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 2021-07-08

00:02:53 Quit sadjhflj333 (Ping timeout: 265 seconds)
00:27:01_bilgusbraewoods, why the addition of the utimebuf struct?
00:27:13braewoods_bilgus: see man utime
00:28:20braewoods_bilgus: i would review the previous request first since the utime one depends on it
00:31:01braewoodsi was just trying to follow a standard approach for it for maximum compatibility
00:32:23_bilgusaren't we targeting posix though thus far?
00:32:37braewoodsyes. utime used to be in posix but was dropped in 2008.
00:32:41braewoodseven so it tends to stick around.
00:33:07_bilgusposix will want a fd and a time struct as well
00:33:22braewoodsbut it uses filenames
00:33:36braewoodsonly the newer ones can use FDs
00:34:17_bilguseh you clearly are more aware of it than I
00:34:41braewoodswell we can make our own but they're not much mroe useful than utime
00:34:50braewoodswe don't support sub-second stuff really
00:35:13braewoods_bilgus: i'll look into it more, just look at my time oncversion
00:36:38braewoodsbut i can change it to allow different dirent stuff to be modded
00:36:48braewoodswould be more flexible
00:36:58braewoodsbut i can't see anything else worth modding in metadata
00:37:28braewoods_bilgus: yea it's weird. readlink doesn't work with FDs.
00:38:30rb-bluebotBuild Server message: New build round started. Revision 49ca4b3e5e, 297 builds, 8 clients.
00:47:48_bilguswhat about making it take two time struct pointers
00:48:11_bilgusatm its gonna be a pain to get into a plugin
00:49:01braewoods_bilgus: why? structs aren't part of the ABI per say. they can just include the header or so.
00:49:54braewoodsi mean if it bugs ya we can just write our own version
00:50:35_bilgusidk I'll have speachy look at it its surely more his forte
00:51:05braewoodsbut afaik all the standard time modifying APIs use structs
00:51:22braewoodsor an array of structs
00:51:38_bilgusit should match to the native stuff
00:51:47_bilgusthats most important
00:52:02braewoodswe can do that at the FAT layer but use something else for the generic api
00:52:09_bilgusand being they are usually linux
00:52:43_bilgusyeah another translation layer
00:52:52_bilgusand two more in lua
00:53:19braewoodsi just wanted this so my ZIP files can update timestamps after finishing their extraction
00:53:35braewoodswe could make an extension API that just wraps utime or so on real targets
00:53:45braewoodsaccess time is pretty irrelevant to us as it is
00:53:52braewoodsmodtime is the only one we really care about
00:57:16rb-bluebotBuild Server message: Build round completed after 1125 seconds.
00:57:18rb-bluebotBuild Server message: Revision 49ca4b3e5e result: All green
01:04:26braewoodsthe other variants of utime...
01:04:40braewoodsuse struct timeval or struct timespec
01:05:29braewoodsso afaik we would need a custom version if we wanted to use a simpler API or so.
01:05:57braewoodsthere's so many damn functions for changing these it's not even funny
01:06:09braewoodsutime, utimes
01:07:32***Saving seen data "./dancer.seen"
01:16:08_bilgusyeah I saw them all
02:20:24desowinbtw. has the same WiFI chip as sansa connect
03:07:36***No seen item changed, no save performed.
05:07:37***No seen item changed, no save performed.
05:36:26 Join ufdm_ [0] (
05:37:25 Quit ufdm (Ping timeout: 246 seconds)
05:38:30 Quit ufdm_ (Client Quit)
05:40:44 Quit spork (Ping timeout: 265 seconds)
05:45:54 Join spork [0] (
06:30:13 Join cockroach [0] (~blattodea@user/cockroach)
06:53:47 Quit cockroach (Quit: leaving)
07:07:41***Saving seen data "./dancer.seen"
07:34:49 Join ufdm [0] (
07:36:19 Join MarcAndersen [0] (~no_znepna@
07:37:25MarcAndersenHi. I now have a physical sansa connect in front of me, and a compiled bootloader and zsitool. Could someone please guide me through installing it so I don't brick it like my c200?
08:01:06MarcAndersenI am now compiling rockbox itself
08:07:48desowinSansa Connect can pretty much always be recovered as the OF bootloader is hardware write-protected
08:09:06desowinto start, enter recovery mode by completely powering down the device (press power button for 10 seconds)
08:09:26MarcAndersenAnd where was the power button, sorry.
08:09:42desowinat the top, left to the hold switch
08:10:19desowinit is round button
08:10:22MarcAndersenshould hold be closest to the edge or to the power button?
08:10:46desowinhold is closest to the edge, power button is in the middle
08:11:15MarcAndersenyes, but shoudl the switch be closest to the edge to make it unlocked?
08:11:33desowinclosest to the edge unlocks
08:13:21desowinto enter recovery mode, press and hold right button (above the wheel) together with volume up button (on the left side)
08:13:40desowinwhile holding the buttons, connect the device to USB
08:13:44MarcAndersenI try, but don't get any usb devices, maybe it's battery is dead?
08:14:21desowinthat's possible, if the battery is low it will shut down
08:14:50desowinif MTP does not come up, then it's probably charging
08:15:11desowinif MTP comes up, then it booted OF (and not recovery mode)
08:15:32MarcAndersennothing yet
08:16:01desowinwhen in recovery mode, lsusb should list 0781:7481 SanDisk Corp. Sansa Connect (in recovery mode)
08:16:19desowinbattery can take like 30 minutes to get the initial charge
08:17:02MarcAndersenI will give it some time then. It was a new one from ebay, he might have had it lie around for years. It is new though.
08:17:05desowinwhen the firmware (OF or Rockbox) is not running the charging is really slow
08:18:04MarcAndersenBut when I get it into recovery, which of all these files do I need to send to zsitool?
08:18:36MarcAndersenso ./zsitool -e bootloader.bin?
08:19:11MarcAndersenI'll report back when I get that far... Thanks. It's always fun to test new ports.
08:19:14desowinit takes a minute or two after zsitool says done for flashing to finish
08:19:38desowinthen you should see mass storage device with access to ext2 filesystem
08:19:58MarcAndersenok. what to put there?
08:20:02desowinto run rockbox you can either use microsd card or format the partition to fat
08:20:44MarcAndersenSo I format ext2 as fat and copy rockbox to that. That sounds easy.
08:22:06MarcAndersenIn an hour or so I will get someone who can see in here if it goes wrong.
08:22:09speachydesowin: g#3528 and g#3529
08:22:12rb-bluebotGerrit review #3528 at : builds: Move the Sansa Connect to unstable by Solomon Peachy
08:22:59MarcAndersenSo then the automatic build system will begin to build it next time?
08:23:50speachyMarcAndersen: it's already being built with every commit; this will add it to the stuff retained nightly, and generate voice files etc.
08:23:53MarcAndersenI can't wait to get this working.
08:25:38rb-bluebotBuild Server message: New build round started. Revision e6ee3dd17c, 297 builds, 9 clients.
08:26:20MarcAndersenWhat is the difference between bootloader.bin and bootloader-connect.sansa? I see it is 8 bytes longer?
08:26:41desowinheader, but it's not accepted by OF bootloader
08:26:49speachydesowin: just added it to the themesite too.
08:27:06desowinrockbox.sansa is the main application
08:27:15MarcAndersenI know
08:27:31MarcAndersenBut still no usb here
08:27:56MarcAndersenI'll give it some more time.
08:28:29desowinusb won't come up automatically, you can either press the power button or replug
08:28:37speachythe www change will land in ... 3 minutes. themesite will refresh on the quarter-hour after the build farm goes idle.
08:28:39MarcAndersenI just tried both
08:29:18MarcAndersenBut I think I can take it from here, thanks a lot!
08:29:21desowinif it boots linux, then it will takes some time before usb shows up
08:31:06desowinand speaking of new port, I wouldn't consider sansa connect port to be new
08:31:25desowinit just took me 10 years to exploit OF in software only way
08:31:37speachythanks to your work it's only now finally becoming usable :)
08:31:58desowin(maybe like 3 weeks over these 10 years, but still)
08:32:49MarcAndersenYes. Also, in which source files is the voice numbers concattenated? I am going to try to make it say numbers like one and twenty instead of twenty one in danish.
08:35:00speachyIgor Poretsky has some patches in his fork that rework how numbers are spoken
08:35:25MarcAndersenOk. I don't know if I am skilled enough since I mostly use autoit... But I will at least take a look on how it does it now.
08:36:42MarcAndersenHow exactly do I get patches? Sorry if this is a stupid question.
08:38:28speachy(ignore the duplicate please)
08:38:56MarcAndersenHow do I apply these?
08:40:48rb-bluebotBuild Server message: Build round completed after 910 seconds.
08:40:50rb-bluebotBuild Server message: Revision e6ee3dd17c result: All green
08:54:32speachyMarcAndersen: give me a few, I'm trying to rebase them against our current tree.
09:02:24MarcAndersenI still have no usb connection unfortunately
09:02:38MarcAndersenMaybe this device is dead...
09:03:20_bilgusI'd start checking cables
09:04:04MarcAndersenIt is this special sandisk cable that came with it. Like the 30 pin connector but not the same.
09:04:40MarcAndersenOh maybe the one from my c200 fits, just a second
09:04:45desowinit might be OF welcome screen, press right (above the wheel) like 10 times and then press upper part of the wheel
09:05:13_bilgusspeachy when you get a chance can you look over g#3527 Its over my head looks fine but would like your expert opinion
09:05:15rb-bluebotGerrit review #3527 at : file/fat: add utime function by James Buren
09:05:59MarcAndersenStill nothing
09:07:10MarcAndersenI'll have sighted assistance here in 10 minutes or so anyway, so we can find out what's wrong with this beast.
09:07:44***No seen item changed, no save performed.
09:08:13speachythere's a 'download' link on the right side below the "merge conflicts" pane, cut-n-paste the "cherry pick" instructions.
09:10:21speachyand the themesite refreshed, 67 themes seem to pass checkwps on the Connect
09:11:09 Join massiveH [0] (
09:11:31MarcAndersenI will look at this time and date thing after I get the thing to boot.
09:12:05speachyit'll appear on on the _next_ build.
09:12:25speachyand voice files will generate tonigh
09:12:27MarcAndersenSo it is merged into master now?
09:13:57speachy_bilgus: I think it looks okay?
09:14:33speachyMarcAndersen: Connect builds, I mean. that timedate patch is not, having only been compile tested on a single target so far
09:15:05speachy(well, it's been in Poretsky's tree for some time, and on my list to port/merge, but I need to at least _run_ it first..)
09:15:32rb-bluebotBuild Server message: New build round started. Revision c174d3a544, 297 builds, 9 clients.
09:16:53MarcAndersenI just got sighted assistance here. It says a
09:17:07speachyMarcAndersen: that patch probably doesn't do what you want, but it should indicate a path forward.
09:17:11MarcAndersensandisk, then low battery, and then it dies.
09:18:58_bilgusbattery is probably too discharged to safely charge
09:19:09desowinkeep it connected, don't mess with it
09:19:15desowinthe power button should be blue
09:19:28MarcAndersenIt is
09:19:32desowindo not press power button as every time you do so it discharges more
09:19:36speachyMarcAndersen: there isn't currently a mechanism to selectively reorder generic numeric voicing.
09:19:48desowinOF bootloader drains more current than gets from USB
09:20:13desowinso the only way to get initial charge is to pateintly wait
09:20:46speachyif anyone could comment on g#3530 I'd appreciate it. I combined two independent patches due to how they were related and stacked.
09:20:48rb-bluebotGerrit review #3530 at : Time and date format unification (Igor Poretsky) by Solomon Peachy
09:20:57_bilgusif a day of charge doesn't bring it back then you can try manually charging it use another liion battery and a resistor in a fire safe area
09:21:26MarcAndersenIs the battery replacable in this thing?
09:21:35desowinwhat is the serial number on the back?
09:21:41_bilgusyes just not easily
09:21:43desowinnext to FCC logo
09:23:24speachyMarcAndersen: would the "five and twenty" spoken ordering for Danish also apply to how years are voiced?
09:24:37MarcAndersenI think so, the only real difference is the ones first and then the tens, so two thousand one and twenty is fine, or maybe two thousand and one and twenty
09:25:20MarcAndersenSerial: UZB-ES20070309
09:26:37_bilgushe even added date and time to chessbox :)
09:26:38speachyrockbox's year voicing currently reads "2021" as "twenty twenty one" −− would it need to be "two thousand one and twenty" ?
09:26:59desowinMarcAndersen: this is FCC ID, the serial is painted above it (in very hard to read color)
09:27:35MarcAndersenI would say. Maybe rasher or someone else from Denmark can verify.
09:28:20speachyI'm curious about the expectations/norms of other languages too
09:30:58speachyand with that patch it becomes "twenty hundred twenty-one"
09:31:04rasherPronouncing numbers for different languages sounds like a hard problem.
09:31:11rb-bluebotBuild Server message: Build round completed after 939 seconds.
09:31:14rb-bluebotBuild Server message: Revision c174d3a544 result: 0 errors 343 warnings
09:31:29rasherIn Danish you'd either want "two thousand one-twenty" or "twenty one-twenty"
09:32:16speachybraewoods: ^^^ 343 warnings. :)
09:32:18MarcAndersenThat sounds ok. It's hard to say without listening to it
09:32:49speachyrasher: and that same convention is used for years?
09:32:52desowinMarcAndersen: it should have the old bootloader so once you get into recovery mode everything should go smoothly
09:33:38desowin(new one in refurbished units should work as well, but I just didn't get anyone to report)
09:34:45speachyI'm thinking it owuld be relatively simple to add a "one-and-twenty" vs "twenty-one" rule, but the question is if there are other quirks that need to be handled too.
09:35:17rasherspeachy: Actually that would be the convention for years. "twenty one-and-twenty" would be odd for a regular number
09:35:55MarcAndersenI always say two thousand one and twenty here
09:35:57rasherSo two-thousand one-and-twenty is the safest bet
09:36:27_bilgus speachy
09:37:15_bilgusoh nm
09:37:20_bilgus0 and '0'
09:37:25speachyso this patch's unifacion of "year" and "number" pronounciations is misguided.
09:38:31speachy_bilgus: I'd think fractions should always be spoken digit-by-digit, so we don't end up with 0.000512 as " point zero zero zero five hundred twelve "
09:38:56MarcAndersenthat's right. after the point everything is digits.
09:39:39MarcAndersenI'm going out for a few hours. Let's hope my sansa is awake when I come back...
09:42:30_bilguswanna a cheap car buy a diesel gate VW
09:44:09_bilgusbuddy has this nice turbo diesel bug I sure hope their rust proofing is good the metal work leaves things to be desired
09:44:43speachyI'm gonig to pull out the fractional fixes into a separate patch.
09:45:14_bilgusoops i'm in the wrong channel sorry
09:55:23rb-bluebotBuild Server message: New build round started. Revision a8fd590003, 297 builds, 9 clients.
09:55:46speachy_bilgus: g#3531 is the fractional rework
09:55:48rb-bluebotGerrit review #3531 at : talk: Always read out fractional values digit-by-digit by Solomon Peachy
09:59:53_bilgusis tbuf limited to numbers only? whats it do with malformed input '/'
09:59:58speachywait, that's what talk_spell already does. that patch is.. pointless.
10:04:25rb-bluebotBuild Server message: Build round completed after 540 seconds.
10:04:28rb-bluebotBuild Server message: Revision a8fd590003 result: 1989 errors 62 warnings
10:08:12rb-bluebotBuild Server message: New build round started. Revision 78eb87f453, 297 builds, 9 clients.
10:19:08braewoodsspeachy: wtf indeed?
10:19:42speachynext question. for voiced 24-hour format time, in English at least the convention is "21:00" is voiced as "twenty-one hundred hours" −− 3530 changes that to "twenty-one o'clock" which isn't correct for English, but what about non-English?
10:19:42_bilgusthe law of unintended consequence?
10:20:14speachybraewoods: the build going now is a reversion of my "fix", and a correct fix is queued.
10:20:36braewoodsspeachy: ah. i didn't know the fat header was even used on non-native ports
10:20:42_bilgusif you do oclock thats only for 12hr time then?
10:20:46braewoodsbecause why would it be?
10:20:49*braewoods boggles.
10:21:14speachyyeah, I tried nuking it as a fix but apparently the dirent structs are used in the file API.
10:21:37rb-bluebotBuild Server message: Build round completed after 805 seconds.
10:21:40rb-bluebotBuild Server message: Revision 78eb87f453 result: 0 errors 343 warnings
10:21:45rb-bluebotBuild Server message: New build round started. Revision 39b82afb45, 297 builds, 9 clients.
10:21:59braewoodswhat's the warning? utimbuf?
10:22:15braewoodsi can revamp my API choice if it'd help
10:22:16speachyyeah, that needs to comie from #including utime.h
10:22:46braewoodsyou can revert my PR and i'll work on a new version
10:22:49speachy_bilgus: o'clock is 12h only, at least for English.
10:22:52braewoodserr commit
10:23:07braewoodsi didn't realize utime would be this muchfuss
10:23:34speachythe current build running should go all green.
10:24:02braewoodsbilgus was wanting some rework done
10:24:15braewoodsi just wanted to wait until today since it was so late :P
10:24:37_bilgusso just pull head and start from there, stick to the prescribed interface will worry about the api later..
10:25:22 Quit massiveH (Quit: Leaving)
10:25:26braewoodsi had an alternative in time, our own "modtime" function or so.
10:25:35braewoodstakes a single time_t and either a path or FD.
10:25:37speachyso can any non-english speakers comment on 12/24h spoken time conventions
10:27:15_bilguswell you said posix pulled it so I'd stick to the native interface
10:27:32braewoodsi'll update it
10:27:53braewoodswe don't really use modtime
10:28:14_bilgusfor good reason..
10:30:02speachy0900 would be "oh-nine-hundred hours". 1923 would be "nineteen twenty three hours"
10:33:18braewoodsi'll work on this some more
10:33:47braewoodsi'll see if this works with an existing FD or not
10:34:02braewoodstruth is i don't know if the closing flush will overwrite my attempts to set the time
10:34:14braewoodsso that's one thing i'd need to consider
10:34:47rb-bluebotBuild Server message: Build round completed after 783 seconds.
10:34:49rb-bluebotBuild Server message: Revision 39b82afb45 result: 0 errors 10 warnings
10:37:20speachyok, those remaining warnings are in teh android port.
10:37:44braewoodsdon't worry about it speachy, i'll work on fixing this need for utimbuf or w/e
10:38:23braewoodswe can implement our own "modtime" for w/e else using the native APIs and don't need to emulate it for native
10:38:59braewoodsjust my first time messing with the FAT internals so not surprised there were some hiccups
10:40:39braewoodswhat do we plan to do when 2038 hits?
10:40:46braewoodsmost of our time_t are 32 bit signed
10:41:00braewoodsassuming rockbox is even still relevant then
10:41:33braewoodsoddly dos timestamps are pretty future proof up to around 2100 or so
10:49:35braewoodsok... basic redesign done. next up...
10:52:24braewoodstime to see if i can use an existing FD
10:54:16braewoodsi'll probably have to trigger a flush before doing my work
10:59:45rb-bluebotBuild Server message: New build round started. Revision a952cf52d5, 297 builds, 9 clients.
11:07:47***Saving seen data "./dancer.seen"
11:13:17rb-bluebotBuild Server message: Build round completed after 812 seconds.
11:13:22rb-bluebotBuild Server message: Revision a952cf52d5 result: 0 errors 10 warnings
11:47:56speachyhuh. the simulator builds fine with LTO.
11:54:12 Join amachronic [0] (~amachroni@user/amachronic)
12:01:45rb-bluebotBuild Server message: New build round started. Revision 0e1a90ea1d, 297 builds, 9 clients.
12:13:56desowinpcap format just switched to unsigned 32 bit which is good until 2106
12:14:24rb-bluebotBuild Server message: Build round completed after 759 seconds.
12:14:33rb-bluebotBuild Server message: Revision 0e1a90ea1d result: 0 errors 10 warnings
12:15:01rb-bluebotBuild Server message: New build round started. Revision 17f7dfb9d2, 297 builds, 9 clients.
12:17:10 Quit Romster (Ping timeout: 268 seconds)
12:21:39 Join Romster [0] (~romster@user/romster)
12:21:47sporksent you a more working link, amachronic
12:23:23amachronicthanks, got it.
12:26:36rb-bluebotBuild Server message: Build round completed after 695 seconds.
12:26:39rb-bluebotBuild Server message: Revision 17f7dfb9d2 result: 0 errors 10 warnings
12:27:29 Quit j-r (Ping timeout: 252 seconds)
12:27:51braewoodsspeachy, _bilgus: small problem in my tests. i managed to write a FD version of "modtime" but it doesn't work at all with write enabled FDs. it only works with read only ones. i have two options. i can add some kind of override inside of FAT to apply a different set of timestamps during the close process or just consider modtime as only a good idea during a separate file phase. in which case i'd
12:27:53braewoodsswitch back to taking a file path since it's pretty pointless to take a FD if you don't work on both kinds.
12:29:20braewoodsi'm leaning towards the latter due to how narrow the use case for this is. i don't think it's a good idea to introduce hacks into the FAT driver to accomodate this.
12:31:23braewoodsit's funny how it's a problem even when i don't write anything during closing
12:31:29braewoodserr before
12:31:37braewoodsi just open for writing
12:34:07desowin g#3535
12:34:09rb-bluebotGerrit review #3535 at : DM320: Regorganize LCD and TTB memory layout by Tomasz Moń
12:34:37desowinthis trades off the (MMU) 1 MiB alignment for 858 KiB larger audio buffer
12:34:58desowinas we don't do anything with MMU, I strongly favor larger audio buffer
12:37:11speachybraewoods: tailor it for the specific use case until we have a need for sometihng more general.
12:38:01braewoodsok. i'll just have it work as a separate open phase.
12:38:20braewoodsthat won't require me to modify how we flush data. i'm concerned i'd introduce new bugs by doing so.
12:41:44 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:f86b:93ab:cfff:61aa)
12:42:29 Quit Romster (Ping timeout: 255 seconds)
12:47:53 Join Romster [0] (~romster@user/romster)
12:49:36MarcAndersenThis stupid connect still says battery low
12:54:41desowinI had mine discharged for few years and then just plugging it for a day did the trick
12:55:05desowinthat is, get from bootloader to firmware and only then it actually charges properly
12:55:18MarcAndersenOk. Waiting time...
12:55:29desowinyou might want to not enter recovery mode before it is fully charged
12:55:45MarcAndersenThat sounds like a good idea
12:56:27desowinjust boot OF and leave it charging, if there's welcome screen just keep pressing right and then upper part of the wheel
12:56:44desowinOF menu scrolling should be audible
12:56:52braewoodsspeachy: did you see this?
12:56:58braewoodsif true oh boy.
12:57:19braewoodstheir software may be assisting in copyright or license infringement.
12:57:33MarcAndersenOnly sandisk, battery low, then loop forever
12:58:28desowinthat's still in bootloader, not even booting linux (which I call OF)
12:58:32speachyyep. I'll leave the folks with deeper pockets to take them to task.
12:58:58MarcAndersenOk. So I should just wait forever?
12:59:41desowinprobably leave it overnight and check tomorrow it is is any better
12:59:47braewoodsMarcAndersen: no idea. your unit's battery may be completely worn out. hard to say at this point.
12:59:47MarcAndersenWill do
13:00:15braewoodsi usually replace batteries in used electronics if i can, especially if i'm planning to keep them.
13:00:18desowinOF bootloader has a bug in it regarding the charging
13:00:40MarcAndersenWhich bug
13:00:55desowingetting out of the low battery is taking very long, then when in proper firmware it works
13:01:07desowinthat they mention in linux init scripts ;-)
13:01:24MarcAndersenMy ipod classic 6g and zen x-fi does something simular
13:02:44desowinalso if you get into recovery mode and leave it connected, it'll eventually drain the battery and get you to the low battery problem again
13:02:57 Join lebellium [0] (~lebellium@2a01:cb10:2e:2000:69ec:1efa:f053:e15)
13:02:58desowinso first step is just to try to boot OF and charge it
13:07:13rb-bluebotBuild Server message: New build round started. Revision a0f1236e88, 297 builds, 9 clients.
13:07:50***Saving seen data "./dancer.seen"
13:18:05 Quit amachronic (Ping timeout: 252 seconds)
13:18:52rb-bluebotBuild Server message: Build round completed after 699 seconds.
13:18:56rb-bluebotBuild Server message: Revision a0f1236e88 result: All green
13:33:51 Join sadjhflj333 [0] (~sadjhflj3@2607:fb90:17c7:aba2:ceab:9048:ad08:6407)
13:48:34 Quit sadjhflj333 (Quit: Konversation terminated!)
13:51:32braewoods g#3536
13:51:34rb-bluebotGerrit review #3536 at : file/fat: rework utime function as modtime extension by James Buren
13:51:49braewoodsThis is my reworking. It should be a minor improvement and good enough for now.
14:20:07braewoods_bilgus: you mentioned wanting to use this for themes. did you have any ideas for how that would work?
14:21:10braewoodsif all the theme files are loaded into RAM, we could just load from the ZIP instead of extracting them.
14:22:01braewoodsfor now i'm just going to focus on my use case
14:22:22braewoodsbut there's nothing stopping us from adding custom loaders that use this to load files directly into RAM from an archive
14:23:10 Quit desowin (Ping timeout: 272 seconds)
14:23:54 Join j-r [0] (
15:07:51***Saving seen data "./dancer.seen"
15:57:18 Join LiteralBrainlet [0] (
15:57:28LiteralBrainlethowdy fellas
15:57:45LiteralBrainletI have a problem installing RB on my iPod Classic 6gen
15:58:31LiteralBrainletI get stuck on
15:58:49LiteralBrainlet"action required press and hold select + menu"
15:59:50LiteralBrainletIn the past I successfully got the software on the iPod manually using a friend's mac, but even after hard resetting the device I'm still stuck with Apple's OS
16:00:14LiteralBrainletA1238 - 160gb version
16:08:00braewoodsLiteralBrainlet: so what does that come after?
16:45:39LiteralBrainletoh fuck disregard, after 2 years of attempts I finally got it to work
16:46:22LiteralBrainletAlthough, whenever I plug the device into my PC Windows prompts me to format it
16:46:56braewoodserm. you usually need to format iPods with windows itunes so it'll be FAT based.
17:07:44LiteralBrainletI did. I originally formatted it as FAT32, but now it keeps telling me to do it again. I don't think it matters because I can still move music files onto the iPod
17:07:55***No seen item changed, no save performed.
17:34:25 Quit lebellium (Quit: Leaving)
17:45:03 Quit ZincAlloy (Quit: Leaving.)
17:53:34speachysweet. Got the X3 compiling cleanly with LTO!
17:54:11 Quit LiteralBrainlet (Quit: Client closed)
17:56:28speachyjust the main binary, plugins are shot.
17:59:13 Join cockroach [0] (~blattodea@user/cockroach)
18:35:32_bilgusbraewoods, for the time being its fine just to unpack to a directory we can integrate later
18:43:30braewoods_bilgus: ok. i'm just waiting for my next review.
18:45:42 Quit cockroach (Quit: leaving)
18:45:59 Join cockroach [0] (~blattodea@user/cockroach)
18:54:00_bilgusi'm looking at it but busy atm
19:03:59rb-bluebotBuild Server message: New build round started. Revision a9f36efa62, 297 builds, 8 clients.
19:05:34_bilgusthat'll be much easier since its custom you might wanna add flags to hit atime or at least have the flag in the proto
19:07:59***Saving seen data "./dancer.seen"
19:21:05rb-bluebotBuild Server message: Build round completed after 1025 seconds.
19:21:08rb-bluebotBuild Server message: Revision a9f36efa62 result: All green
19:54:22rb-bluebotBuild Server message: New build round started. Revision 2c984f3714, 297 builds, 8 clients.
20:09:12rb-bluebotBuild Server message: Build round completed after 891 seconds.
20:09:17rb-bluebotBuild Server message: Revision 2c984f3714 result: All green
20:37:33braewoodsspeachy: isn't it a bit early for LTO?
20:50:32speachythe core is solid. the X3 main binary shrinks by ~4%.
20:50:50speachyI have some codecs and plugins building, but others are going boom for a reason I don't yet understand.
20:55:01_bilgushow does that work with plugins and codecs sharing the same address space?
20:55:36speachyit's just fancy compile-time crap; it still has to confirm to the appropriate link map
20:57:02_bilgusI'm surprised it works at all tbh
20:57:05speachywhat's going on (I think) is that the LTO is stripping out the codeclib's memcpy()
20:57:19speachybecause we force -nostdlib etc
20:57:51speachywhat surprised me is that LTO on the main (X3) binary only generated a single warning.
21:03:44rb-bluebotBuild Server message: New build round started. Revision 2e9b93dc5d, 297 builds, 8 clients.
21:08:00***No seen item changed, no save performed.
21:16:04rb-bluebotBuild Server message: Build round completed after 740 seconds.
21:16:06rb-bluebotBuild Server message: Revision 2e9b93dc5d result: All green
21:16:09braewoodsspeachy: what does rockbox do with trailing slashes in path?
21:16:24braewoods/a/b/c the same as /a/b/c/?
21:16:51speachyif it's a directory, I don't think it matters. but that trailing slash is an illegal filename.
21:19:05speachybut, as always, use the source, luke..
21:41:05braewoodsstrange seeing a function return ssize_t when it can't return negatives by design
21:44:00_bilgussome of those are because of sticking to prototypes others might be erroneous
21:45:14braewoodsit's kinda moot anyway since unless we deal with RAM above 2GB we'd never hit it
21:47:02braewoodsfound a small optimization for my code for path construction
21:47:27braewoodsby writing the root directory only once i can reduce some extraneous writes, saving a bit of time
22:14:11 Quit cockroach (Quit: leaving)
23:08:03***Saving seen data "./dancer.seen"
23:19:14 Quit F3l1x_10m (Read error: Connection reset by peer)
23:35:17 Join F3l1x_10m [0] (~Al3x_10m@user/f3l1x-10m/x-3393542)

Previous day | Next day