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 2021-06-21

00:08:31 Join speachy1 [0] (~speachy@
00:09:06 Join akaWolf1 [0] (
00:09:36 Quit akaWolf (*.net *.split)
00:09:36 Quit speachy (*.net *.split)
00:16:54 Quit fmlatghor (Remote host closed the connection)
00:17:18 Join fmlatghor [0] (~lcoogan@2601:5cd:8100:2890:9220:3aff:fe1a:350d)
01:34:10***Saving seen data "./dancer.seen"
01:51:57 Quit markun (Ping timeout: 268 seconds)
01:57:05 Join markun [0] (
02:12:03 Nick akaWolf1 is now known as akaWolf (
03:34:12***Saving seen data "./dancer.seen"
03:37:49braewoodsspeachy1: i just realized that bitwise XOR with -1 (all bits on) produces the same logical output as bitwise NOT. so why do people prefer XOR over NOT for this effect? i see that in some crc algorithms as the final step.
03:56:52 Join TheLemonMan [0] (~lemonboy@irssi/staff/TheLemonMan)
05:10:20 Join advcomp2019 [0] (~advcomp20@user/advcomp2019)
05:25:04 Quit advcomp2019 (Ping timeout: 250 seconds)
05:34:15***Saving seen data "./dancer.seen"
05:41:08braewoods_bilgus: g#3491
05:41:10rb-bluebotGerrit review #3491 at : mi4: replace chksum_crc32 with crc_32r by James Buren
05:41:31braewoodsi'll test it on my iriver h10 but that's my only PP unit right now i can easily get to
05:59:51 Quit akaWolf (Ping timeout: 268 seconds)
06:02:23 Join akaWolf [0] (
06:04:00braewoods_bilgus: ... it's back. the h10 is freezing again on latest dev stuff. unrelated to my changes too.
06:06:21braewoodstesting the 5gb model
06:09:03braewoodsrofl the 5gb model is immune?
06:28:01braewoodsweird. the 5gb bootloader made from git doesn't even boot. the OF boot rom thinks its corrupted or so.
06:28:18braewoodsunrelated since latest master w/o my commits is same end result
06:55:09braewoodsok found the problem
06:55:25braewoodsturns out I need the MTP mi4 container which isn't mentioned any where
06:55:43braewoodsso the crc32 change appears to be working correctly
06:55:48braewoodsas my preliminary tests showed
06:55:58 Quit speachy1 (Ping timeout: 268 seconds)
07:02:43 Join speachy [0] (~speachy@
07:02:43Mode"#rockbox +v speachy" by ChanServ (
07:07:12braewoodsspeachy: wb
07:07:21speachyUPS died. :/
07:07:35braewoodsspeachy: care to review my changes?
07:07:36speachyrefuses to power up, batteries are good.
07:07:42speachycan't for a few hours
07:07:58braewoodssounds like it could be a hardware failure then
07:08:46braewoodstldr; i staged 2 commits for implementing a space optimized version of the mi4 checksum algorithm
07:10:07braewoodsalready tested it on the h10 5gb
07:10:22braewoodsworks with both the new/old bootloader and new/old main; only used during booting anyway
07:10:46braewoodsit's a bit slower but it's only used when booting so probably not that big of a deal
07:12:55 Quit TheLemonMan (Quit: "It's now safe to turn off your computer.")
07:29:22speachyLi-Ion UPSes command a 7x price premium still. sigh.
07:31:48braewoodsspeachy: what kind of UPS broke?
07:33:25speachytripp-lite smart1500lcd to be precise
07:33:54speachy4 yr old or so. been through one battery replacement already.
07:34:01braewoods$227 on amazon
07:34:17***Saving seen data "./dancer.seen"
07:34:28braewoodsyou were looking for something not based on SLA?
07:35:02speachycurious about pricing alternatives
07:35:07speachybut yeah
07:35:32braewoodsi've heard some people modded their SLA UPS to take a lifepo4 pack
07:35:59braewoodsonly problem is lifepo4 packs usually have low discharge amps compared to SLA
07:36:18braewoodsat least at the sizes UPS often use
07:36:20speachyafter two different battery explosions due to not using stock battery chemistries I'm leery of doing such things.
07:36:57braewoodsit is an option. some people have managed to use lifepo4 for a drop in replacement.
07:37:19braewoodsbut i'd not do that unless i knew everything about the design of the battery it expects to have
07:37:33braewoodsbut if it does CC/CV only then it may be compatible
07:37:36braewoodsfor charging
07:37:45braewoodsas for discharge...
07:37:59braewoods12V needs a lot of amps to work
07:38:03speachythe limiting factor is most likely the UPS's inverter anyway.
07:38:07braewoodsi expect high end UPS would use higher voltage
07:38:24braewoods12V DC would need a shit ton of amps to support 500+ W
07:38:55speachy37.5A max (@24V) on this thing, for example (assuming 100% efficiency)
07:39:24braewoods4S lifepo4 is fairl compatible with SLA voltage range
07:39:27braewoodsat 12V
07:39:39braewoodsbut you're dealing with 24V...
07:39:50braewoodsmy 7S lithium pack is 29.4V peak
07:40:00braewoodsaround 21V at the low end
07:40:44braewoodsi suspect loads don't care about the battery chemistry as long as it's within its voltage range tolerances and can handle the discharge rate of the load
07:41:05braewoodsthe main problem is probably charging which does have to care about the chemistry
07:41:19speachyyep, charging is the real key.
07:41:54braewoodsi'm working on installing a DC online UPS for my networking gear
07:41:58speachywell, figuring out the remaining capacity too −− LiFePO4 has a _very_ flat curve. looks great until it's nearly completely dead
07:42:10braewoodsoh, right. it won't know the load.
07:42:19braewoodsso it'll throw off any instruments and anything relying on that data.
07:42:56braewoodsi imported a proline modular UPS that lets you connect most any SLA battery to it
07:43:01braewoodsit only does CC/CV type charging
07:43:04speachythe UPS I had explode on me (twice) was set up with some deep cycle marine wet cell batteries; I spec'd it for ~5 hours of runtime.
07:43:22speachybut the charging characteristics ended up cooking the cells after about 2 years.
07:43:38braewoodsi see.
07:43:47speachyafter the second such bang I decided it wasn't worth the risk again.
07:43:50braewoodsdon't UPS normally take regular SLA batteries?
07:44:01braewoodsnot those special purpose built ones
07:44:39speachyIf I really wanted to go nuts I could use some 4-cell golf cart batteries. :D
07:44:47speachythose things are built for abuse
07:44:51speachybut $$$
07:45:22braewoodsi'm not sure if it's wise but i was considering using my DC UPS on a 4S lifepo4 battery pack
07:45:23speachyanyway. this is way OT, and not even high on the priority list for the day.
08:42:47rb-bluebotBuild Server message: New build round started. Revision 581081a3df, 297 builds, 8 clients.
08:44:21braewoods_bilgus: thanks
08:44:42braewoods_bilgus: care to test how much free space you now have on the c200 and e200?
08:44:46_bilgusnp ill stay long enough to make sure it sticks
08:45:05braewoodsor i can calculate it if you know the space limit
08:45:17_bilgusnot at my dev machine
08:45:47braewoodswhich reminds me, i'm replacing my old dev machine with a newer system... first upgrade of it in 5 years
08:46:04braewoodsdouble the core count and quadruple the RAM
08:46:14_bilgusI went to the micro pcs really liking it
08:46:29braewoodsah, my dev machine is also my server NAS and desktop
08:46:43braewoodswhy i still use a full size tower
08:47:21_bilgusnew micro pc is specd like my giant one just in 10% of the space
08:47:46braewoodsyea that works too but i really enjoy the power I got in this one
08:48:13braewoodsit will be interesting to see how well this new ATX12VO psu standard works out
08:48:26braewoodsit'll be a drastic departure from the old ATX systems
08:48:27_bilgusyeah I guess as I get older the clutter becomes annoying
08:49:01braewoodsit gets rid of the 3.3V and 5V rails and replaces it all with 12V
08:49:04_bilgusthe hp biz machines already do as a general rule
08:49:42_bilgusI spun up boards to convert so I could use real power supplies
08:50:04braewoods_bilgus: how much barrel jack power does the mini pc use?
08:50:07braewoods19V i assume?
08:50:29_bilgusits lenova so same as the laptop bricks
08:50:35braewoodsso 19V.
08:50:51_bilgussounds right
08:50:54braewoodsso they basically took the laptop model and applied it to a mini pc
08:51:27_bilgusyeah infact the mini pcs are like laptop boards w/ no monitor
08:51:45braewoodsif you need a UPS for it there's a few options for DC only power; they have no inverters
08:52:10braewoodsjust kinda interesting if nothing else
08:52:43braewoodsi saw one on alibaba or so that provides 19V regulated output; uses lifepo4 cells
08:52:51braewoodsnot much runtime but interesting in any case
08:53:00_bilgusmy power has gone out here 1x in 10 yrs
08:53:08braewoodsah i see
08:53:15_bilgusthe ups's die before they get an event
08:53:42_bilgusI wanna say that 1x was a brown out
08:53:53braewoodsthe desktop PC is evolving
08:54:07braewoodsmost new cases have no front side drive bays
08:54:38braewoods3.5" internal is being phased out in favor of other stuff
08:54:44_bilgusI thought I'd miss cd drives or dvd but no
08:55:04braewoodswell i still miss the front side drive bays; i did manage to find a case finally
08:55:39braewoodsbut i'm not using mine for optical drives
08:55:55braewoodsone is a status LCD connected over USB
08:56:08braewoodsthe other is a front panel USB hub
08:56:17_bilgusmy old dev machine has like 5 disks it makes 5tb of storage far more feasible than the mini pc 2 drive bays + m2
08:58:44_bilgus2.5" drive bays I think I found 3tb drives but they are $$ and likely less reliable
09:02:49braewoodsdo these build rounds normally take this long?
09:03:19_bilgusdepends whos on
09:03:42braewoodsi see
09:04:33braewoods_bilgus: do you have any recommendations for converting car voltage range to a regulated 12V?
09:04:45braewoods10.8 to 13.8 volts here
09:04:55braewoodsthough it'll be at 13.8V most of the time
09:05:12_bilgusthe under voltage is the one that makes it harder
09:05:21_bilgusso you need buck and boost
09:05:31braewoodsindeed since the voltage i need is in the middle
09:05:54_bilgusthose ebay modules are fine depending on the amps
09:06:01rb-bluebotBuild Server message: Build round completed after 1394 seconds.
09:06:04rb-bluebotBuild Server message: Revision 581081a3df result: All green
09:06:07_bilgusconsider it half the listing
09:06:59braewoodsyea, the proline i bought can handle uh..
09:07:17braewoodsit provides 10.8V to 13.8V but is always at 13.8V when AC is connected
09:07:22braewoodsthe range is when the battery kicks in
09:07:49_bilguswell tbh the most regulated dc you will ever have is another battery
09:08:09_bilgusif you need big amps you be better off that way
09:08:29braewoodswell the max load amps is 6.5Aish at 13.8V
09:08:33_bilgusbut 10A or < then buck boost
09:09:02braewoodsmore than enough
09:09:41braewoodsgiven the problems with barrel jack connectors i'm planning to distribute my heaviest load to one regulator and the lighter 2 to a different one
09:09:58braewoodsthe proline i have provides 2 DC barrel jacks
09:10:13_bilgusor switch to the connectors made for rc cars
09:10:42braewoodswell it's more the fact that barrel jacks don't really advertise their amp loads well
09:11:01braewoodsif i could find a buck boost inline converter i'd use that but
09:11:06braewoodsthey're usually buck or boost only
09:11:23braewoodsinstead i have to buy one and wire it up for my purposes myself
09:11:59braewoodsin any case this ended up being more cost effective for me to replace my crappy UPS for my networking gear
09:12:06_bilgusIts not a complicated circuit but there are ready made ones
09:12:50_bilgushave run into one that was buck only even though they claimed both
09:13:09_bilguswell 5 of them but they eventually got used
09:19:15_bilgusbraewoods, it was double because of the stacked commits, duh
09:19:33braewoods_bilgus: lol
09:19:36_bilgusin hindsight
09:19:50braewoods_bilgus: well, i'd already tested it to minimize the chance of random failure
09:20:06braewoodsside note, i discovered the H10 is freezing again during boot
09:20:08braewoodsnot sure why this time
09:20:42_bilgusIdk whats changed since then but ugh.
09:20:46braewoodsthe H10 20GB has got to be the most fickle rockbox port i've ever used
09:21:07braewoodsyou want to take another look at it?
09:21:19_bilgusno time atm
09:21:31braewoodsi'll let speachy know it's not booting for now
09:21:47braewoodsoddly the 5GB model works just fine
09:21:59_bilgusI assume you updated the fw?
09:22:06braewoodsthe OF firmware?
09:22:16_bilgusno rb
09:22:28braewoodsyes, i tested the latest dev build and it froze during boot
09:22:45braewoodsi'll look into it more later
09:22:58_bilgusah ok so some commit broke it
09:22:58braewoodsi ended up using the 5GB model to test my crc changes
09:23:35braewoodsthanks for your help merging it; i can now return to working on my rockzip module
09:23:43_bilgusthere can't be many that actually touched h20 specific code
09:24:07braewoodsi just needed a viable crc32 algorithm and this solves that problem
09:24:33braewoodsi'm trying to make rockzip as lightweight as possible so we can use it on even low ram targets (in theory)
09:24:39_bilgusnp I can still read code without my dev machine
09:24:54braewoodsmain problem is i need at least 128KB of memory for buffer space
09:25:00braewoodsdue to how ZIP is designed
09:25:26braewoodsbut that is best dynamically allocated so it doesn't need to be permanently reserved
09:25:29_bilguskill the playback buffer
09:25:51braewoodsyea, i decided to make allocating the buffer up to the user of the library
09:26:00braewoodsso i don't need to care how it is allocated in my code
09:26:18braewoodsthat way i can use a static allocation in the bootloader and dynamic in rockbox itself
09:26:32_bilgusin the case of a plugin itd also depend on bin size
09:27:09braewoodsi'm just optimizing for size since i expect rockzip to become part of the firmware
09:27:16_bilgusyou might be able to satisify both with only the plugin buffer
09:27:48_bilgusitd be nice to enable theme files to be loaded as zip
09:28:04braewoodswell for now i'm just implementing basic zip file support
09:28:12braewoodshow it gets used is a matter of creativity
09:28:36braewoodsmy library will handle everything except what to do with the files and data payload
09:28:56braewoodsi plan to support only the basic 2 storage methods; STORE and DEFLATE
09:29:08_bilgusis pass protect part of that spec?
09:29:23braewoodsmaybe but i don't plan to add support for it because it's not really necessary for what we do
09:29:56braewoodswhy would we need support for encrypted ZIP files?
09:30:32braewoodsthe only two parts that seem necessary to use a third party library for is the crc32 algorith and DEFLATE
09:30:37_bilgusnothing in particular just curious
09:31:05braewoodsif you want to look into other uses for rockzip, be my guest. i'm desinging it for a port specific problem.
09:31:19braewoodsbut i conceived of using it to extract rockbox archive payloads and such sent over MTP or UMS
09:31:42_bilguswell assuming it can unzip the then it'd make user stuff easier
09:32:02_bilgusand builds could be installed like that too
09:32:04braewoodsit'll be able to extract them to the internal storage
09:32:09braewoodsor w/e you decide to do with it
09:32:21braewoodsmaybe you can just load them into RAM from the archive
09:32:36braewoodsinitially i'm only going to have a basic function that extracts to internal storage
09:32:50_bilguswith ROLO for sure to internal but the themes I hope can just load from ram
09:33:09braewoodsplus the option of processing the ZIP file from a RAM buffer or from a file descriptor
09:33:20_bilguswe already cache theme files
09:34:03braewoodsthe advantage of using ZIP is we'll be able to leverage the compressed data to reduce RAM requirements for loading the archive into RAM
09:34:21***No seen item changed, no save performed.
09:35:20_bilgusit still has to expand though to use it so how is that an advantage
09:35:47braewoods_bilgus: it is if you're planning to write it to disk
09:35:59braewoodsbut in either case
09:36:08_bilgusoh you mean while creating an archive?
09:36:25braewoodswell i'm only planning to support read-only access
09:36:32braewoodsat least initially
09:36:50braewoodsi'm stripping it down to a minimal subset so the requirements won't overwhelm rockbox
09:37:02_bilgusI'm not understanding how the zip file gets in ram then?
09:37:25braewoods_bilgus: someone reads the archive into RAM. they process the file contents in RAM before writing to disk.
09:37:34braewoodsby reading the archive into RAM they avoid the double-io penalty
09:37:45braewoodsfrom trying to read/write at the same time
09:38:00_bilgusah ok
09:40:24_bilgusI suppose with flash storage at least you would really want nice long blocks too
09:40:53braewoods_bilgus: and zip files also don't really lend themselves well to sequential access all the time
09:41:13braewoodsto parse them fully you end up jumping around a lot.
09:41:29_bilgusyeah i'm not not overly familar clearly
09:41:32braewoodsthe data payload is stored in a different region from the basic directory payload
09:42:01braewoodsyou first look for the end of directory record at the end of the file
09:42:19braewoodsonce you find it you find the offset of the first regular file entry
09:42:36braewoodsit also details how many exist in this contiguous region
09:42:50braewoodsthis is enough if you just want to iterate over the metadata
09:43:09braewoodsbut if you want to process the file data, you need to jump to wherever it stored it as indicated by each entry's LF offset field
09:43:22braewoodsit's usually stored at the start of the archive
09:43:31_bilgusare you planning to add that to the file browser?
09:43:42braewoodsmaybe, but not right now.
09:43:50braewoodsVFS support seems a bit further along.
09:44:12braewoodsbut it's not impossible to add
09:44:35_bilguswouldn't need to do that just make a file/dir list
09:44:46braewoodsi see
09:44:47_bilgusthen you unzip
09:44:57braewoodsso support for extraction from the UI or so
09:45:36braewoodswell i'll finish my first use case and we'll see where that leaves us
09:45:39braewoodsone thing at a time and all
09:46:25braewoodswe can also experiment with faster crc32 if zip support is too slow or so
09:46:32braewoodsor even disabling crc checks
09:47:03braewoodsbut for now i'm just going to turn on all my validation checks
09:47:24_bilgusyeah make it right first :)
10:03:03braewoods_bilgus: which also reminds me, by having a single source for crc32 we could also swap in a faster version if we don't mind the waste
10:03:13braewoodsmake the choice at compile time or something
10:03:44braewoodsbut that's for later if we deem it useful
10:03:58braewoodswe don't use these much as it is so a faster version won't do us much good
10:04:42braewoodsin theory we could also use these to replace code in some of the apps too
10:05:24braewoods_bilgus: i'll give you more stuff as i work on this rockzip code.
10:05:36braewoodsfor now i'm developing it for my desktop target so i can easily test my code
10:05:47braewoodsjust using designs that will easily translate to rockbox
10:06:46_bilguswell with plugins just have them carry it
10:07:33braewoodsi guess if they need a speed optimized version that would probably be best
10:07:38braewoodsrockbox core doesn't need one
10:07:55_bilgusand its a mp3 player lol
10:08:14_bilgusjust zip support is amazing
10:08:46braewoodsi'm working on this because i need at least some ZIP support to make delivery of the first installation of rockbox on gigabeat S smoother
10:09:03braewoodsso i thought i should make it something reusable while i'm at it
10:09:40braewoodsbeing able to upload the rockbox files right before the bootloader payloader would make it much easier
10:10:30 Join TheLemonMan [0] (~lemonboy@irssi/staff/TheLemonMan)
10:11:00 Join advcomp2019 [0] (~advcomp20@user/advcomp2019)
10:58:58 Quit TheLemonMan (Quit: "It's now safe to turn off your computer.")
11:34:24***Saving seen data "./dancer.seen"
11:49:06 Join lebellium [0] (~lebellium@2a01:cb10:2e:2000:1d19:cbf1:d0a:86e3)
11:56:59 Quit lebellium (Ping timeout: 252 seconds)
12:05:38 Nick Ckat is now known as cate (~Ckat@xn--z7x.xn--6frz82g)
12:06:03 Nick cate is now known as Ckat (~Ckat@xn--z7x.xn--6frz82g)
12:50:10 Quit advcomp2019 (Ping timeout: 252 seconds)
13:15:06 Quit akaWolf (Ping timeout: 252 seconds)
13:17:55 Join TheLemonMan [0] (~lemonboy@irssi/staff/TheLemonMan)
13:21:58 Join akaWolf [0] (
13:34:25***Saving seen data "./dancer.seen"
13:49:12 Quit speachy (Ping timeout: 252 seconds)
13:58:44 Quit TheLemonMan (Quit: "It's now safe to turn off your computer.")
14:11:31 Join speachy [0] (~speachy@
14:11:31Mode"#rockbox +v speachy" by ChanServ (
14:28:20_bilgusbraewoods, I assume you tested g#3316 on the H20 can we assume thats a known good point?
14:28:22rb-bluebotGerrit review #3316 at : pp: fix adc mistake where base 10 was used instead of base 16 by James Buren
14:54:51braewoods_bilgus: probably.
14:55:07braewoodsonly one way to know; it's been awhile
15:14:51 Quit akaWolf (Ping timeout: 268 seconds)
15:17:20 Join akaWolf [0] (
15:19:14 Quit fmlatghor (Remote host closed the connection)
15:19:34 Join fmlatghor [0] (~lcoogan@2601:5cd:8100:2890:9220:3aff:fe1a:350d)
15:34:26***Saving seen data "./dancer.seen"
16:19:29 Join ufdm_ [0] (
16:20:38 Quit ufdm (Ping timeout: 252 seconds)
16:57:56 Join petur [0] (
17:34:30***Saving seen data "./dancer.seen"
18:26:18 Quit petur (Remote host closed the connection)
19:09:59 Nick ParkerR_ is now known as ParkerR (
19:13:30 Quit tchan (Ping timeout: 240 seconds)
19:13:36 Quit TorC (Ping timeout: 268 seconds)
19:14:14 Join TorC [0] (~Tor@fsf/member/TorC)
19:14:24 Join tchan [0] (
19:34:33***Saving seen data "./dancer.seen"
20:10:23 Quit akaWolf (Ping timeout: 265 seconds)
20:18:40 Join akaWolf [0] (
20:33:55 Quit SammysHP (Quit: *wuff*)
20:34:14 Join SammysHP [0] (
21:04:29 Quit ufdm_ (Quit: Leaving)
21:34:34***Saving seen data "./dancer.seen"
21:40:59 Join ufdm [0] (
21:56:33 Quit fmlatghor (Remote host closed the connection)
21:56:56 Join fmlatghor [0] (~lcoogan@2601:5cd:8100:2890:9220:3aff:fe1a:350d)
23:09:25 Join advcomp2019 [0] (~advcomp20@user/advcomp2019)
23:34:05 Join advcomp2019_ [0] (~advcomp20@user/advcomp2019)
23:34:40***Saving seen data "./dancer.seen"
23:36:17 Quit advcomp2019 (Ping timeout: 265 seconds)
23:37:24__builtinso I've had commit access on this project for ~5 years at this point... and I *just* found out what we have working USB serial support :/
23:56:23__builtinanyway, I need some ideas: I've got a heisenbug that's manifesting as a mysterious yet consistent memory corruption of a particular variable
23:56:37__builtin(only on hardware, of course...)
23:57:18__builtincan you all think of any tricks for quickly tracking this down?

Previous day | Next day