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 2020-05-23

00:09:04 Join beencubed [0] (~beencubed@
01:42:22 Join dys [0] (
01:51:19***Saving seen data "./dancer.seen"
02:02:12 Join S|h|a|w|n [0] (~shawn156@unaffiliated/shawn156)
03:30:10 Join lebellium [0] (
03:36:31 Quit dys (Ping timeout: 246 seconds)
03:51:21***Saving seen data "./dancer.seen"
04:21:20 Quit S|h|a|w|n (Quit: Leaving)
04:52:05 Join MrZeus [0] (
05:08:08 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
05:32:23 Quit livvy (Ping timeout: 240 seconds)
05:51:25***Saving seen data "./dancer.seen"
07:09:15 Join livvy [0] (~livvy@gateway/tor-sasl/livvy)
07:51:28***No seen item changed, no save performed.
08:14:47speachyforums, translate, and themsite all https enabled properly now
08:17:29speachythemes (and download) are the only sites not forced to be https-only
08:17:44speachyuntil rbutil can handle them sanely
08:19:21pamauryspeachy: hey, I didn't answer on the mailing but I want to say thank you for the all the hard work you are doing and all the transition :)
08:20:46speachythis isn't entirely altrusitic, but you're welcome. :)
08:21:32speachythe eventual motivating goal is to remove impediments to contributing.
08:22:51speachyputting aside the fact that reverse-engineering/hacking baremetal real-ish-time hardware+software can ever not have a daunting learing curve..
08:25:01BilgusI hope we get an ipod contributer It might just be that they are popular but I'm thinking they are just a support nightmare
08:25:21speachythey practically defined the entire market
08:25:49speachyIs apple still selling "classic" ipods?
08:25:56BilgusF NO
08:26:55speachyIME supporting anything-Apple is a nightmare.
08:27:23Bilguslast I saw they wanted to sell this touch screen almost a phone or maybe it was a phone monstrosity
08:27:45Bilgusbut thats been years ago
08:28:00speachythe ipod touch is/was basically an iphone minus the cellular modem.
08:28:11 Join ac_laptop [0] (~ac_laptop@
08:28:41Bilgusyeah thats the last one I saw and it did not give me warm fuzzies lol
08:28:54speachyhave one of them floating around here; inerited it from $job-1 where it was used for Bluetooth LE interop testing.
08:28:57livvyI love my classic - shame they dont make them like they used to
08:31:30speachybut yeah, would be really nice to have someone here that knows the ipod fairly well.
08:33:02BilgusYeah they are still selling the touch
08:33:27speachy...well enough to help with that PP data abort that's holding up the gcc494 move.
08:33:41livvyspeachy: ive been trying to learn in earnest but I found finding the right documentation to be a challenge
08:34:32BilgusNot that any other company does either most everything here was reverse engineered
08:34:48BilgusI should note 'By people smarter than I'
08:35:02speachythere aren't terribly many "onboarding" tasks with rockbox. A goodly number of the issues are hardware-specific bugs, and the rest.. well, the codebase is well-written but there's a _lot_ of it.
08:35:46BilgusI can't wait to get rid of HWCodec and knock out 15%
08:36:22 Quit koniu (Remote host closed the connection)
08:36:46livvyBilgus: heh. i'm out of my depth with my current knowledge (didn't focus at such a low level).
08:36:49 Join koniu [0] (~koniu@gateway/tor-sasl/koniu)
08:36:54speachypersonally I want to rip hwcodec out after the toolchain uprev lands.
08:37:58speachythe recorderv1 "retirement" has been a good dry-run so far
08:38:12livvydoes anyone have any books/resources/suggestions for reverse engineering hardware? hah
08:39:31speachyhardware RE is generally mechanical in nature, trace out the PCB etc. that's necessary, but one can't do that to the chips themselves.
08:39:53speachyThis is a decent book for SW:
08:41:29speachyto be effective you have to know the basic CPU architechure reasonably well (especially with respect to memory layouts), how structures are constructed in memory..
08:44:38fs-bluebot_Build Server message: New build round started. Revision 084aa9c, 292 builds, 10 clients.
08:45:04livvyTY. I can use a pen, paper and my brain reasonably well, just needed to motivation to learn the technicals since I'm programming at high level in my job. Thanks for the resource and pointers though speachy
08:46:10livvyI'll see what I can learn before the end of the year :)
08:47:16BilgusI'm going to get these AMS clock patches out of my tree I kinda forgot about them
08:47:34speachylivvy, if you aren't already familar with the rudiments of microcontrollers, that's a good place to start −− pick up an arduino or something along those lines and do something "useful" with it.
08:48:31Bilgusor esp8266
08:48:34speachyand step outside of the IDEs to see what's really going on −− things like "write a for loop, see what it looks like in asm" sort of thing.
08:48:59speachyIMO the ARM Cortex-M are the most pleasant micros to work with
08:49:38speachyas an added bonus you can drive them with pure C using FreeSoftware from end-to-end.
08:50:40livvy> and step outside of the IDEs to see what's really going on −− things like "write a for loop, see what it looks like in asm" sort of thing
08:50:43livvythis is good suggestion
08:51:02livvy*is a
08:51:52speachyyou will also get to see how interrupts are handled, and how hardare peripherals are driven.
08:52:22Bilgusand why people don't write in ASM anymore :P
08:52:25speachydoes wonders for clarifying your thinking about how to write this stuff. and eventually recognize these patterns well enough to reverse-engineer things
08:53:03*speachy has strong opinions that nobody should be allowed to write code in a JS browser framework without going through a bare-metal education process.
08:56:15speachylivvy, it helps that we usually have access to manufacturer datasheets for various components, including the SoCs. But not always.
08:58:36*gevaerts also has opinions about that, but due to limited experience with ASM he thinks C is bare-metal enough :)
08:59:55BilgusI think C forces you to at least understand some of the internal side but people hate that apparently
09:00:01gevaertsC is at least low-level enough that operations that are expensive start to look expensive
09:00:02livvyspeachy: yes, i were looking at some for the S5L8702 but were stumped trying to find details about the DW usb controller. but I were probably grasping at straws more than not so I need to some studying for sure
09:01:09livvygevaerts: :')
09:01:14gevaertsI *like* being able to concatenate strings by having a + between them, but I also think it's good to know that it's a lot more expensive than adding up numbers
09:01:23gevaertsAnd C shows you that
09:01:31gevaertsASM shows it even more, of course
09:02:00fs-bluebot_Build Server message: Build round completed after 1042 seconds.
09:02:34livvygevaerts: as a novice haskeller, i probably couldn't be further from adhering to your philosophy
09:03:40livvybut I do love C. suckless software has a close place in my heart :P
09:04:00fs-bluebot_Build Server message: New build round started. Revision 6ed38c8, 292 builds, 10 clients.
09:04:24speachyC is rarely insufficent, even for what we do.
09:04:50speachythe DW USB controller deserves to die a horrid death.
09:06:30BilgusI would argue that this is the perfect place for C I can't imagine interfacing with all these devices in any of the higher level languages and still have it be intuitive
09:06:43Bilgusnot to say it couldn't be done...
09:07:00livvyspeachy: finding the registers seemed to prove pretty hard for that thing
09:07:00speachyyou could make reasonable argument for C++.
09:07:13gevaertslivvy: I really have nothing against languages that make such things easy, that's a *good* thing. What's bad is the natural assumption of people who haven't seen the other side that what *looks* cheap *is* cheap
09:07:50speachyRust too, though I'm not sure how well it will work for true bare-metal applications.
09:07:53gevaertsSo I claim that if you're familiar with C, you're going to write better haskell
09:08:55livvygevaerts: yeah, I guess it's the "React devs" who've just gone though a 12 week coding bootcamp that I worry about :P
09:09:33livvys/worry about/have sympathy for/
09:10:41BilgusSpeachy are the devices in the build page able to be aranged differently or are they strictly alphabetic?
09:10:44*gevaerts nods
09:11:10speachy"autobuilder status" page you mean?
09:11:34Bilgusyeah I guess they are arranged based on position in the build list
09:11:38speachythey're alphabetic based on the long-form name out of the 'builds' file. the size listing is alphabetic based on the target name
09:12:48Bilgusjust slightly annoying all the blank spaces maybe they can disappear when the table compacts again
09:14:29speachyI added a test to show the column if any builds in it are missing. I could tweak it to exclude the retired builds.
09:19:27Bilgusthat would work I mean if they don't get build why have them highlighted in the list
09:19:59fs-bluebot_Build Server message: Build round completed after 958 seconds.
09:21:45Bilgusthat was quicker
09:22:44speachypartially fixed.
09:29:09speachydtange how the size didn't change for those last two clip commits. I'd have expected _some_ movement...
09:33:02speachybin size changed, but not the ram size.
09:38:41speachythoughts on how I can change the visualization? two tables, one bin, one ram that toggle like the short/full build list?
09:41:07Bilguscurrently it does some adding of the two no?
09:41:45speachyonly one or the other is actually displayed. both are available in the tooltip
09:41:57Bilgusbins didn't change because they are just getting inlined once something else calls them I'd expect that to change a bit
09:42:10speachywhat was displayed was ram usage, not bin size.
09:42:16BilgusOH i thought it added the change together
09:42:23speachyI regenerated it showing the bin
09:42:31speachywhich matches the descriptive text
09:43:44BilgusI think the bin size is probably more important but two tables wouldn't be terrible
09:44:23fs-bluebot_Build Server message: New build round started. Revision 4f49d21, 292 builds, 10 clients.
09:44:25speachymost targets, bin size goes up, so must ram usage.
09:45:46 Quit advcomp2019_ (Ping timeout: 246 seconds)
09:46:14 Join advcomp2019_ [0] (
09:46:14 Quit advcomp2019_ (Changing host)
09:46:14 Join advcomp2019_ [0] (~advcomp20@unaffiliated/advcomp2019)
09:49:18BilgusI do notice the sizes don't seem to be bouncing around anymore
09:50:02speachyI don't recall if the size list used to include sim/etc builds, but it doesn't any more.
09:50:20speachylost all the historical build data
09:50:24Bilgusoh maybe it did but I'm not thinking so
09:50:26speachysize data, I mean
09:50:45speachybut now sizes are stored in the DB properly.
09:51:30***Saving seen data "./dancer.seen"
09:57:09Bilgusbuild page looks much better less OCD food there, thanks!
09:59:13fs-bluebot_Build Server message: Build round completed after 890 seconds.
10:00:16speachydamnit, why is that last row still acting all screwed up.
10:03:16Bilguswas it just skipping them or something or maybe a formatting issue?
10:03:55Bilgusanyone object to g#1706?
10:03:57fs-bluebot_Gerrit review #1706 at : Sansa Clip+ Clip V1, V2 updated simulator images by William Wilgus
10:04:28BilgusI mean they've only been sitting there for ~3 years
10:04:32speachyno, it queries numrounds * numbuilds rows from the db, except numbuilds comes from the file and not the db. so when we nuke/retire builds, we end up pulling too few rows.
10:04:56speachythe proper solution is to write a smarter db query
10:05:18speachyI was going to ask about 1706 actually. I see no reason it can't go in
10:05:33speachyglad to see a little bit of gerrit cleanup
10:05:43BilgusI just kinda forgot about everthing on the second page
10:05:46 Part assa ("WeeChat 1.6")
10:06:43speachyheh, second of 17 pages. :D
10:07:04Bilguslol no I meant my second page :p
10:10:40Bilgusactually I went traipsing through gerrit I think theres quite a bit that just need to get abandoned
10:11:02speachyyeah. I think I have the rights to do that now but I had enough on my plate already
10:11:11Bilguspreferably by the original author but that seems unlikly
10:11:24speachythere are something like 300 old patches in flyspray too
10:11:27gevaertsI think you can safely ignore my stuff there. IIRC it's all fun experiments but nothing finished or maybe even "serious"
10:12:09speachyfirst round will need to be stuff that's already been merged one way or another.
10:12:25fs-bluebot_Build Server message: New build round started. Revision 2244305, 292 builds, 10 clients.
10:13:03Bilgusyeah superseded makes at least a few percent
10:13:54gevaertsHmm, there are a few I remember *nothing* about...
10:14:19gevaerts"Add a setting to allow preferring file-based album art"?
10:14:47Bilgusas opposed to folder based?
10:15:03gevaertstag, presumably
10:15:17gevaertsI mean, I only wrote that code, I have no idea what it does :)
10:15:30gevaerts g#756
10:15:32fs-bluebot_Gerrit review #756 at : Add a setting to allow preferring file-based album art. by Frank Gevaerts
10:15:40speachyhmm, this is interesting. g#272
10:15:42fs-bluebot_Gerrit review #272 at : Drop size entries from voicefiles. by Dominik Riebeling
10:18:56gevaertsspeachy: given other stuff you've done recently, g#1126 might be a quick yes/no for you
10:20:56speachygevaerts, looks like that change is already committed.
10:21:08gevaertsOh, great :)
10:22:57*gevaerts closes it
10:23:38gevaerts272 is interesting indeed. Seems like a good idea
10:25:13blbro[m]Hmm. Why did that never get in?
10:25:28*blbro[m] tries to remember
10:26:13speachy g#282 might be redundant now
10:26:14fs-bluebot_Gerrit review #282 at : Reworking playlist viewer to use memory more efficiently. by Purdea Andrei
10:26:20 Join johnb4 [0] (
10:26:22blbro[m]Be aware that this might need a rebuilt Rockbox Utility, or break its voice files with old releases.
10:26:51gevaertsThat's probably it. Maybe not enough review, and then the logistics of rolling it out
10:26:55fs-bluebot_Build Server message: Build round completed after 870 seconds.
10:26:56speachyew. I keep forgetting about rbutil's out-of-band nature.
10:28:22speachyon the size table, does it make sense to exclude the 0-size deltas from the averaging, so it only averages those that changed?
10:30:22fs-bluebot_Build Server message: New build round started. Revision 326b7a5, 292 builds, 10 clients.
10:30:27gevaertsInteresting question
10:30:31blbro[m]Rockbox Utility could easily create boths formats if there is a way to tell which one to use.
10:31:27blbro[m]Once that cryptopp thing is sorted I can also create an updated macos binary. I've also looked into creating an appimage for linux lately.
10:31:42gevaertsThere are two reasons a delta can be 0. One is alignment and other randomness, in which case it should count I think, the other is that that target doesn't have the changed feature, in which case counting it is somewhat misleading
10:32:15BilgusRe: G-282 is so old that most of it doesn't even exist anymore
10:33:44gevaertsLooks like I have the dubious honour to have the oldest unmerged change request :)
10:36:24Bilguslol 2012 huh
10:38:33Bilgussomeone was asking about functionality provided by G#174
10:38:35fs-bluebot_Gerrit review #174 at : Play music or complex tones with square wave channel. by Michael Sevakis
10:38:45Bilgusmaybe that was on the forum
10:44:26fs-bluebot_Build Server message: Build round completed after 845 seconds.
10:44:50 Join Infinite777 [0] (~Infinite7@
10:45:39speachyat build time we don't know why a delta is 0, only that it is.
10:45:55speachymost of the time it's that nothing changed.
10:46:12gevaertsDoes anyone actually really care about the average though?
10:47:05speachygood question. IMO it's useless, but I left it in.
10:48:44gevaertsYes, it doesn't hurt a lot to just have it there
10:49:26gevaertsBut changing what it means requires an informed opinion I think, not a hunch from someone who doesn't look at the number :)
10:56:13 Join krabador [0] (~krabador@unaffiliated/krabador)
11:00:29speachyI mean, one can't infer anything useful from the delta of 1 across those two clip commits you made.
11:00:39speachybeyond "something changed"
11:02:05gevaertsA quick glance at the delta table is mostly useful to spot unexpected changes or to check that expectations match reality if you did intend a change
11:02:29gevaerts"Why did the Archos Player increase for this bitmap-only change?"
11:31:54speachysorry, meant to say "average delta of 1"
11:32:27speachydeltas are quite useful, I find myself agreeing that the averages are questionable.
11:36:15*gevaerts nods
11:38:39speachymade the change to only average the ones that changed
11:51:33***Saving seen data "./dancer.seen"
11:58:30speachythat's odd.. build a8ae936 isn't showing up. utterly failed across the board due to an ISP hiccup.
12:05:01 Quit michaelni (Read error: Connection reset by peer)
12:07:28 Quit pamaury (Ping timeout: 272 seconds)
12:09:46 Quit MrZeus (Ping timeout: 260 seconds)
12:12:29 Quit johnb4 (Quit: Nettalk6 -
12:13:37speachyfixed. blank row is ugly but it's important that we know that a round simply didn't happen.
12:16:50fs-bluebot_Build Server message: New build round started. Revision a8ae936, 292 builds, 10 clients.
12:17:07speachylet's see how badly this breaks the master db. :D
12:21:52 Join michaelni [0] (
12:33:00fs-bluebot_Build Server message: Build round completed after 971 seconds.
12:35:23speachyexcellent. Just had to correct the round timestamp so the ordering was correct
12:39:07 Quit krabador (Remote host closed the connection)
13:03:26 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
13:38:04gevaertsspeachy: looks like the "rockbox forums" image on the forums is still loading over http
13:40:47gevaertsI'm trying to find where that's set...
13:43:39gevaertsFound it! I'll fix it
13:44:54gevaertsAdmin -> Configuration -> Current Theme -> Logo image URL
13:46:08gevaertsI suspect that can be a relative path, actually, but I didn't want to risk breaking stuff
13:51:37***Saving seen data "./dancer.seen"
14:02:47 Join S|h|a|w|n [0] (~shawn156@unaffiliated/shawn156)
14:10:14speachysweet, thanks
14:10:53speachysmf has a lot of wonkiness like that. For example the base URL for themes is different for some reason.
14:43:53gevaertsFinally having that on https is good
14:50:07BilgusEspecially with the way things are going forced https and whatnot
15:01:32 Join krabador [0] (~krabador@unaffiliated/krabador)
15:06:45 Join MrZeus [0] (
15:08:21Bilgushmm html manual is broken for the clipzip yet the pdf version is fine :/
15:09:13 Quit riku1 (Ping timeout: 264 seconds)
15:09:36 Join riku1 [0] (
15:32:56Bilgusoh great its been so long since I built a m,anual I nolonger have laTEX :/
15:37:49speachywas about to say, the nightly manual builds work for all targets..
15:38:56Bilgusno it is borken
15:39:23Bilgusyet the exact same section is fine for the clip+
15:39:51Bilgusbut only the html version is messed
15:40:05speachywhich section?
15:41:15Bilgusfuzeplus is broken in the same place :/
15:41:19speachyhuh, guess the nightl manual builds don't properly error out.
15:41:57Bilgusyet the pdf versions are fine?? WEIRD
15:42:31speachyhtmltex has its own quirks
15:42:47Bilguswell in 40 - 60 mins maybe ill be able to build a manual
15:45:14speachy'Environment exampleredirect undefined'
15:45:23Bilguspretty sure its been broken for 10 months since I updated the manual for multiboot
15:46:51Bilguswonder wtf its failing on that
15:47:50speachythat 'exampleredir
15:48:11Bilgusit has a /begin tag just prior
15:48:14speachyit's barfing on that.
15:48:56Bilgusill just make it quote it
15:50:43Bilgusthanks speachy i'll still have to wait for this install to build it but atleast I can do my patching
15:51:14speachyfrustrating that 'make manual-html' isn't erroring out properly.
15:51:41***Saving seen data "./dancer.seen"
15:52:37Bilgusoh I see the issue begin is suppose to denote a list and I bet it is picking up the slash as such
15:54:49Bilgushmm maybe not
16:32:20 Join bzed [0] (
16:36:33 Quit bluebrother (Disconnected by services)
16:36:38 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
16:37:16 Join fs-bluebot [0] (
16:39:23 Quit livvy (Ping timeout: 240 seconds)
16:39:57 Quit fs-bluebot_ (Ping timeout: 265 seconds)
16:41:18 Join livvy [0] (~livvy@gateway/tor-sasl/livvy)
16:41:34 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
16:41:35 Join fs-bluebot_ [0] (
16:43:19 Quit fs-bluebot (Ping timeout: 246 seconds)
16:43:49 Quit bluebrother^ (Ping timeout: 265 seconds)
16:51:53Bilguswow that was a lot of warnings
16:52:12fs-bluebot_Build Server message: New build round started. Revision b935475, 292 builds, 10 clients.
17:05:40fs-bluebot_Build Server message: Build round completed after 808 seconds.
17:10:16Bilgusah crap I guess I'll have to wait till tomorrow to see if it built proper
17:18:43speachyseemed to build ok for me
17:19:08speachych13 now goes all the way to 13.5.5 and navigation buttons
17:19:34BilgusYeah I built it locally first but I think it only updates on the main site once daily?
17:20:31speachyI've been thinking of making the manual builds farmed out like the others
17:20:59BilgusI think i'd leave it like that
17:21:20Bilgusthat way we have some time if something breaks horribly
17:21:44speachyyou mean leave it as-is?
17:22:01speachyat the moment it goes months before anyone notices. :)
17:22:36speachytex4htm failing without an appropriate error code is a problem.
17:22:37Bilguslol yeah but I do check for my own stuff at least once a year!
17:23:27BilgusI try to play naive user every once in a while and sometimes I forget enough that its actually true
17:24:15Bilgusthough try as I might today I couldn't reproduce the failure to reformat a 128GB exfat => Fat32
17:31:17speachythe nightly manual build is a little wonky though. "version rUnversioned directory-200523"
17:32:11speachylook at page two of any of the PDFs up right now
17:33:44Bilgushuh maybe something with LaTex 's cfg?
17:34:54BilgusI should go look at one of the old ones
17:38:03speachythis has to do with the incestuous symlinks all over the place; is getting invoked from a path outside its git repo
17:41:58BilgusI only have old html versions unfortunately
17:43:06speachyor the splash page of any html manual on the site now too
17:49:53speachyfingers crossed. let's see if this succeeds..
17:50:09BilgusI have a html manual built in 2016 that has the same
17:50:20Bilgusso its not new
17:50:20speachywell, it's getting fixed now.
17:50:31BilgusGo Speachy!
17:51:42***Saving seen data "./dancer.seen"
17:53:04*speachy needs relatively non-braintaxing stuff to distract him from the post-surgical colors.
17:57:30 Quit user890104 (Remote host closed the connection)
18:05:55speachyokay, fixed.
18:06:22speachyrebuilt only this one but the rest will get picked up tonight
18:06:34 Quit pamaury (Ping timeout: 272 seconds)
18:15:23 Join user890104 [0] (
18:16:34 Join petur [0] (~petur@rockbox/developer/petur)
18:21:23 Quit lebellium (Quit: Leaving)
18:59:11 Quit ac_laptop (Ping timeout: 260 seconds)
18:59:50 Join ac_laptop [0] (~ac_laptop@
19:24:18 Quit petur (Quit: Leaving)
19:33:23 Join skolem [0] (~skolem@
19:34:19 Quit skolem (Client Quit)
19:51:46***Saving seen data "./dancer.seen"
20:55:18 Join Piece_Maker [0] (
20:56:07 Quit Acou_Bass (Ping timeout: 265 seconds)
20:56:29 Nick Piece_Maker is now known as Acou_Bass (
21:08:26 Quit sakax (Quit: Leaving)
21:10:10 Quit MrZeus (Ping timeout: 256 seconds)
21:21:42 Quit krabador (Remote host closed the connection)
21:51:49***Saving seen data "./dancer.seen"
21:57:54 Quit ac_laptop (Ping timeout: 240 seconds)
23:05:38 Join J_Darnley [0] (
23:06:10 Quit jdarnley (Read error: Connection reset by peer)
23:33:02 Quit J_Darnley (Read error: Connection reset by peer)
23:35:19 Join J_Darnley [0] (
23:42:27 Quit beencubed (Ping timeout: 260 seconds)
23:51:51***Saving seen data "./dancer.seen"
23:54:38 Join beencubed [0] (~beencubed@

Previous day | Next day