:58:38mendel_munkisI think it just needed js for all the features, but I don't remember for sure
22:00:52mendel_munkis3478 looks sane to me
22:13:30 Join ufdm_ [0] (
22:13:30 Quit ufdm (Read error: Connection reset by peer)
22:14:52speachyjust made another couple of tweaks, stripped out a little more dead code
23:18:15 Quit dconrad (Remote host closed the connection)
23:27:51Bilgusso it looks like to me it originally went H0H1H2H3 H0:H1 H0:H2 H0:H3 and now it goes H0:H1 H1:H2?
23:28:05Bilgusre 3478
23:28:24 Join dconrad [0] (~dconrad@
23:28:26Bilgussorry H2:H3 was the last iter
23:30:09***Saving seen data "./dancer.seen"
23:33:00 Quit dconrad (Ping timeout: 252 seconds)
23:34:41speachyBilgus: No, it did Hn:Hn+1
23:35:21speachy(ie it would seek forward from the current position to try and peek at the next header, before seeking back)
23:35:46speachythe new code does Hn:Hn-1
23:36:34speachy...which... is wrong now that I explained it in that way. It should return Hn-1, not Hn.
23:38:48Bilgusbut it used to seek back to the last header?
23:39:06Bilgusmaybe i'm misreading it quite possible
23:39:09speachyit would seek "back" to the current header
23:39:48speachyold code returned the first of a valid pair, new code retursn the second of the pair.
23:40:14speachyI need to change it to return the first, and seek back to where that header lived.
23:42:49Bilgusbut the old code how did it keep track of position as it seeked back to the last known header
23:43:13speachyit knew how forward it seeked, so it just reversed it.
23:44:08speachyok, that's fixed.
23:45:09Bilgusno i'm saying in the case of invalid header wouldn't it get stuck looking at the same two bad headers till pos maxed out the loop
23:53:11speachyold or new?
23:53:39speachyI mean, if it doesn't find anything it considers valid it'll eventually max out pos and exit the loop, sure.
23:54:15speachybut the old code didn't rewind to the *start* of the header, it rewinded to just after the header.

Previous day | Next day