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 2021-07-09

01:08:06***Saving seen data "./dancer.seen"
01:29:32 Join ufdm_ [0] (
01:32:36 Quit ufdm (Ping timeout: 272 seconds)
01:47:26 Quit pixelma (Quit: .)
01:47:26 Quit amiconn (Quit: - Chat comfortably. Anywhere.)
01:49:56 Join pixelma [0] (
01:49:56 Join amiconn [0] (
02:14:18 Join desowin [0] (
03:08:07***Saving seen data "./dancer.seen"
04:24:31 Quit j-r (Ping timeout: 246 seconds)
04:36:41 Quit akaWolf (Ping timeout: 252 seconds)
05:08:09***Saving seen data "./dancer.seen"
05:15:54 Quit yosafbridge (Quit: Leaving)
05:16:38 Join akaWolf [0] (
05:16:39 Join yosafbridge [0] (
05:24:19 Join j-r [0] (
05:24:40rb-bluebotBuild Server message: New build round started. Revision 60e2cd6de9, 297 builds, 8 clients.
05:37:21rb-bluebotBuild Server message: Build round completed after 760 seconds.
05:37:23rb-bluebotBuild Server message: Revision 60e2cd6de9 result: All green
07:04:05 Quit akaWolf (Ping timeout: 255 seconds)
07:08:12***Saving seen data "./dancer.seen"
07:34:01 Join akaWolf [0] (
07:49:54 Join massiveH [0] (
08:00:19speachywell, LTO on the X3 did introduce at least one regression −− rolo broke.
08:03:09braewoodsspeachy: which would imply it also broke the BL, no?
08:03:48braewoodsin any case it sounds like LTO has a lot to work before we could turn it on
08:05:11speachyTheres' going to be a long tail of regressions.
08:05:30braewoodsso why do all this then?
08:06:04braewoodssome of the LTO issues may be resolved or changed by newer toolchain
08:06:05speachysmaller binaries and generally better performance/optimization
08:06:38braewoodsshould we aim for a new toolchain after the next release?
08:06:48speachyeven modern GCC 11 with sim builds doesn't LTO properly yet.
08:07:08braewoodshuh. i'd expect hosted builds to work the the easiest
08:07:32braewoodsi never touched LTO since it seemed to cause way more issues than it solved
08:07:56speachyI've been involved in several embedded projects that simply wouldn't have fit without LTO.
08:08:23braewoodsah. well i've mainly developed for Linux so that's why i think differently.
08:09:08braewoodsi have more resources so i want to side with maturity a lot more; i don't have patience to workout random issues due to LTO issues.
08:09:44speachymost of the LTO "issues" are ultimately due to bugs (or at least a reliance upon undefined behavior) in the code.
08:09:52braewoodsspeachy: btw i found an mpio hd300 and i bought it
08:10:01braewoodsi'll be able to test it
08:10:20braewoodsonce it arrives
08:10:21speachyeg definitions that don't match, namespace collisions
08:10:31braewoodswe don't have namespaces so
08:10:47braewoodsbut we can still have prototype mismatches
08:10:50braewoodsor so
08:10:53speachyand being able to trace code execution across "modules" allows for a lot of other issues to be uncovered.
08:11:01speachynamespace meaing "compilation unit"
08:11:25braewoodsC standard calls those translation units
08:11:54braewoodsso you're saying we've got a lot of code to fix before we can use LTO.
08:13:19braewoodsi may be engaging in some UB just to get my code to work... :|
08:13:26braewoodsmostly with casts
08:13:57braewoodssometimes i discard const because i know the original data is truly not const and i just want it to stop warning
08:14:06MarcAndersenI'm sorry to say that my connect has still not powered on. I think this thing is totaly dead.
08:14:10braewoodswhen dealing with function parameters that want a "const char**"
08:14:29braewoodsor rather, fake it being const. the originaly is char**.
08:14:56braewoodsMarcAndersen: sorry to hear that.
08:15:14braewoodsif you lived in the US i'd send you one of my extra rockbox units. :|
08:16:05MarcAndersenI got this one on ebay from us.
08:17:28braewoodsi mostly own non-ipod hardware
08:17:52MarcAndersenI will just let it be connected and see if it magicly turns on some day, but I wish there was a way for it to ignore the battery and turn on by the help of usb power and then charge. All of my laptops can do that when I connect them to the mains.
08:17:55braewoodsi have 4 gigabeat S units i've acquired by watching for deals on ebay; i've repaired and upgraded the 3 i could open
08:18:31braewoodsif that's what you'd like, i suggest the gigabeat ports. both can be powered from either direct DC power or battery
08:18:49braewoodsGigabeat S and Gigabeat F (or X if you're lucky enough to find one)
08:18:59speachyMarcAndersen: alas, ultimately it depends on how the power supply was designed. :/
08:19:21braewoodsthe gigabeats have a battery switch too even
08:19:24MarcAndersenMaybe that's what i should look after, thanks.
08:19:54braewoodsi'll be doing an upgrade of a gigabeat F later this week
08:20:14braewoodsYou can get some good deals on these F series from time to time.
08:20:33braewoods$30-50, compared to the $100+ some idiots want.
08:20:52braewoodsthe S series is pretty rare to find but I have found 2 for $50 or less in the last few months.
08:21:01MarcAndersenOh, I also have a small problem with my nano 1g. It takes a very long time to start transfering files and to delete them, like it needs to do it 2 times. Is that something you can confirm?
08:21:21speachyusing the latest builds?
08:21:32speachyI'm not aware aof any specific problems with the nanos
08:21:46MarcAndersenOh, sorry, maybe it's on 3.15.
08:22:01MarcAndersenI will try updating if it is.
08:22:19speachythere have been a ton of general improvements but the nanos were unaffected by the iflash woes of their spinning rust bretherens
08:23:28braewoodsseems the entire EU ebay lacks any gigabeats
08:23:29 Quit ufdm_ (Quit: Leaving)
08:23:32braewoodsall US it seems
08:23:40 Join ufdm [0] (
08:23:52speachybut since the nano doesn't have replaceable internal storage their days are all numbered.
08:23:57desowinSansa Connect OF bootloader won't let you through unless the HDQ reports that battery is fine
08:25:05braewoodsMarcAndersen: unfortunately most of the options i've mentioned are mainly a good idea if you can get them repaired and upgraded. you'd probably need a sighted person to do that first.
08:26:50braewoodsin any case i'd be willing to do that at some point if i know what you require
08:28:57desowinMarcAndersen: if you don't mind shipping to Poland I can check if the battery can be recovered
08:30:04MarcAndersenThat would be fine. Should we make a private chat or something?
08:30:25speachyMarcAndersen: one question −− was there a reason you chose the Connect specifically?
08:30:45braewoodswe don't need to know the details and best it doesn't show up in the logs
08:30:46MarcAndersenYes, to test the new port mostly.
08:31:46speachyah, okay, so you weren't after it due to some physical consideration. (Some hardware is definitely better than others for visually-impaired folks)
08:31:57braewoodsof which i'm no expert on
08:32:10speachywe definitely appreciate your willingness to suffer...
08:32:16MarcAndersen/msgnickserv identify marc7002
08:32:20braewoodsbut anything touch based is probably a challenge
08:32:23MarcAndersensorry sorry sorry
08:32:33braewoodsMarcAndersen: may want to change your password now
08:32:36MarcAndersencan you delete that from the log?
08:32:45braewoodsit's automated to the web site
08:32:51MarcAndersenI will change my pass
08:32:52speachyMarcAndersen: yes but not until it rotates.
08:33:05braewoodswell speachy can
08:33:09braewoodsi don't have access to logs :P
08:33:15speachy(and of course who knows how many eyeballs and/or other bots have seen it)
08:37:49 Quit massiveH (Quit: Leaving)
09:02:35_bilgusMarcAndersen, when a lithium ion battery goes too low it can actually flip polarity and when you go to charge it it is now backwards and has a melt down
09:03:04_bilgustherefore they put protection circuits to disconnect when its too low
09:04:13_bilgusby taking 2 standard AA batteries in series to make 3v you can get some power in it safely-ish
09:06:39_bilgusI usually have a 3.3v supply with a resistor and let it charge for a bit on a metal plate (don't make it encased in a pressure vessel)
09:07:37_bilgusI've yet to have one blow but you do need to make sure its not getting hot or expanding
09:08:00_bilgusas for results IDK more good than bad
09:08:14***Saving seen data "./dancer.seen"
09:09:42_bilgusoddly the worst are cordless tool batteries - one pack every cell was .001 v I didn't even try it
09:19:48_bilgusMarcAndersen, I kinda searched around here is a guide that fits pretty well with my method
09:20:17_bilgusthey use a dedicated charger which is probably a good idea
09:21:03MarcAndersen_bilgus: I will check it out
09:21:55_bilgusI've charged a cell battery on a dedicated liion charger for 18650s its not rocket surgery
09:22:34_bilgusyou just need to be aware of the potential for damage to your surroundings and do it in an appropriate area
09:23:08 Quit aquijoule__ (Remote host closed the connection)
09:23:33 Join aquijoule__ [0] (
09:24:36_bilgusits usually just a bit too low, 2.6v or something just shy of allowing it to work you get some current in there and they usually start working
09:25:04braewoods_bilgus: wouldn't a CC/CV power supply work just as well?
09:25:11braewoodsassuming you set it correctly
09:25:20_bilgussure a little benchtop supply
09:25:21MarcAndersenI'm going to send my connect to desowin
09:26:33 Quit j-r (Ping timeout: 252 seconds)
09:26:33braewoodsMarcAndersen: if you end up needing a repaired rockbox unit, let me know.
09:26:50braewoodsi usually repair the ones i buy to use for development as i need stable hardware to work with
09:28:59_bilguslike I said I use a beefy 3v3 supply and a large-ish resistor 10k? to great effect
09:29:18 Quit pablocastellanos (Ping timeout: 252 seconds)
09:29:20braewoods_bilgus: resistor to control current flow?
09:29:44_bilgusjust a little trickle
09:29:44 Join j-r [0] (
09:29:45braewoodsafaik batteries generally have nothing to stop it
09:29:51braewoodsso they need external help
09:30:10braewoodsthat's usually an issue mainly in charging
09:30:50_bilgusthe other thing is that it keeps the temp low
09:31:14braewoodstrue. most simple lithium packs lack temperature sensing
09:32:04_bilgusyou'd be bypassing it anyways I'm talking of keeping the heat low to keep from melt down
09:32:19braewoodsthermal runaway
09:32:46braewoodsunder normal operation you'd not let it get that low
09:33:17_bilgusyeah Ive yet to actually do that on purpose to a liion battery the videos were enough of a warning
09:34:09braewoodsi've had devices use a very low current to charge a dead battery for awhile
09:34:17braewoodsand then it spikes
09:34:20braewoodsafter awhile
09:34:29_bilgusyeah thats a safety circuit
09:34:37braewoodslike 0.1 or 0.2 A from a 5V input
09:35:03braewoodsi usually use a dc watt meter to monitor the charging current
09:35:19braewoodsthat tells me a lot about how it is progressing
09:35:29_bilgusthose battery chargers are a lot smarter than the lead acid 'smart' chargers ever were
09:35:55braewoodsSLA chargers are never disconnected automatically iirc
09:36:08_bilgusbut they have to be liion is a lot of unstable energy in a small package
09:36:35braewoodswhat about lifepo4?
09:36:36 Quit advcomp2019 (Ping timeout: 252 seconds)
09:36:43braewoodsisn't it safer?
09:36:51_bilgusIDK I haven't ran across any cells yet
09:37:07braewoodsah. i'm planning to build a lifepo4 pack
09:37:10braewoodsi have the cells already
09:37:12braewoodsthey're 26650
09:37:58_bilgusoh? are they the same voltage range or just don't get full capacity if used in non life devices?
09:38:00braewoodstheir voltage range is different so you need a different type of BMS and charger for them
09:38:11braewoodstheir max voltag is around 3.6 or 3.7V
09:38:31braewoodssome people use 4S setup to replace SLA packs
09:38:38_bilgushmm then maybe using the same dimensions as existing liion is BAD juju
09:38:56braewoodsthey're 26650 for a reason
09:39:11braewoodsthose are larger and primarily used by lifepo4 cells
09:39:26braewoodsthere are 18650 lifepo4 cells but
09:39:27_bilgushave 3 flashlights that say otherwise
09:39:30braewoodsthey're pretty rare
09:39:43_bilgusno 21700s are rare
09:40:31braewoodsi'm planning a build using those 21 cell boxes they sell from china or so
09:40:45_bilgusah 21650 not 26650 I think
09:40:47braewoodsthey can fit 21 18650s in a cell holder or 10 26650s
09:40:56_bilguslet me look
09:41:11_bilgusno 26650
09:41:23 Quit j-r (Ping timeout: 252 seconds)
09:41:30braewoodsif you're interested, i think the daly bms is the best overall option that i could find
09:41:55braewoodsyou can order them with thermistors too
09:41:55 Join j-r [0] (
09:42:17braewoodsit was the closest i could find for a "brand name" bms from china
09:42:49braewoods_bilgus: you can also buy those headway lifepo4 cells
09:42:50_bilgusIve been toying with making a liion pack for my UPS esp after speachys 'experiments' with LA
09:42:58braewoodsno soldering
09:43:11braewoodsor welding
09:43:19braewoodsthe cells use screw on caps
09:43:21_bilgussoldering is fine I need to make a spot welder anyway
09:43:33_bilgusah like those caps of old
09:43:51braewoodswhich seems fine but those cells are expensive
09:43:56braewoods$8+ a pop
09:44:11speachymy wet-cell failures were more due to a lack of maintainene than anything else. I didn't keep the cells sufficiently, um, wet.
09:44:34_bilgusour equipment at work does the same
09:45:07braewoodslifepo4 is the only real option for a drop in replacement
09:45:14braewoodssince it has a compatible voltage range
09:45:20braewoodsin 4S
09:45:21speachybut LiFePO4 is $$$$ and requires active regulation at the battery or completely different charging circuits. I wouldn't trust one with a charger designed for SLAs
09:45:39braewoodsi've read people do that though speachy.
09:45:50braewoodsas long as the UPS only does regular CC/CV
09:45:53_bilgusno if anything I figure I can use a circuit to fake the onboard
09:45:53braewoodsit should be ok
09:46:15speachywhen you're talking 100s of amp-hours that's a lot of boom should something go wrong.
09:46:17braewoodsbut i wouldn't attempt it without understanding how the UPS charger works
09:47:08_bilgusa battery explosion is not a fun experience
09:47:40speachyI think the solar charge controller I have in my camper can hendle LiFePO4; it has different charge modes for wet cell, SLA, and Li
09:47:40braewoods_bilgus: in any case, i'd check out the daly BMS store on aliexpress if you need a BMS for a custom lifepo4 or lithium pack.
09:47:41_bilgusI was deaf for a day or two when one exploded in a bulldozer I was attempting to start
09:48:16braewoodsi use primarily the single port options
09:48:24braewoodsso i can use them in a regular battery case
09:48:38braewoodswhere the same port is used for charge and discharge
09:49:11braewoodsjust need to ask the seller to enable their thermistor part for sale if you want to order one with an optional thermistor
09:49:44braewoodsit's soldered to the BMS so it has to requested prior to shipment
09:49:50speachybefore I was using wet cells I was using very oversized SLAs, but got sick of paying through the nose every couple of years to replace 'em. Gotta love Florida's power quality.
09:50:14speachyif I were do it again I'd move to golf cart batteries instead. they're designed for abuse.
09:50:46braewoodsmust be a large scale UPS
09:50:54_bilgusjesus guess florida has a lot of em
09:50:57braewoodsmost UPS i've used can't handle random SLAs
09:50:59speachyI'd sized it to handle a five hour outage.
09:51:13braewoodsthey have to fit a small profile
09:51:23 Join pablocastellanos [0] (~pidgin@user/pablocastellanos)
09:51:28speachythis one couldn't; I rigged up custom cables. sat on the floor under my server "table" :D
09:51:51braewoodsmost UPS can't be used with external batteries that i can tell
09:52:08braewoodswhat voltages do these higher end UPS use?
09:52:11braewoods12V still?
09:52:29braewoodsi would expect for the kind of loads you have that they'd use higher voltage to reduce the wiring costs
09:52:34speachyone of my printers here has visible acid damage from the more recent BOOM
09:53:27speachy1300VA or so tends to be 24V. higher-end stuff (>2KVA) is often 48V.
09:53:34_bilgusI guess the other option is a vented cabinet
09:53:40_bilgusvented to outdoors
09:53:45braewoods_bilgus: one thing to note if you try to build a lithium pack replacement. it may have to be external since many UPS expect 12V range.
09:54:07braewoodslifepo4 doesn't handle high discharge current well
09:54:16braewoodsmy cells are only rated for about 10A each
09:54:33braewoods500W ups easily needs like
09:54:38braewoods40A or so
09:54:55braewoodsSLAs can handle that
09:55:19braewoodsyou'd need a larger lifepo4 to replace it due to discharge current
09:55:26_bilgusI have surplus shipboard UPSs from the 90s they are more like 15-20A @ 120 so 2000W
09:55:33MarcAndersenMaybe this is a stupid question, but does the gigabeats have touchscreens?
09:55:43braewoodsMarcAndersen: no. neither do.
09:56:07MarcAndersenOk. Maybe that's what I should look after next...
09:56:19braewoodsMarcAndersen: Gigabeat S has a directional button pad below the screen. Same with the F.
09:56:30braewoodsthe rest are side buttons or switches
09:56:37MarcAndersenWhich is most accessible?
09:56:46braewoodsdefine accessible
09:56:56MarcAndersenI mean for the blind.
09:57:11rb-bluebotBuild Server message: New build round started. Revision 8de163b8ae, 297 builds, 9 clients.
09:57:12speachybraewoods: "usable without being able to see where your fingers are"
09:57:13braewoodswell, most of our targets don't use touch screens.
09:57:16_bilgusare the rockers alrady too hard to get?
09:57:20braewoods_bilgus: yep.
09:57:47MarcAndersenIs the best processor in the gigabeat s?
09:58:02braewoodsMarcAndersen: probably the S. the F has a sunken in pad that can be hard to trigger consistently.
09:58:09braewoodsbad news if you can't see well
09:58:25_bilgussee I think we should be looking at ease of replacemnt as well
09:58:29braewoodshow much storage you need?
09:58:38_bilgushow easy is replacimg the battery or storage
09:58:45speachyMarcAndersen: IMO the best option for blind users is the original xDuoo X3. funky buttons that are very easy to tell apart by touch.
09:58:47braewoodsthe gigabeat S is pretty easy actually
09:58:53speachyand the crappy screen doesn't matter. :)
09:59:12braewoodsfor a sighted person to repair anyway
09:59:27braewoodsi've finished repairing 3 gigabeat S to functional state
09:59:40braewoodsupgraded their storage too
09:59:48MarcAndersenDo any of the gigabeats or xduoo have external storage?
10:00:00braewoodsthe gigabeat S has no external storage
10:00:06braewoodsonly the internal
10:00:13braewoodsbut it is replaceable
10:00:13speachyX3 has dual SD slots. no internal storage.
10:00:24_bilgusjust saying already got burnt once with an older device one that is repairable would make that less terrible
10:00:47_bilgusIF you can get a new X# or new-ish
10:00:58braewoodsthe gigabeat S is a pretty beefy device for its age
10:00:59speachythe newer X3ii has a single SD slot, but not as accessible due to its hosted nature (bootloader is not accessible)
10:01:16braewoodsi installed a 128GB SSD in all 3 of the ones i repaired
10:01:19braewoodsand new batterie
10:01:23MarcAndersenDoes any of them have a built in speaker which works in rockbox?
10:01:29braewoodsSadly no.
10:01:35braewoodsmost players need headphones
10:01:58MarcAndersenThe connect has one, right?
10:02:04braewoodsare there any rockbox players with built in speakers?
10:02:08braewoodsno idea i'll look
10:02:12MarcAndersenthe zen x-fi
10:03:00MarcAndersenI got that in 2009 and tested it during development in 2013.
10:03:56braewoodssansa connect has some holes on theback; it might support a speaker
10:03:59braewoodsi'd ask desowin
10:04:26MarcAndersenBut first I need to get it to start up...
10:04:55MarcAndersenStupid peace of crap
10:05:15braewoodsthe connect is not easy to repair from what i saw
10:05:23braewoodsthe battery has no aftermarket replacements
10:05:40MarcAndersenMaybe I should just drop sending it to desowin then?
10:05:42speachythere's always aliexpress
10:06:05braewoodsspeachy: really... i didn't see any.
10:06:15braewoodsi had to import my gigabeat S batteries
10:06:20speachywell, yeah.
10:06:47MarcAndersenMaybe I should just throw it in the trash and get another model?
10:06:50braewoodsMarcAndersen: up to you.
10:07:04braewoodsMarcAndersen: that's just my take on it a a sighted person
10:08:01rb-bluebotBuild Server message: Build round completed after 649 seconds.
10:08:03rb-bluebotBuild Server message: Revision 8de163b8ae result: All green
10:08:32desowinSansa Connect has built-in speaker
10:08:40braewoodshuh. interesting.
10:08:48braewoodsthat's pretty rare to find.
10:09:08desowinit is mono, and OF (and Rockbox) switches to it when there are no headphones connected
10:09:26 Join advcomp2019 [0] (~advcomp20@user/advcomp2019)
10:09:28braewoodsare there any other ports with it?
10:09:46desowinMarcAndersen: it is up to you
10:10:01speachyoh! that reminds me.
10:10:30speachydesowin: can you grab an appropraite voice file and make sure it's sane?
10:11:50desowinI would say the speaker is pretty bad. and I don't think you can use it to output something else than on the headphones
10:12:13braewoodsMarcAndersen: the gigabeat S is still under development but i can send you one of the repaired units. it's basically done but still lacks rbutil integration, main reason it's not "stable".
10:14:05MarcAndersenbraewoods, I think that would be better then.
10:14:36braewoodsi can preinstall rockbox for you if you want. currently they run the OF.
10:14:56MarcAndersenbraewoods, Yes please do that
10:15:06 Quit MarcAndersen (Remote host closed the connection)
10:15:07braewoodsi'll pm you then for the rest.
10:15:29braewoodswell they'll be back at some point if they want it
10:15:30 Join MarcAndersen [0] (~no_znepna@
10:15:38MarcAndersenMy client crashed
10:15:53braewoodswe saw the quit message
10:16:00braewoodsi'll pm the rest
10:16:02MarcAndersenwhat did it say?
10:16:46braewoods"MarcAndersen (~no_znepna@ has quit (Remote host closed the connection)"
10:17:09MarcAndersenOh that's my ip
10:19:38desowinspeachy: I just used the voice menu for the first time
10:20:22desowinI seems to work (with Polish voice pack), just clearly audible it is synthesized
10:22:10speachydesowin: it's generated using espeak-ng's defaults, if there are better-sounding settings to use that's easily changed.
10:25:48braewoodsspeachy: do you know what processor the ipod 6G uses?
10:25:51braewoodsi can't find it on
10:26:48speachyit's on the wiki. some samsung arm9xx part
10:27:51MarcAndersenNo matter what it's better than these 2 portalplayer devices I have left which actually works
10:28:25braewoodsthe gigabeat S has 64MB of RAM.
10:28:31braewoodsan armv6 SoC
10:28:41braewoodsand some other features most ports lack
10:28:46braewoodslike FM Radio RDS
10:28:54braewoodsbut i think that's moot for Europe; didn't they do away with FM there?
10:34:41MarcAndersenOk. It doesn't matter which one you send, just the one you use the least.
10:36:06speachyone thing that concerns me is that we can only spell out ascii words.
10:36:26braewoodscan we do better with our limited resources?
10:36:45speachybut I don't really see a good way to improve that.
10:36:49braewoodsi'm no expert on TTS.
10:37:21speachythe idea is you'd generate appropriate voice snippets for your specific files.
10:37:34speachyfalling back to spelling out words
10:37:44speachybut that fails for non-english.
10:37:56speachy(or rather, non-ascii)
10:38:14braewoodsthe main problem is what TTS solutions could you use with rockbox's limited resources?
10:38:54braewoodsi'd assume we'd use a third party one if we were trying to improve it
10:44:12braewoodsspeachy: how do i build voice files?
10:44:22braewoodsi don't normally use them.
10:44:24MarcAndersenjust press v in advanced build
10:44:36speachyconfigure script, advanced, lets you generate voices.
10:44:38braewoodsdoes it have to be part of the normal build process? hm.
10:44:42MarcAndersenand make voice
10:44:43speachyit's not.
10:45:23speachybuilding voice files from the cmdline is vastly improved from what it used to be but it's still relatively crappy
10:45:33MarcAndersenI'm soon going out for a while, but I'll read the logs when I come back.
10:46:16speachythe build infra piggybacks on top of the configure script with options taken from
10:47:00speachybut generally speaking the expectation is that users will generate voices (and file talk clips) using rbutil, which has its own quidks.
11:00:45MarcAndersenI'm going out now just so you know
11:03:25braewoodsspeachy: syntax error from configure while doing that
11:03:35braewoods../tools/configure: 1156: [: =: argument expected
11:04:18braewoods../tools/configure: line 1156: [: : unary operator expected
11:04:23braewoodswhen i use bash
11:06:03speachyat what step?
11:07:06speachypretty sure that's because you dn't have any of the supported tts engines instealled
11:07:07braewoodsafter choosing the english voice or so
11:07:16speachyshouldn't be a syntax error though.
11:07:20braewoodseven so it should still give a proper response
11:07:31braewoodsnot leave me wondering what's wrong
11:07:34speachyah found it, a missing space
11:08:02speachyedit the script, line 1156, find the "] and change it to " ]
11:08:15***Saving seen data "./dancer.seen"
11:08:21speachy"$MIMIC"] -> "$MIMIC" ]
11:09:28braewoodsthat should be patched in a commit though
11:09:56braewoodswhat software is recommended?
11:10:16speachyespeak-ng is what the nightly voice builds utilize
11:10:28speachyfestival might be the best for English
11:10:59rb-bluebotBuild Server message: New build round started. Revision 663539619c, 297 builds, 9 clients.
11:14:42braewoodssheesh. no wonder this isn't done much. building the voices takes a long time.
11:15:22braewoodshow do i make a zip archive for the voice stuff...
11:21:44rb-bluebotBuild Server message: Build round completed after 644 seconds.
11:21:46rb-bluebotBuild Server message: Revision 663539619c result: All green
11:23:31braewoodsok found it
11:25:41 Quit JanC (Ping timeout: 265 seconds)
11:28:31 Join JanC [0] (~janc@user/janc)
11:39:28braewoodsinteresting, i'm going to have to investigate this
11:39:41braewoodsspeachy: i've produced panics when using the remote with gigabeat S.
11:39:50braewoodsi'll see if i can find out the issue later.
11:52:38 Quit advcomp2019 (Read error: Connection reset by peer)
11:53:22 Join advcomp2019 [0] (~advcomp20@user/advcomp2019)
12:02:47 Join ZincAlloy [0] (
13:08:18***Saving seen data "./dancer.seen"
13:58:29braewoodsi've got a head scratcher...
13:58:51braewoodsmy code hangs in the callback or something...
14:04:33braewoodsspeachy: i've got a mystery. maybe you know something i don't. so, i pass a context pointer to my zip callback... i've confirmed via printf logging that the pointers are exactly the same before i call it and during..
14:04:55braewoodsbut when i try to print the string buffer i wrote to, the contents are totally different once i'm in my callback.
14:05:18braewoodsit should say "/extract/" like it was earlier but it just says "zip" once in the callback
14:05:36braewoodsis it a coincidence that "zip" is the name i gave to core_alloc_ex?
14:05:43braewoodsyet wth is goingon
14:05:56braewoodsit's like my allocation got moved or something
14:06:41braewoodsit just don't make no sense.
14:07:57braewoodsis my code right now, the extract prototype i haven't finished yet
14:09:22braewoodsis my log output
14:10:02braewoodswth am i doing wrong? or is rockbox doing?
14:10:23braewoodsthis behavior just doesn't compute to me
14:10:28braewoodsbut it does explain my issue
14:10:34braewoodsjust not the why
14:10:40braewoodssomething is happening to my allocation.
14:17:50braewoodsi'll see if refetching the pointer changes anything
14:21:30braewoodsok it's not moving afaik...
14:21:55braewoodsonly one other possibility... the data was written to somehow between the two states.
14:28:14desowincryptomemory read commands do not comply with I2C (Atmel calls it TWI)
14:28:32desowinit is connected to the same lines as proper I2C devices in Sansa Connect
14:29:19desowinI2C is bitbanged so I would pretty much have to add new function in generic_i2c
14:30:06desowinquestion is, how should I name it? i2c_cryptomemory_read()?
14:56:16bluebrotherbraewoods: well, my idea was to factor out bootloader installation into separate binaries. That also has the advantage that we can run Rockbox Utility non-root, and then use pkcon / elevation for the bootloader installation. So Rockbox Utility wouldn't need to do any mtp stuff.
14:56:56braewoodsbluebrother: yea that's what i'm working towards.
14:56:58bluebrotherinstead we'd install the bootloader, then wait for the bootloader to come up in usb mode, and automatically figure the correct drive to use.
14:57:18braewoodsi actually had a different idea
14:57:24braewoodsslightly different
14:57:55braewoodswe can send the rockbox files as an archive payload before the bootloader and the bootloader can extract them on first boot
14:58:21bluebrotherI've started to look into how to split out the bootloader installation into separate binaries, but got stuck on the communication issue. Right now f.e. ipodpatcher functions are used to detect the correct disk. But those require sudo / elevation, so won't work anymore like it does now.
14:58:55bluebrotherof course we could start the separate ipodpatcher during detection, but then would need some communication.
14:59:22bluebrotherand similar issues for others. Had quite a lot of other stuff on my plate lately, so no real progress on this
14:59:24braewoodsbluebrother: ok, fair enough. i'm just looking at a solution for MTP only ports (initially)
14:59:43bluebrotherwell, beastpatcher could do that. First send the bootloader, then
14:59:46braewoodsi'm working on coding zip support for rockbox so it can extract the ZIP archive
14:59:49bluebrotherso nothing lost :)
15:00:01braewoodsso we can keep the first install simple
15:00:37braewoodsi found that it has support for TAR already but it's easier if i just code ZIP support for RB i think
15:00:43braewoodsso we don't need to repack anything
15:00:52bluebrotherthough I didn't deem it necessary to automatically install Rockbox in the first run. More user friendly though :)
15:01:11braewoodswell i decided to go this route since it has a lot of other use cases too
15:01:20braewoodsZIP support would open up new potential if i can get this working
15:01:28braewoodsright now tracing some weird issue
15:01:33bluebrotherright. Simple zip support shouldn't be too much of an issue. Though I remember issues from years back when an updated zip created slightly different archives.
15:02:47braewoodsi also plan to add decompression support but right now just trying to get the basic skeleton to work
15:02:54braewoodsi suspect i got a memory stomping bug
15:03:11braewoodsmy core_alloc allocations change contents somehow and the data hasn't been moved
15:03:59braewoodsok, changing the context pointer to a global changees nothing.
15:04:05braewoodssounds like an overflow
15:04:07braewoodsor something
15:04:29braewoodsguess i get to pick apart it and see when it changes
15:04:56braewoodschanges sometime after calling my read function
15:08:21***No seen item changed, no save performed.
15:08:30braewoodsunrolled my read function, time to test it out
15:28:27braewoodsSERIOUSLY? THAT'S THE ISSUE?
15:29:05braewoodsi didn't request enough memory. sizeof(zip_extract)
15:29:19braewoodsit's taking the size of my function address
15:29:24braewoodsi forgot to add struct to it
15:41:11braewoods g#3543 for whoever has time to review a trivial change
15:41:13rb-bluebotGerrit review #3543 at : zip: rename modts field to mtime by James Buren
15:47:48speachybluebrother: how does /rockbox/bot/utility sound for rb-bluebot?
15:48:08speachyfor a cloak, I mean
15:49:11blbro[m]Sounds good to me :)
15:50:54speachyok, request put in for that and the logbot
15:51:53blbro[m]Nice, thanks
15:59:15desowinspeachy: can you get me a cloak?
15:59:53*braewoods gives desowin the emperor's new clothes.
16:07:38speachydesowin: sure, /rockbox/developer/desowin ?
16:17:41 Quit rb-bluebot (Changing host)
16:17:41 Join rb-bluebot [0] (~rb-bluebo@rockbox/bot/utility)
16:17:53 Quit desowin (Changing host)
16:17:54 Join desowin [0] (~linux@rockbox/developer/desowin)
16:24:50speachyapparently I neglected to register rb-logbot with nickserv..
17:08:25***Saving seen data "./dancer.seen"
17:56:24 Quit ZincAlloy (Quit: Leaving.)
18:37:25 Join cockroach [0] (~blattodea@user/cockroach)
19:08:27***Saving seen data "./dancer.seen"
19:23:56 Quit wolfshappen (Quit: later)
19:25:09 Join wolfshappen [0] (
19:54:23braewoodsspeachy: were inline functions not a thing in early rockbox? i see a lot of macros that could be inline functions. they don't need the typeless nature of macros.
19:54:54braewoodsinstead they're using a GNU extension to simulate a function expression
19:55:03braewoods({ })
19:55:56speachycan't comment on the thinking of all of the baggage but for a while there one really couldn't trust gcc to inline things with any sense of coherency
19:56:29braewoodswould you be opposed if i replaced function macros with inline functions where they're a drop in replacement?
19:56:44speachyand if the stuff wasn't inlined the performace would be worse
19:56:57braewoodsstatic inline should still be honored
19:57:14braewoodseven at -Os
19:57:23braewoodsbut not quite sure
19:57:32speachyat -Os it'll most likely _not_ inline them actually
19:57:40braewoodsi see
19:57:53braewoodsi've seen GCC inline my static functions at times
19:58:03speachysome archs have relatively high overhead for function calls
19:58:16speachy(remember this whole thing started on sh)
19:58:24braewoodswhich we dropped
19:58:49braewoodsoddly pathfuncs.h already uses some inline functions
19:58:50speachyyep, but there's still a ton of code/structure that was optimized for the sh targets.
19:59:16speachygenerally I'd like to see some sort of tangible improvement to change stuff that's already working.
19:59:19braewoodsi think we should consider replacing some of this code with inline functions since those may be better optimized than a function macro
19:59:29braewoodseven so
19:59:38braewoodsi'll come back to this later
20:01:27braewoodsi see this macro defined in rockbox headers but nothing uses it
20:01:52braewoodsi wonder what it was added for
20:03:36speachygit blame?
20:04:12braewoodsspeachy: do you have any recommendations for how i should split paths for relative file paths? i need to split them so i can use mkdir on the components... except for the last part which may be a file
20:04:39braewoodsthe rockbox pathfuncs seem a bit confusing to use for this
20:05:17speachyI can't comment on the rb path functions, but typically one walks forward, finding the path separators, and using that to walk/create the directory tree
20:05:41braewoodsyea, i have constructed the full path and i just need to step through all but the last part.
20:05:55braewoodsi can null terminate them to allow mkdir to work before reverting to the original char
20:14:21 Quit wolfshappen (Quit: later)
20:46:19braewoodsspeachy: huh, i just realized. GOBBLE_PATH_SEPCH and GOBBLE_PATH_SEPCH are specialized versions of strspn and strcspn
21:08:30***Saving seen data "./dancer.seen"
21:25:07 Quit akaWolf (Ping timeout: 246 seconds)
21:27:03 Join akaWolf [0] (
21:59:57 Quit cockroach (Quit: leaving)
22:49:14 Quit MarcAndersen (Quit: I was using NightOwl 0.2.)
23:08:32***Saving seen data "./dancer.seen"

Previous day | Next day