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 2011-02-02

00:03:55 Join FBI_Guy [0] (
00:05:48 Quit FBI_Guy (Client Quit)
00:06:30 Join mebus [0] (
00:06:53mebuscan I somehow leave the multimedia and keypad mode and stuff on the ipod when connected via usb and go back to normal rockbox?
00:07:26 Quit dfkt (Quit: -= SysReset 2.53=- Sic gorgiamus allos subjectatos nunc.)
00:07:37JdGordon1mebus: RTFM.... hold menu while connecting usb
00:07:47mebusah ok
00:08:08mebusbut this would mean I have to reconnect but am charging a dead battery :-(
00:12:12 Quit komputes (Remote host closed the connection)
00:17:20pixelmaJdGordon1: I don't get your response here at all (including the locking),25169.msg177526.html#msg177526
00:17:56 Quit wodz (Remote host closed the connection)
00:18:55 Join PaulJam [0] (
00:21:54 Join FBI_Guy [0] (
00:21:55 Quit FBI_Guy (Client Quit)
00:25:00 Quit GeekShad0w (Read error: Connection reset by peer)
00:25:20 Join GeekShad0w [0] (
00:26:03*foolsh still stands by FS #11903 as ready for serious testing ;)
00:28:08 Nick scorche` is now known as scorche (~scorche@rockbox/administrator/scorche)
00:31:51 Quit Judas_PhD (Quit: This is a quitting message)
00:39:54 Quit amiconn (Remote host closed the connection)
00:39:54 Quit pixelma (Read error: Connection reset by peer)
00:40:10 Join pixelma [0] (quassel@rockbox/staff/pixelma)
00:40:10 Join amiconn [0] (quassel@rockbox/developer/amiconn)
00:45:10JdGordon1arg, so I just found the "enable automatic resume" menu item.... thats really not a nice place for it
00:45:29JdGordon1sideral: ^ (I know, its post commit, but it should be made consistant with the rest of the menus)
00:46:21sideralsure. where do you think would be a better place for it?
00:46:46 Join krazykit [0] (
00:48:35sideralJdGordon: BTW, this will eventually move into a submenu
00:48:56JdGordon1for the furute, anything that affects tha languages should be finalised before being commited
00:49:02sideralThat menu should sit next to the bookmarking menu, I thought
00:49:04JdGordon1moving/chaning things after the fact causes problems
00:49:31JdGordon1anyway, I would have expected it under playlist or playback options and named "automatic resume" (i.e without enable)
00:50:36JdGordon1there is possibly an argument to be amde for a "playback resumption" submenu with all the relevant settings ("rewind before resume", "auto resume", etc)
00:52:39sideralI proposed "automatic resume" w/o "enable" to be the name of the menu, and "enable automatic resume" to be its first entry. See last comment of FS #11748
00:53:21sideralI actually had this in the playback options first, but thought it would best live next to bookmarks
00:54:06sideralI like the idea of collecting everything resume related in one menu
00:54:09JdGordon1I dont know of *any* other options that say "enable" in the setting menu entry
00:55:37 Quit kugel (Remote host closed the connection)
00:55:40amiconnHmm, EQ?
00:55:50 Nick Ypsy is now known as YPSY (
00:56:16JdGordon1is that the only one?
00:56:26amiconnNot sure
00:56:49JdGordon1that probably has it because just "EQ" as the top item would look odd
00:56:49pixelmaI was complaining about the "requires initialised database" in the option name too a while ago, found it unusual ;)
00:56:55sideralWell, I find not including "enable" quite confusing, as in "Crossfeed" in the "Crossfeed" menu
00:57:05JdGordon1dithering, timestretch, compressor all in the same submenu dont have it
00:57:40sideralpixelma: That will go away, see last comment of FS #11748
00:58:11sideralbut if that's the standard, I'll comply
00:58:31sideralI mean not including "enable"
00:58:39JdGordon1out of curiosity, when was this commited?
00:58:57JdGordon1(I dont particularly care that it went in, just want to see when)
00:59:03siderala couple of weeks back, by saratoga
00:59:05pixelmaa while agp by saratoga
00:59:21*JdGordon1 has been out of it for a while
00:59:38sideralJan 2
01:01:21sideralJdGordon: I suggest to reshuffle the resume options into their own menu in a second step, after I've completed committing the agreed-upon autoresume feature set
01:02:33JdGordon1sideral: ok, but really, anything with translation requirements shouldnt be commited in steps, it causes waste and annoying g issues later
01:02:46JdGordon1if the strings are stabelised before the next release that would be great
01:03:34sideralAdvice taken. But I don't think it would have been possible to commit anything from this feature without cutting it into pieces first
01:04:29sideralAnd someone could have told me about the "enable" policy earlier :)
01:04:32 Quit Topy44 (Read error: Connection reset by peer)
01:04:51 Quit domonoky (Quit: Leaving.)
01:04:53pixelmaI don't thinke there's an official policy ;)
01:06:37JdGordon1its not a policy per-se.... its just trying to keep tihngs consistant
01:06:56sideralSo, where should I put the autoresume menu? Can it live next to the bookmarks menu for the time being, or should it move to the playback options?
01:07:32JdGordon1for the time being it is fine, its better to work on finishing the rest of it than wasting time with moving the menu
01:08:16sideralJdGordon: Do you have an opinion on my configuration / feature set proposal in the last comment of FS #11748?
01:09:44JdGordon1not really
01:10:11 Join bonkers23234 [0] (
01:10:37sideralOK, thanks for having a look!
01:10:51 Join bonkers23213 [0] (
01:11:08bonkers23213I have an issue
01:11:14bonkers23213with my ipod
01:11:22sideralJdGordon: I should be done well before the next release, likely this week
01:12:36bonkers23213I don't know how it happened, but the songs on my ipod copies themselves
01:12:43bonkers23213but not as files
01:12:46bonkers23213on the database
01:12:56bonkers23213i updated the database, and it did it again
01:13:05bonkers23213so I removed the music folder
01:13:11bonkers23213and im going to replace them
01:13:25bonkers23213but it will take like 30 mins
01:13:27bonkers23213so ill be back
01:13:29S_a_i_n_tplease don't press enter after ever few words.
01:13:36bonkers23213gonna play some minecraft..
01:13:39S_a_i_n_tit makes it harder to read.
01:13:50pixelmatry to initialise the database, not update it
01:15:12 Quit bonkers23234 (Quit: CGI:IRC (Ping timeout))
01:15:33pixelmathere's are sometimes reports of duplicated entries in the database, maybe a bug but not reproducible at will so hard to fix. There is also the possibility that the database finds copies in e.g. a trash folder
01:18:21 Quit PaulJam (Ping timeout: 240 seconds)
01:20:01 Quit bonkers23213 (Quit: CGI:IRC (EOF))
01:31:40 Quit pamaury (Remote host closed the connection)
01:31:44 Part toffe82
01:32:22 Quit DerPapst (Quit: Leaving.)
01:36:15 Quit S_a_i_n_t (Disconnected by services)
01:36:16 Join [Saint] [0] (S_a_i_n_t@
01:40:53 Join kronflux [0] (~kronflux@
01:41:41kronfluxquestion for anyone who can answer it. is there any active development for Rockbox on the Sandisk Sansa View? The forum thread has been inactive for ages, and I was really looking forward to a port.
01:42:21[Saint]kronflux: Not that I know of.
01:45:17 Quit sideral (Quit: Leaving.)
01:45:20 Join L-Strife89 [0] (~Strife89@
01:47:04 Nick L-Strife89 is now known as Strife1989 (~Strife89@
01:47:12***Saving seen data "./dancer.seen"
01:47:29 Nick Strife1989 is now known as L-Strife89 (~Strife89@
01:59:50 Join mordocai [0] (~mordocai@
01:59:58 Quit Strife89 (Read error: Operation timed out)
02:00:31 Quit mudd1 (Ping timeout: 246 seconds)
02:00:34 Join Strife89 [0] (~Strife89@
02:00:35 Join quem_ [0] (
02:03:15 Quit BHSPitMonkey (Remote host closed the connection)
02:04:34 Quit quem (Ping timeout: 240 seconds)
02:04:34 Nick quem_ is now known as quem (
02:04:39 Join Strife1989 [0] (~Strife89@
02:04:50 Quit L-Strife89 (Disconnected by services)
02:04:55 Nick Strife1989 is now known as L-Strife89 (~Strife89@
02:07:35 Join Keripo [0] (
02:14:48 Quit kronflux (Quit: Leaving)
02:23:00 Join ntkm [0] (
02:24:31 Quit ntkm (Client Quit)
02:25:02foolshhi htkm can I help you?
02:33:45 Join mordocai_laptop [0] (~mordocai@
02:38:34 Quit antil33t (Read error: Connection reset by peer)
02:38:39 Join antil33t [0] (
02:40:39 Quit t0rc (Ping timeout: 260 seconds)
02:45:38 Quit designate72 (Quit: Leaving)
02:46:06 Quit GeekShad0w (Quit: The cake is a lie !)
02:52:31 Join Strife1989 [0] (~Strife89@
02:52:50 Join FBI_Guy [0] (
02:54:38 Quit Keripo (Quit: Leaving.)
02:55:45 Join Keripo [0] (
02:56:10 Join Judas_PhD [0] (
02:56:24 Quit L-Strife89 (Ping timeout: 260 seconds)
02:57:18 Quit Keripo (Client Quit)
03:02:40 Quit Judas_PhD (Ping timeout: 276 seconds)
03:04:32 Quit FBI_Guy (Ping timeout: 272 seconds)
03:06:05 Join Keripo [0] (
03:13:18 Nick Strife1989 is now known as L_Strife89 (~Strife89@
03:13:22 Nick L_Strife89 is now known as L-Strife89 (~Strife89@
03:15:03 Join Judas_PhD [0] (
03:16:39*foolsh dreamed about opengl plugins in rockbox last night
03:19:03 Join navysealltblue [0] (
03:21:31 Join fdinel [0] (
03:22:38[Saint]foolsh: And...what devices would you run those on?
03:22:55[Saint](keep in mind, plugins are aimed at all platforms)
03:28:43 Quit navysealltblue (Quit: CGI:IRC (EOF))
03:33:14 Join MaxHR [0] (~chatzilla@2001:5c0:1000:b::8881)
03:37:22MaxHRHello, using rockbox 3.6 with sansa, recent strange problems have been happening, player stops playing and becomes frozen, zero files show up when connecting to computer via usb... same issue when running regular sansa firmware, any ideas to fix?
03:39:40MaxHRactually, I can sometimes browse by usb when using the sansa firmware, but when rockbox is booted, it will not show up at all when connected
03:41:07MaxHRjust now I connected by sansa firmware, saw all folders, then double click on music folder, get zero files, then go up to root, now also show no files
03:45:40MaxHRany way to reset my player or fix these errors?
03:47:16***Saving seen data "./dancer.seen"
03:47:45 Quit Judas_PhD (Ping timeout: 240 seconds)
03:48:26 Join webguest26 [0] (
03:49:08 Quit MaxHR (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014])
03:49:10 Quit webguest26 (Client Quit)
04:01:42 Join Judas_PhD [0] (
04:06:11 Nick JdGordon1 is now known as JdGordon (
04:06:23 Quit JdGordon (Changing host)
04:06:23 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
04:06:49 Quit Judas_PhD (Ping timeout: 240 seconds)
04:07:14 Join ClaytonBraasch [0] (
04:07:51 Part ClaytonBraasch ("Coffee break.")
04:13:57 Quit Kitar|st (Ping timeout: 250 seconds)
04:15:06 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:15:06 Quit amiconn (Disconnected by services)
04:15:10 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:15:10 Quit pixelma (Disconnected by services)
04:15:23 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:15:25 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:17:39 Join quem_ [0] (
04:18:43 Quit quem (Ping timeout: 240 seconds)
04:18:51 Nick quem_ is now known as quem (
04:19:27 Join Kitar|st [0] (
04:27:36 Quit TheSeven (Ping timeout: 255 seconds)
04:31:22 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:35:21 Join AlexP_ [0] (~alex@rockbox/staff/AlexP)
04:35:47 Quit AlexP (Ping timeout: 240 seconds)
04:41:59 Quit markun (Ping timeout: 272 seconds)
04:52:35 Join Barahir_ [0] (
04:55:57 Quit Barahir (Ping timeout: 255 seconds)
04:57:02 Join Joat1 [0] (~fraser@
04:59:27 Part Joat1
05:11:50 Join Rob2223 [0] (
05:13:20 Join markun [0] (~markun@
05:13:20 Quit markun (Changing host)
05:13:20 Join markun [0] (~markun@rockbox/developer/markun)
05:15:40 Quit Rob2222 (Ping timeout: 264 seconds)
05:21:38 Quit Keripo (Quit: Leaving.)
05:24:16 Join Keripo [0] (
05:34:43 Join quem_ [0] (
05:36:42 Quit Horscht (Quit: Verlassend)
05:38:59 Quit quem (Ping timeout: 272 seconds)
05:39:00 Nick quem_ is now known as quem (
05:45:34 Quit L-Strife89 (Quit: Shutting down for the night)
05:47:20***Saving seen data "./dancer.seen"
05:52:15 Quit mordocai_laptop (Quit: Leaving)
05:53:22 Join Judas_PhD [0] (
05:55:48 Quit Keripo (Quit: Leaving.)
06:01:12 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
06:07:27 Join Keripo [0] (
06:14:51 Quit MethoS- (Remote host closed the connection)
06:25:09 Quit quem (Read error: Connection reset by peer)
06:25:30 Join quem [0] (
06:43:07 Join dxrt [0] (~dxrt@2607:f358:1:fed5:11:0:1399:1)
06:45:19 Part dxrt ("Leaving")
07:10:28 Quit TheSeven (Ping timeout: 240 seconds)
07:23:57 Join kugel [0] (~kugel@
07:23:58 Quit kugel (Changing host)
07:23:58 Join kugel [0] (~kugel@rockbox/developer/kugel)
07:47:24***Saving seen data "./dancer.seen"
07:51:15 Quit JdGordon (Quit: leaving)
08:00:19 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
08:05:02 Join LinusN [0] (
08:05:02 Quit LinusN (Changing host)
08:05:02 Join LinusN [0] (~linus@rockbox/developer/LinusN)
08:08:42 Quit kugel (Quit: Yaaic - Yet another Android IRC client -
08:08:48 Join esperegu [0] (~quassel@
08:08:59 Join Keripo1 [0] (
08:09:17 Join kugel [0] (~kugel@rockbox/developer/kugel)
08:10:34kugelI did get a mail (for the first time) for reported post in the forums. Did anyone else get it too?
08:10:53 Quit Keripo (Ping timeout: 264 seconds)
08:13:55 Quit Judas_PhD (Quit: This is a quitting message)
08:16:34 Join PaulJam [0] (
08:21:16 Join Judas_PhD [0] (
08:23:57 Join sideral [0] (~sideral@unaffiliated/sideral)
08:31:04 Join quem_ [0] (
08:33:13 Quit quem (Ping timeout: 240 seconds)
08:33:21 Nick quem_ is now known as quem (
08:34:12 Quit [Saint] (Disconnected by services)
08:34:13 Join S_a_i_n_t [0] (S_a_i_n_t@
08:40:18 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
08:42:13 Join fyrestorm [0] (
08:43:18 Join mudd1 [0] (
08:44:37*LinusN flashed peturs h120 last night with his tblcf bdm interface - it took over 8 hours!!!
08:45:00LinusNthe tblcf is so much slower than the parallel port wiggler
08:45:41LinusNi expected it to be slower, but wow...
08:46:14LinusNand the best part is that i flashed the wrong image, so i have to do it again :-(
08:47:08LinusNi better spend some time to write a bdmctrl/flashlib script to do the flashing on the player
08:47:46kugelLinusN: how much is that in bit/s?
08:48:23LinusNwell, the image was 0x1e2000 bytes, do the math :-)
08:48:36Slasheriwhat was wrong with that player?
08:48:47LinusNSlasheri: he flashed a bad bootloader
08:48:54Slasheriah, ok
08:49:18Slasheribut wouldn't it have been enough to flash just the bootloader and reset vector? :) or is that possible..
08:49:54LinusNSlasheri: the problem is that the rockbox bootloader is located at the end of the flash
08:50:09LinusNbut yes, i could flash those two parts separately
08:50:21SlasheriLinusN: true, so the bdm cannot skip that empty flash space..
08:50:43LinusNit can do anything, i wrote the flashing program
08:51:23Slasherioh, nice :)
08:51:42Slasheribut then it should be much faster if it would be able to flash just the bootloader
08:52:10LinusNand we have plugins for reflashing the entire flash, right?
08:52:16*LinusN has a weak memory
08:52:42Slasheriiriver_flash can flash bootloader and program memory separately
08:52:56 Quit Judas_PhD (Quit: This is a quitting message)
08:54:00LinusNhmm, worth a try
08:54:17LinusNi just wish i had more time to work with rockbox...
08:54:23sideralSlasheri: I have two questions related to the tagcache. Do you have some time now or later for a quick chat?
08:54:28Slasheriit can also restore the original firmware file, if that is provided in binary memory dump format (but it will not overwrite rockbox bootloader9
08:54:44LinusNSlasheri: nice
08:55:21Slasherisideral: please ask :)
08:55:29 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
08:56:15sideralcool! here's my first one:
08:56:15sideralCould you briefly explain the intention behind the commit ID? In the DB debug screen, I have never seen a commit ID other that 0...
08:57:03Slasherisideral: the commit id should increase every time a tagcache commit has been executed. The purpose is that you could search for recently added tracks
08:57:49sideralwhen are such commits done?
08:58:22sideralthis is actually related to my 2nd questions: Why are DB changes not reflected immediately in the tagtree (DB) browser?
08:59:13sideralI've noticed that when the playcount is updated, the result screen for a DB query taking playcount into account updates only half a minute later
09:00:59 Join ender` [0] (
09:02:31sideralwhereas reading the resume offset always reads the latest value due to my changes to check_virtual_tags (r28645 / FS #11723)
09:02:47Slasherisideral: commit is done after a new files have been found during tagcache update scan (if there is enough free memory) or on the next boot up
09:03:46Slasherisideral: changes cannot be reflected immediately because the commit is quite complex and cpu&disk intensive process. It requires re-sorting all tables and re-generating index file
09:05:02sideralHmm, that change I mentioned does seem to allow tagtree_buffer_event to read the latest stats and resume offset −− why?
09:05:02kugelplaycount increment needs re-sorting and re-generating the index file?
09:05:07Slasheritagcache updater basically creates a temporary file reflecting changes to the db that needs to be done. And the commit process reads and processes that file
09:05:29Slasherikugel: no, only when adding new tracks
09:05:47Slasherikugel: or updating existing tags
09:06:47Slasherisideral: statistics are fixed length and have a fixed location inside a tracks index entry. And no sorting is necessary for statistics
09:07:31Slasheritables must be sorted because some very low memory targets (archoses) don't have enough memory and power to do sorting of long lists
09:07:43Slasheriso pre-sorted lists along chunked browsing can be used
09:07:48sideralI see
09:07:48sideralI'm concerned only about updated numeric tags. Updating them has immediate effect for tagtree_buffer_event but not for the DB browser queries −− why?
09:08:44Slasherisideral: that's because updates are buffered in ram and committed to disk when disk will spin up
09:09:23Slasheriif tagcache is loaded to ram, you could basically flush those changes to ram immediately.. but iirc, that is not done yet
09:10:10 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
09:10:49sideralhmm, tagtree_buffer_event uses check_virtual_tags to read numeric data, which I changed to use store-to-load forwarding, so no write to disk is needed. Does the DB browser use a different interface to get at the same data?
09:13:15Slasherisideral: i don't remember exactly, you should check tagtree.c to see how much it differs
09:13:15 Quit quem (Read error: Connection reset by peer)
09:13:27 Quit bluebroth3r (Ping timeout: 240 seconds)
09:14:14 Part domonoky
09:14:21sideralOK, I'll have a look at it. Thanks Slasheri!
09:16:14 Join Judas_PhD [0] (
09:16:42 Join mrkiko [0] (~mrkiko@
09:18:36 Join quem [0] (
09:19:02 Join petur [0] (d408b802@rockbox/developer/petur)
09:19:13 Quit mordocai (Quit: sleep time!)
09:27:03 Join n1s [0] (~n1s@rockbox/developer/n1s)
09:31:13 Quit kugel (Remote host closed the connection)
09:31:25 Join quem_ [0] (
09:31:33 Join wodz [0] (
09:31:49 Join wodz_ [0] (
09:31:59 Quit wodz_ (Client Quit)
09:32:10 Quit wodz (Client Quit)
09:32:26 Join wodz [0] (
09:33:55 Join efyx [0] (
09:35:02wodzLinusN: I see you had moderate success with tblcf. When I was playing with it the upload speed was something like few kB/s.
09:35:55 Quit quem (Ping timeout: 260 seconds)
09:35:55 Nick quem_ is now known as quem (
09:37:10Zagorwodz: according the the numbers, linus' speed was 68 bytes/second...
09:38:04 Join einhirn [0] (
09:39:08wodzZagor: yes that's what strike me
09:42:08wodzI was uploading to DRAM (or IRAM if this was bootloader)
09:42:25wodzmaybe there is something wrong with the flashing program
09:45:11wodzI remember that dumping content of the 1MB flash in HD200 was long but no near THAT long
09:47:28***Saving seen data "./dancer.seen"
09:54:28 Join Buschel [0] (
09:59:02 Quit sasquatch (Quit: WeeChat 0.3.2)
09:59:27 Join sasquatch [0] (
10:19:00 Quit Zagor (Quit: disconnecting from stoned server.)
10:19:14 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
10:20:21LinusNwodz: i use host flashing
10:20:56 Join pamaury [0] (
10:20:56 Quit pamaury (Changing host)
10:20:56 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
10:21:14LinusNwhich means 5 bdm commands per word
10:21:49LinusNi'll try the bdmctrl/flashlib approach when i have more time
10:24:03wodzLinusN: I would rather upload small flashing program + binary dump to sdram and let it self reflash
10:24:24LinusNwodz: ...which is the bdmctrl/flashlib approach
10:24:59 Join kugel [0] (~kugel@rockbox/developer/kugel)
10:25:03wodzLinusN: have you checked what gdb reports when doing simple load elf to iram?
10:25:22LinusNwodz: nope, i haven't used gdb yet
10:26:02LinusNif all goes well, all i need to do is write the bdmctrl script to set up the dram controller, upload the flash programming code and upload the binary to sdram
10:26:08wodzLinusN: I recall now that the main drawback of tblbdm is the latency in command response due to the way how usb packets are scheduled
10:26:17LinusNwodz: yup
10:27:10wodzgood luck :-)
10:27:28LinusNi have two goals with this. 1) restore petur's h120 before friday, and 2) create a bdmctrl script to perform faster flashing
10:28:00LinusNso other people can unbrick iriver devices with a tblcf
10:28:50CIA-7New commit by Buschel (r29185): Do not parse into 'ilst' atom if size is 0. Fixes playability issue in FS #11916.
10:29:15wodzregarding bad flash - was it error in flashing or you just uploaded wrong file?
10:29:29LinusNuploaded the wrong file
10:29:54LinusNbtw, my flashing program had problem reading the CSR
10:30:40LinusNit's been years since i worked with this, is the CSR adress different on each cpu?
10:31:10wodzbear in mind that tblcf seems to quite sensitive - I had to put ferrite beads on the flat cable between tblcf and my target to work more or less stable
10:31:34LinusNmine seems to be quite stable
10:31:55LinusNit's just that CSR reads all zeroes
10:32:33wodzLinusN: you have to check in the gdb scripts I send you I don't remember from the top of my head
10:33:02CIA-7r29185 build result: All green
10:33:53 Quit factor (Read error: Connection reset by peer)
10:33:59 Join DerPapst [0] (
10:34:18pixelmadidn't pondlife also have a broken bootloader H300?
10:34:27LinusNpixelma: he's next :-)
10:34:35pixelmaah :)
10:38:27CIA-7New commit by Buschel (r29186): Recognize AAC-HE SBR with upsampling and correct duration, bitrate, seek and resume behaviour for such files. When SBR upsampling is used the decoder ...
10:41:46mystica555hm interesting commit.. :) has aac+ become realtime yet on the sansa e200v1 ?
10:42:21CIA-7r29186 build result: All green
10:43:17Buschelmystica555: no, this change just fixes the handling of duration for such files
10:43:34mystica555Buschel: well, i was meaning in the broader scope of the codec support
10:43:39mystica555last time i tried was 2 years ago
10:44:47Buschelmystica555: it is a lot faster than 2 years ago, but still not realtime of PP processors
10:49:48mystica555granted, i still am not a fan of the harsh sound, but i figure itd be nice maybe to pack more songs onto all the tiny memorycards ive got lying around from cellphone purchases and upgrades >.>
10:50:33gevaertsLinusN: do you happen to have spare flatcables? If so, maybe you could just leave petur's h120 in ready-to-connect state and he gets mine :)
10:50:53*gevaerts wants to have a bdm-ready h320 and h120
10:51:28 Join factor [0] (~factor@
10:52:05LinusNfor some reason, i have lost all my connectors, i have to order new ones
10:52:25LinusNi have cables, but no spare connectors
10:53:34LinusNand the ones i use to order are no longer manufactured, so i have to order another type and try it out
10:54:10LinusNwhat connector do you have in the tblcf end?
10:55:44LinusNi made an adapter for the tblcf with the same connector as on the h120, so i can use a standard flat cable in between
10:55:48gevaertsno idea actually. I'd get wodz' equipmenty (or a copy of it)
10:56:02*gevaerts is new to all of this
10:56:42LinusNgah, the bdmctrl program can only load elf's, i want to load a binary image. :-(
10:58:14wodzgevaerts: so you decided you want to borrow my tblcf finally ;-)
10:58:22gevaertsyes :)
10:58:34gevaertsOr get you to build a second one and buy it :)
10:59:03pixelmakugel: yes, the reported post mails sometimes get through to all staff instead of admins or global mods only. I've seen it before and also got the mail tonight
10:59:16pixelmaor this night rather
11:02:37kugelI never had such a mail
11:05:54wodzgevaerts: if you would like me to build a new one tblcf I will have to order PCB for it. I can design the PCB with whatever connector you like (or a few actually) as this will almost have no impact on final PCB cost (when we count this in single pieces).
11:06:28gevaertswodz: do you have time for that?
11:06:37 Quit petur (Quit: Page closed)
11:07:17gevaertsI would prefer to have my own equipment in the end so I can do things like test iriver bootloaders in the long term and not just once
11:07:35 Quit iq (Ping timeout: 255 seconds)
11:07:46gevaertsBorrowing yours of course works, but you may want it back at some point :)
11:07:54wodzgevaerts: Of course no :-). But I can postpone hacking rk27xx DAP and design the PCB.
11:07:56 Quit Keripo1 (Quit: Leaving.)
11:08:02gevaertsAnd I definitely don't feel up to building this myself
11:08:50 Quit kugel (Ping timeout: 240 seconds)
11:09:53wodzgevaerts: I still don't think I would need this tblcf in the easy to describe future.
11:10:44gevaertswodz: There's of course also the option that you send me yours and when you need it I send it back while you start building a second one :)
11:11:45LinusNwodz: it would be cool if you could build a tblcf PCB with the iriver 20-pin connector :-)
11:12:12wodzgevaerts: or even simpler - If I'll need the one you just refund me new PCB.
11:12:28gevaertswodz: that works too, if you don't need it *too* fast
11:15:55wodzgevaerts: 1) I don't need BDM for enything else but rockbox hacking. 2) I have at least 3 non CF rb targets to listem music if I screw my MPIOs up
11:16:00 Join petur [0] (d408b802@rockbox/developer/petur)
11:16:35gevaertswodz: just wait until MPIO users start doing things wrong and sending their players to you :)
11:16:50gevaertsBut we can decide that when you need it next
11:18:24wodzThe only question is if you can easily interface standard BDM goldpin connector in my tblcf with irivers
11:19:20*gevaerts leaves the answer to that to wodz and LinusN
11:19:45LinusNi made an adapter for that, using an old prototyping PCB i had lying around
11:20:28wodzwait - iriver has 20 pin 0.5mm pitch flat cable?
11:21:18wodzok I can send you my hacky "adapter" than. You will rewire it to your needs
11:21:35wodzthe flat cable also
11:22:34 Join kevku [0] (~kevku@2001:7d0:0:f9af:215:c5ff:fe7e:794d)
11:23:06wodzgevaerts: Let me conclude - I'll search the pod and when I find it I'll mail it to you - that is the most simple approach
11:23:59gevaertswodz: ok. I'll email you my address
11:24:15wodzhave to go work, bye
11:24:20 Quit wodz (Quit: Leaving)
11:24:33 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
11:29:51 Quit foolsh (Ping timeout: 240 seconds)
11:30:50 Quit markun (Ping timeout: 240 seconds)
11:39:00 Quit Dreamxtreme (Ping timeout: 250 seconds)
11:41:26 Quit esperegu (Ping timeout: 265 seconds)
11:45:41 Join esperegu [0] (~quassel@
11:46:40S_a_i_n_tJdGordon|: playlist/view current playlist
11:47:29***Saving seen data "./dancer.seen"
11:51:24 Join webguest09 [0] (c391b46a@gateway/web/freenode/ip.
11:52:42webguest09Buschel: Hello. In a recent change to metadata/mp4.c, the comment says "at least a size of 8", but the condition uses ">8". Shouldn't it be ">=8"?
11:53:04 Join dfkt [0] (dfkt@unaffiliated/dfkt)
11:55:54 Join kugel [0] (~kugel@rockbox/developer/kugel)
11:57:40 Quit webguest09 (Quit: Page closed)
12:04:43 Quit kugel (Remote host closed the connection)
12:05:19 Quit quem (Ping timeout: 245 seconds)
12:05:44Buschelwebguest09: (just if you should read the logs) you're right, the comment is 100% matching the code. but: in fact size must be much larger than 8 to contain any reasonable metadata. at least 8 byte are needed to read the next atom, this is without any further content. we could also check for >0 and assume any size >0 will be correct...
12:06:24 Join quem [0] (
12:11:23 Join quem_ [0] (
12:11:59 Quit quem (Ping timeout: 245 seconds)
12:12:11 Nick quem_ is now known as quem (
12:24:50 Join iq [0] (~iq@unaffiliated/iq)
12:25:21 Join JdGordon1 [0] (
12:27:03 Quit mystica555 (Ping timeout: 240 seconds)
12:28:31BuschelLear: (for the logs) I would like to close FS #8923 (several problems with m4a) as it describes issues with large file which I cannot reproduce in sim and on my iPod Video
12:29:43 Quit JdGordon1 (Client Quit)
12:39:45 Join chattr [0] (
12:41:58 Part chattr
12:45:41 Quit Barahir_ (Ping timeout: 255 seconds)
12:47:22 Join Barahir [0] (
13:00:39 Quit petur (Ping timeout: 265 seconds)
13:02:26 Quit mudd1 (Ping timeout: 240 seconds)
13:04:36 Join petur [0] (d408b802@rockbox/developer/petur)
13:09:25 Quit Rob2223 (Quit: Rob2223)
13:09:52 Join Rob2222 [0] (
13:11:46 Join mudd1 [0] (
13:32:41n1sBuschel: the long files not playing was (is?) due to a table of offsetts being kept by the container parser becomming too large
13:32:44 Quit esperegu (Remote host closed the connection)
13:32:49n1sdid you fix that?
13:33:30n1sit has been discussed in here numerous times so the logs should have all gory details if you are interested :)
13:33:51 Quit alexbobP (Ping timeout: 255 seconds)
13:34:10n1svery cool that you've tackled these old bugs btw :)
13:34:34Buscheln1s: ah, understood now. the arrays are still loaded...
13:35:45n1si think saratoga looked into it but he wasn't certain his patch was correct (IIRC)
13:36:17 Join Dreamxtreme [0] (~Dre@
13:38:32Buschelthis will be too easy... we need to shrink / merge the tables...
13:38:54Buschelof course I wanted to say -> this will _not_ be too easy
13:40:22*Buschel found a file which give wrong duration with hsi latest changes
13:40:57Buschelit is an SBR file that signals and uses SBR_upsampling but already states it uses a frame size of 2048
13:41:27Buschelis it ok to assume only sizes of <=1024 will be upsampled to 2048?
13:47:33***Saving seen data "./dancer.seen"
14:17:06 Quit GeekShadow (Read error: Connection reset by peer)
14:17:30 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
14:19:47 Join kugel [0] (~kugel@
14:19:47 Quit kugel (Changing host)
14:19:47 Join kugel [0] (~kugel@rockbox/developer/kugel)
14:25:25 Quit Buschel (Ping timeout: 250 seconds)
14:31:02 Join shai [0] (
14:47:25 Join mystica555 [0] (
14:51:12 Quit sideral (Ping timeout: 240 seconds)
14:52:10 Join sideral [0] (~sideral@
14:52:10 Quit sideral (Changing host)
14:52:10 Join sideral [0] (~sideral@unaffiliated/sideral)
14:57:08CIA-7New commit by kugel (r29187): Android/Java: Disable implicit compilation of dependecies to make the life easier for make.
14:57:36kugeldionoea, S_a_i_n_t: make confusion fixed
14:57:51kugelmy suspicion was right and javac has a switch for it
15:01:02CIA-7r29187 build result: All green
15:01:36 Quit sideral (Remote host closed the connection)
15:01:51 Join Buschel [0] (
15:01:58 Join sideral [0] (~sideral@
15:01:58 Quit sideral (Changing host)
15:01:58 Join sideral [0] (~sideral@unaffiliated/sideral)
15:11:51 Quit sideral (Remote host closed the connection)
15:12:32 Join sideral [0] (~sideral@
15:12:32 Quit sideral (Changing host)
15:12:32 Join sideral [0] (~sideral@unaffiliated/sideral)
15:18:27 Join evilnick_B [0] (0c140464@rockbox/staff/evilnick)
15:26:45 Quit mc2739 (Ping timeout: 272 seconds)
15:27:52 Join komputes [0] (~komputes@ubuntu/member/komputes)
15:29:55 Part LinusN
15:32:01 Join quem_ [0] (
15:35:51 Quit quem (Ping timeout: 240 seconds)
15:35:51 Nick quem_ is now known as quem (
15:36:32 Join LambdaCalculus37 [0] (
15:36:32 Quit LambdaCalculus37 (Changing host)
15:36:32 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
15:39:56 Quit mrkiko (Quit: leaving)
15:43:54 Quit Judas_PhD (Ping timeout: 255 seconds)
15:46:17 Join L-Strife89 [0] (
15:47:35***Saving seen data "./dancer.seen"
15:49:40 Join FedericoLang [0] (
15:50:46FedericoLangi want upload a file fmpreset from caracas venezuela
15:52:40gevaertsFedericoLang: ah, you presumably need wiki access?
15:55:12 Quit fyrestorm (Quit: Ur skills' fireproof like a wooden panel -- U got feds talking leet on your IRC channel!)
15:56:15 Join quem_ [0] (
15:57:07 Join Judas_PhD [0] (
15:58:04dionoeakugel: thanks
16:00:11 Quit quem (Ping timeout: 240 seconds)
16:00:11 Nick quem_ is now known as quem (
16:06:13 Join MethoS- [0] (~clemens@
16:08:40 Quit user890104 (Ping timeout: 245 seconds)
16:13:02CIA-7New commit by Buschel (r29188): Find a more consistent and resilient way to handle SBR upsampled files. The detection is only done in one place (the metadata parser) and takes into ...
16:14:12 Join FBI_Guy [0] (
16:17:44 Quit n1s (Ping timeout: 255 seconds)
16:18:20CIA-7r29188 build result: All green
16:18:50 Quit jae (Read error: Connection reset by peer)
16:19:07 Quit FBI_Guy (Client Quit)
16:19:26 Join FBI_Guy [0] (
16:19:43FedericoLanggevaerts, yes ai need wiki acces
16:22:05 Quit FedericoLang (Quit: CGI:IRC (EOF))
16:24:23 Join FedericoLang [0] (
16:28:18gevaertsFedericoLang: should be done now
16:28:38 Quit kugel (Remote host closed the connection)
16:28:57 Join Barahir_ [0] (
16:31:51 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
16:32:00 Quit Barahir (Ping timeout: 272 seconds)
16:34:52 Quit FedericoLang (Quit: CGI:IRC (EOF))
16:39:04*Buschel somehow fears the iPod Video big-HDD "keep spinning" bug is back
16:41:08 Quit liar (Ping timeout: 255 seconds)
16:42:57 Join liar [0] (
16:43:22Buschelgevaerts: weren't you working on the iPod Video build unification?
16:43:41gevaertsBuschel: a while ago, yes
16:43:57Buschelhmm, how is MEM defined for those builds now?
16:44:11 Quit mystica555 (Ping timeout: 240 seconds)
16:46:40*gevaerts doesn't find that
16:47:45gevaertsMEMORYSIZE is 64, but I don't see MEM
16:47:50 Quit kevku (Ping timeout: 272 seconds)
16:49:02BuschelI am not 100% yet. But it seems like our good old ATA hack for the iPods Video with bigger HDDs has been removed/disabled :/
16:49:26Buschelsee r17682
16:49:53gevaertsah, yes, that one
16:49:55Buschelthe code in ata.c is still there, but it needs MEM==64. the code in thread.c is removed...
16:50:47gevaertsWe did discuss that code at the time, but I don't remember what was decided
16:51:10gevaertsExcept that it wasn't entirely correct since really hard disk size isn't related to memory
16:52:16Buschelthe code in thread.c was removed inr26479 and not moved to arm/thread-pp.c
16:52:57 Join Highlander [0] (
16:54:13Buschelno wait. it is in thread-pp.c ;)
16:54:56Buschelat least it looks like...
16:55:56*Buschel summons jhMikeS
16:56:15gevaertsBuschel: where is/was MEM defined?
16:56:41BuschelI do not really know. I just used it
16:56:47 Quit petur (Ping timeout: 240 seconds)
17:00:35*Buschel will just call a build for his iPod Video
17:01:42 Join toffe82 [0] (
17:01:59Buschelin fact it *is* still set to 64
17:02:46 Join kevku [0] (~kevku@2001:7d0:0:f9af:215:c5ff:fe7e:794d)
17:03:14gevaertsBut where is it set?
17:03:16 Join mystica555 [0] (
17:03:21*gevaerts really can't find that!
17:05:05*Torne was fairly sure at the time that we had preserved allt hat hack code
17:05:19Torneand we had just therefore expanded its use to the 32MB ipod video build as well.
17:05:22Buschelheh, look at plugin.h, line 31 :)
17:05:31TorneYou could jsut drop the MEM condition on it
17:05:37Torneand see if the build changes :)
17:06:03 Quit mc2739 (Quit: leaving)
17:06:19BuschelTorne: MEM is set to 64 and the build uses the code. but thread-pp.c has changed a lot since the split
17:06:43TorneWhen we were looking at this before I never managed to figure out what the hack actually *does*
17:06:57Torneit looks like it's an alternate mutex mechanism but I'm not sure what's different about it or why it's needed
17:08:16Buschelafaik nobody really understood why it was needed...
17:12:16 Quit FBI_Guy (Quit: leaving)
17:15:16 Join Keripo [0] (
17:15:42 Quit Xerion (Ping timeout: 246 seconds)
17:16:12 Quit L-Strife89 (Ping timeout: 276 seconds)
17:17:57Buschelif this user who report this issue does not see the failure with v3.6 that could also be a hint (v3.6 was branched off before the thread.c was splitted)
17:19:03 Join kugel [0] (
17:19:04 Quit kugel (Changing host)
17:19:04 Join kugel [0] (~kugel@rockbox/developer/kugel)
17:20:44 Quit Highlander (Remote host closed the connection)
17:20:51 Join GeekShad0w [0] (
17:20:53 Quit GeekShadow (Read error: Connection reset by peer)
17:21:56 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
17:24:33*gevaerts claims that if MEM is defined to 64, it must be defined *somewhere*
17:24:43 Part Zagor
17:25:02Buschellets make this a job for GSoC 2011 :)
17:25:48Torneaha, command line
17:25:56Buschelgreat :/
17:25:57gevaertsOf course grep -w MEM didn't find it...
17:26:04gevaertsAnd yes, that should be cleaned up
17:26:16 Join Strife89TX [0] (~cstrife89@
17:26:30Buschelwe should remove this thing and use MEMORYSIZE in all places. it is a pain to search for "MEM"
17:26:43*gevaerts agrees
17:27:26Buscheland we should use a IPOD_VIDEO define instead of checking for MEMSIZE for the ata hack
17:27:49gevaertsnot sure
17:28:10gevaertsMaybe the MAX_PHYS_SECTOR_SIZE check is enough
17:28:32gevaertsI mean, what happens if someone puts one of the affected disks in some other player?
17:28:33TorneWell, that depends why the hack exists, no?
17:28:45Tornewell, that case too
17:28:51Buschelhmm, ipod 6g uses MAX_PHYS_SECTOR_SIZE as well
17:29:00Torneiw as thinking "do we need the hack because of large physical sectors or do we need the hack because of the specific toshiba drive"
17:29:33Torneis it a workaround for a limitation o the sector emulation code, or is it doing something magical for the benefit of that specific drive?
17:29:49Torneif the former then it shohuldn't be ipod video specific :)
17:29:54Torneif the latter then it's debatable
17:29:59Buschelmaybe jhMikeS can explain when he's available
17:30:16Torneother people do builds with MAX_PHYS_SECTOR_SIZE on various players
17:30:30Tornebecause there are, sadly, 4096 byte sector drives out there that have the same failure to emulate 512 byte access
17:30:34Buscheldo those builds use MEM==64 as well?
17:30:43TorneNot unless they happen to be using ipodvideo
17:30:45Torneno idea if they are
17:31:07Tornebut yeah, my boot time detection of whether emulation is needed only lets you use 4096 byte drives that emulate correctly
17:31:19Tornepeople still have to do custom builds for the occasional 4096 byte drive that doesn't.
17:31:34TorneI had hoped that that didn't exist but it turns out they do
17:31:42Torneand guess what toshiba made them
17:35:37 Join foolsh [0] (
17:44:03 Quit guymann (Ping timeout: 255 seconds)
17:44:55 Join guymann [0] (~charles@
17:46:27 Quit Judas_PhD (Ping timeout: 240 seconds)
17:47:31 Quit DerPapst (Quit: Leaving.)
17:47:39***Saving seen data "./dancer.seen"
17:54:41 Quit Keripo (Quit: Leaving.)
17:57:13 Quit PaulJam (Quit: .)
17:59:35 Quit Strife89TX (Quit: Switching apps.)
18:01:07 Join Judas_PhD [0] (
18:02:41sideralAnyone here interested in debugging the ClipV2 background noise? FS #11915 −− I'd appreciate if someone could post an audio register dump from the Clip+
18:04:27 Join designate72 [0] (
18:12:03*Buschel works on removing MEM
18:13:39 Quit Barahir_ (Ping timeout: 240 seconds)
18:15:26 Join user890104 [0] (~Venci@2001:0:5ef5:73b8:24e9:2dc9:2b16:7ff3)
18:19:08 Quit TheSeven (Ping timeout: 276 seconds)
18:19:21 Join DerPapst [0] (
18:20:20 Quit sideral (Ping timeout: 245 seconds)
18:21:57 Quit liar (Ping timeout: 255 seconds)
18:24:51 Join webguest806 [0] (
18:25:32 Join mrkiko [0] (
18:26:01 Quit webguest806 (Client Quit)
18:26:13BuscheliPod Video build compiles and runs w/o -DMEM (it's a model with 32 MB RAM)
18:27:34 Join liar [0] (
18:29:41 Quit liar (Remote host closed the connection)
18:31:18 Join quem_ [0] (
18:31:48kugelthis action.c workaround again
18:31:56kugelit's messing up things up on touchscreen as well
18:32:20 Quit Judas_PhD (Ping timeout: 272 seconds)
18:33:02foolshkugel: just tell me where
18:34:01foolshor tell me what I'm on it
18:34:48 Join icheyne [0] (
18:35:12foolshI have noticed the framebuffer feels almost too fast, but that not a bad thing
18:35:31 Quit quem (Ping timeout: 260 seconds)
18:35:32 Nick quem_ is now known as quem (
18:35:32icheyneHi all. Does timestretch work on the Clip+?
18:35:43kugelfoolsh: what do you mean by framebuffer too fast?
18:36:12foolshicheyne: I don't know, have you checked the wiki?
18:36:57foolshkugel:I mean like if you did not release the screen or button soon enough you got an extra button press
18:38:00foolshmaybe I should slow step rockbox again on button presses, but that causes trouble :(
18:39:42 Quit icheyne (Ping timeout: 240 seconds)
18:40:28kugelan extra (ghost) button press is a bug and has nothing to do with speed
18:40:52kugelI currently fight eaten button presses which is equally buggy
18:41:27foolshyou coded in a faulty interrupt kugel read through FS #11903
18:41:30kugelTorne: do you remember the exact reason for the hack there (action.c:208)?
18:41:50Tornethe eaten one on context change?
18:42:00TorneAnd I have suspected for some time that it's *wrong*.
18:42:10Tornenot just on touchscreen targets :)
18:42:11kugelfoolsh: no point in reading through it, 90% of it is your crazy comments which don't make sense to me
18:42:26TorneI think it's broken for cases where you change context by doing a button combination as well
18:42:31 Join icheyne [0] (
18:42:40icheyneDoes not work for me
18:42:46kugelI simply go to the context menu by long pressing the screen
18:42:53icheyneWiki implies it works
18:43:10kugelthere should be a realease event but it's not seen somehow
18:43:12foolshkugel: no hard feelings, but since we can recode the virtual hardware fixing the interrupt is easy
18:43:16icheyneTimestretch would be good
18:43:20Torneicheyne: define "does not work"
18:43:23Tornewhat happens?
18:43:34CIA-7New commit by Buschel (r29189): Clean up multiple definitions of RAM size. Remove -DMEM (make) and MEM (code), use the already defined MEMORYSIZE instead.
18:43:45icheyneTimestretch shows 102%
18:43:53icheyneAnd then stops
18:44:03icheyneNo speed up
18:44:33 Join Judas_PhD [0] (
18:44:56icheyneJust wanted to cfheck before upgrading
18:45:20icheyneIt is a nice feature for podcasts
18:47:45foolshkugel: I will look into onlongKeypress or some other java vodo, and see what I find.
18:47:57pixelmaI believe you're using the wrong control keys and maybe only the "plain" pitchscreen, not enabled timestretch yet
18:48:05icheyneThe timestretch indicator reads 100% initially rises to 102% then stops
18:48:38icheyneTimestretch is enabled in sound options
18:49:01CIA-7r29189 build result: 49 errors, 0 warnings (Buschel committed)
18:49:24icheyneI know because there is only Rate number
18:52:40 Join AndChat| [0] (
18:52:55pixelmaicheyne: and which of the keys are you using (compare to (and scroll UP, the table I mean is right above the headline I linked to)
18:53:34AndChat|Sorry I was disconnected
18:53:54AndChat|Irc over android is not ideal
18:54:42 Quit icheyne (Ping timeout: 240 seconds)
18:54:49AndChat|Does timestretch work for you on your clip+
18:55:08CIA-7New commit by Buschel (r29190): Fix red. Replace remaining MEM's mit MEMORYSIZE.
18:55:16 Quit AndChat| (Client Quit)
18:56:33 Join Xerion [0] (
18:56:48kugelBuschel: mit? :)
18:57:00Buscheloops ;)
18:57:11Buschelmy germish came through :)
18:59:55CIA-7r29190 build result: All green
19:00:01foolshkugel: walk me through, step by step, your exact complaint and problem. Use the FS comments so I can track it farther down the rabbit hole.
19:09:15kugelfoolsh: I think haven't understand how our button engine works yet
19:12:38foolshkugel: just file a complaint on the tracker so I can fix this for all androids and stop being such a pussy
19:13:14 Quit foolsh (Quit: needs a cigarette.)
19:23:02 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
19:24:16 Quit Judas_PhD (Ping timeout: 276 seconds)
19:24:57 Join liar [0] (
19:27:33CIA-7New commit by kugel (r29191): Reduce excessive debug output during mp3 parsing.
19:29:05 Join Barahir [0] (
19:29:27 Join TheLemonMan [0] (~lem0n@
19:30:11CIA-7New commit by kugel (r29192): Redo r29168. Use similar but existing mechanism in the action system which seems to work better. ...
19:32:20CIA-7r29191 build result: 7 errors, 0 warnings (kugel committed)
19:36:09 Join Judas_PhD [0] (
19:36:34CIA-7r29192 build result: All green
19:37:53 Join solexx_ [0] (
19:39:09CIA-7New commit by kugel (r29193): Touchscreen: Fix minor redraw after entering the context menu.
19:39:51 Quit yawny (Ping timeout: 240 seconds)
19:40:41 Quit solexx (Ping timeout: 240 seconds)
19:43:38CIA-7r29193 build result: All green
19:47:06 Join elcan [0] (
19:47:42***Saving seen data "./dancer.seen"
19:50:41 Join Keripo [0] (
19:50:47 Quit Judas_PhD (Ping timeout: 276 seconds)
19:53:32 Join mordocai [0] (~mordocai@
19:54:57 Join sideral [0] (~sideral@
19:54:57 Quit sideral (Changing host)
19:54:57 Join sideral [0] (~sideral@unaffiliated/sideral)
19:56:36CIA-7New commit by bluebrother (r29194): Fix some unwanted empty lines in the manual. ...
20:00:39CIA-7r29194 build result: All green
20:02:04 Join quem_ [0] (
20:02:15 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
20:02:47 Nick mordocai is now known as mordocai_afk (~mordocai@
20:06:23 Quit quem (Ping timeout: 276 seconds)
20:06:23 Nick quem_ is now known as quem (
20:12:51 Quit user890104 (Ping timeout: 245 seconds)
20:17:56AlexP_bluebrother: Cool, I was wondering how to get rid of those lines a while ago :)
20:18:02AlexP_Noted for future reference
20:19:15bluebrotherAlexP_: there's another solution: use \newcommand{\firmwarefilename}{\opt{fuze}{fuzea.bin}...}, then use that commands in the code block
20:19:41bluebrotherbut since I already had the current solution in place (and it's useable in this case) I kept it :)
20:19:41AlexP_ah, local to that file?
20:19:59 Join Keripo1 [0] (
20:20:17bluebrotherwell, not really local. But you can simply start that in the file. Maybe use some kind of prefix to avoid clashes with other "local" commands
20:21:36 Quit Keripo (Ping timeout: 245 seconds)
20:23:16 Quit GeekShad0w (Ping timeout: 245 seconds)
20:25:25 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
20:26:29pixelmathat command looks horrible though
20:26:46pixelmathe one in place now
20:33:03pixelmabluebrother: we already do have a "firmwarefilename" in the platform files for the rockbox binary - not important for the current solution though ;)
20:33:08*kugel wonders if there's any way to speed up compilation of java files
20:33:26kugelthe java files alone need 40s on my notebook
20:44:33 Join thomasjfox [0] (
20:46:41 Join quem_ [0] (
20:50:32 Quit quem (Ping timeout: 240 seconds)
20:50:33 Nick quem_ is now known as quem (
20:50:51 Join bertrik [0] (
20:50:52 Quit bertrik (Changing host)
20:50:52 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
21:00:48jhMikeSBuschel: I wonder if"> works well for you. It's initially based off your FS #8264.
21:02:38Buschelthat was loooong ago ;)
21:03:27jhMikeSI decided to revisit the idea though. :) It looks better overall, imo.
21:06:37BuscheljhMike: what is the use case you want me to test? and what shall I observe?
21:07:54gevaertskugel: have you tried giving all source files to the compiler in one go?
21:08:22 Quit Rob2222 (Quit: Rob2222)
21:08:24gevaertsThe java compiler is written in java, and java startup time isn't very fast
21:08:31kugelbut the implicit compilation wasn't any faster
21:09:02 Join Rob2222 [0] (
21:09:08gevaertsDid you have fewer javac invocations?
21:09:11jhMikeSBuschel: whatever inspired the patch in the first place I guess. slow display too (since I have no targets that are really slow at rendering)
21:09:20kugeljavac −−help is fast, I don't think it's the startup time
21:09:35BuscheljhMikeS: I have :) the good old iPod Video :)
21:09:37 Join astonishing [0] (
21:09:57kugeljavac by default compiles dependencies of classes before the class itself. that also confused make
21:10:14 Nick astonishing is now known as enth (
21:10:24kugelI hoped there was some kind of ccache for java but it doesn't appear so
21:10:28gevaertsOh, definitely
21:10:41jhMikeSBuschel: it should less "droppy" since it also attempts to stop the dropping if it catches up before the countdown goes to 0
21:12:00 Quit evilnick_B (Ping timeout: 240 seconds)
21:14:49kugelgevaerts: hm, you seem right. all files in one go is 3.5s
21:15:36gevaertsOf course a proper makefile rule for that isn't trivial
21:16:07kugelbut that's strange, even a single file at longer than all in one
21:16:29kugelI suspect it's loading android.jar what takes so long
21:16:53BuscheljhMikeS: testing now
21:18:02jhMikeSBuschel: thanks. I don't always expect better throughput, just more even, with tighter sync
21:18:36gevaertsWell, that and the actual compiler code
21:19:55 Quit factor (Ping timeout: 264 seconds)
21:21:10 Join user890104 [0] (
21:21:26 Nick mordocai_afk is now known as mordocai (~mordocai@
21:21:47kugelgevaerts: yes defintily, loading android.jar (i.e. the entire android framework with all its apis) is the culprit
21:26:11*Buschel is testing with Cpt. Future
21:26:22BuschelI love this old stuff :)
21:26:43Buschelthe music is great
21:28:06kugelgevaerts: no idea how to make that observation into the makefile :(
21:30:37*jhMikeS wonders why the theme to that sounds like it has a hint of StarTrek lol
21:31:40 Quit Dreamxtreme (Ping timeout: 240 seconds)
21:31:55kugelI guess it's not doable to load android.jar in some kind of final link stage as java doesn't use the headers concept
21:32:20gevaertsno, forget that
21:32:23*gevaerts tries something
21:32:49 Quit niekie (Read error: Operation timed out)
21:33:54BuscheljhMikeS: hmm, the framerate is much lower now. but the movements are more even.
21:34:23BuscheljhMikeS: framerate dropped from 19 -> 17 (16:9) or 14 -> 11 (4:3)
21:34:33Buschel(for my test cases)
21:35:35jhMikeSI might've expected it. Which is better to watch?
21:37:25jhMikeSIt attempts to not run the frames up really late and then snap back suddenly, but the short term sync cost a bit in the long term
21:37:33 Join niekie [0] (quasselcor@CAcert/Assurer/niekie)
21:38:31kugelthis site suggests compiling 500 java files should not take longer than 14s
21:38:46 Quit Bagder (Ping timeout: 250 seconds)
21:39:33 Join Bagder [0] (~daniel@rockbox/developer/bagder)
21:40:56bluebrotherpixelma: I agree that it doesn't look nice. Feel free to create the required macros to make it work without that :)
21:41:40bluebrotherbut I like it better without the extra lines in the output than with code that isn't complicated but gives extra lines in the output :)
21:42:15gevaertskugel: on a reasonably fast machine, that's probably accurate. You have to do it in one invocation then though
21:42:42BuscheljhMikeS: hard to say what is better. I like smoother moves, but there are definately less images...
21:43:02kugelgevaerts: oh I read it wrong
21:43:30kugelI thought he meant 14s for 500 files (each seperate) and 5s for all in one
21:44:12BuscheljhMikeS: it's judders more but even now :)
21:44:54jhMikeSBuschel: I thought you just said it was more even ??
21:45:28BuscheljhMikeS: I'm missing the words to clearly describe... let me try again
21:46:10 Quit quem (Remote host closed the connection)
21:46:11 Quit TheLemonMan (Quit: free(me))
21:46:28 Join TheLemonMan [0] (~lem0n@
21:46:31 Join quem [0] (
21:47:29jhMikeSI did change that 1/2 GOP thing for P frame, instead it learns how many and holds off for total remaining
21:47:31BuscheljhMikeS: with svn there a lots of minor judders and twice a seconds there is big jump (= uneven judder). with your patch the juddering is constant but with larger steps (= even)
21:47:46***Saving seen data "./dancer.seen"
21:50:03 Join foolsh [0] (
21:50:36jhMikeSI suppose the goal is to simply make it look like a lower framerate movie as much as possible, not sure if you mean that it's approaching that
21:51:10foolshkugel: thanks buddy ;) now can I leave java behind me?
21:51:25kugelI don't know
21:51:47jhMikeSBuschel: well, lower fr for as long as it's playing catchup
21:51:54*foolsh hunts for dog food
21:53:46BuscheljhMikeS: yes, the patch seem to do this job.
21:55:24 Quit Rob2222 (Remote host closed the connection)
21:55:35gevaertskugel: I can't think of a better way than that page describes, so unless some magic lightbulb suddenly appears above Mr. Someone's head, the tradeoff described there will apply: a slow "traditional" build with full dependency tracking, or an unconditional build that's faster if all files need to be compiled but has a fixed overhead
21:55:44 Join Rob2222 [0] (
21:56:43kugelit would be nice if at least "don't re-compile if *none* of the java changed" would be possible
21:58:11 Quit w0m (Ping timeout: 276 seconds)
21:58:22 Join w0m [0] (~wom@
21:59:05kugelthe overhead of compiling all over one file (in one go) is insignificant (but one file is still noticeable) while the overhead for each javac invocation is incredible huge
21:59:36*gevaerts has another go
21:59:53gevaertsNo promises though
22:00:02kugelno problem :)
22:00:26 Quit Rob2222 (Client Quit)
22:01:22kugelrecompile all in one if any of the .java files is newer than any one of the .class files (they should all have the same timestamp) should work (if make allows)
22:01:27 Join Rob2222 [0] (
22:04:06 Quit liar (Ping timeout: 245 seconds)
22:04:48kugelor perhaps s/.class files/classes.dex/
22:05:09kugelclasses.dex is the end result of the java compilation
22:05:41 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
22:05:57 Join vnl [0] (
22:06:17 Quit Keripo1 (Quit: Leaving.)
22:06:43gevaertsI can think of ways that work for most cases, but *not* when you delete one of the class files
22:07:08 Quit mrkiko (Quit: leaving)
22:07:48 Part vnl
22:07:55kugelcan that work if you use classes.dex?
22:12:47jhMikeSBuschel: Goal met then. :) thanks again. I suppose skipping is becoming less of an issue with more recent targets anyway.
22:13:03 Join liar [0] (
22:13:49BuscheljhMikeS: no prob :)
22:14:19gevaertskugel: works except for deleting class files (which doesn't get detected)
22:14:31gevaertsYou can also do the javac in the same target as dx instead
22:14:46gevaerts(with the same problem)
22:14:58gevaertsOr use zip instead of jar
22:16:00kugelwell, there's no make target for .class files anymore, so I guess it's impossible to detect deleted ones
22:16:15gevaertsit's impossible in general
22:16:44gevaertsYou could maybe compare to $(JAVA_OBJS) somewhere, but $(JAVA_OBJS) doesn't have the inner classes
22:17:02gevaerts(things like RockboxPCM$PCMListener.class)
22:17:21 Join Horscht [0] (
22:17:21 Quit Horscht (Changing host)
22:17:21 Join Horscht [0] (~Horscht@xbmc/user/horscht)
22:17:30kugelyea that issue is in svn too
22:17:53kugelbut I think your patch is sufficient (assuming it works), nobody deletes java files :)
22:18:02gevaertshm, let me try one more thing...
22:19:50BuscheljhMikeS: btw, did you follow the earlier discussion regarding the ata hack for the iPod Video HDD?
22:20:44jhMikeSBuschel: no, I thought that was all forgotten about long ago :)
22:21:06jhMikeSIt's awful and I'd like to get rid of it though
22:21:26*Dhraakellian seems to be getting "Undefined instruction at 3002A1E0" a lot these past couple days
22:21:40Dhraakellianon track change
22:21:46*Dhraakellian suspects corruption somewhere
22:21:47Buschelit just came up my mind as a forum user describes such issue again. I was wondering whether the latest changes in ata.c and/or thread-pp.c impacted this hack?
22:24:30jhMikeSIt never seemed to be a thread or ata issue per se. I have to remind myself which changes those were.
22:25:18Buschelquestion to devs: is anybody against FS #11918? it adds a bit more detail to the codec types.
22:26:46 Join Keripo [0] (
22:27:58 Quit benedikt93 (Quit: Bye ;))
22:28:12jhMikeSmost recent change besides yours to thread-pp.c is just one for the EABI switch, which were it broken would crash everything immediately
22:28:12gevaertskugel: I can't think of a better way, except (needs more thought) a dirty hack (that still won't work for those inner classes)
22:30:09jhMikeSfor ata.c one only changes it for i.MX31 and the other is just not sending a message that wasn't doing anything, so I find it highly unlikely the two would relate
22:34:31kugelgevaerts: not worth it IMO
22:35:18*jhMikeS still guesses the hack is needed simply to prevent the core from sleeping at some moment during disk spinup or access, which make the WFI mode stick rather than wake when it should
22:38:17jhMikeSOne of the nanos had that same issue but iirc it was resolved by allowing a long driver powerup time
22:40:38gevaertskugel: right. I had one more idea, but it's getting complicated as well
22:43:30 Join Dreamxtreme [0] (~Dre@
22:46:47jhMikeSBuschel: I read the logs. I'd say it's for iPod video specifically. I've built other targets with MAX_PHYS_SECTOR_SIZE without such a bug surfacing. So, I'm going with "power/clocking setup related on iPod video".
22:46:52gevaertskugel: is the same thing, except that it adds a check for the existence of all expected class files
22:47:10gevaertsIt will still miss the inner classes, but I do think it's an improvement
22:47:26gevaertsThe net result is that if you delete a class file, the build fails
22:49:34 Quit Dreamxtreme (Read error: Connection reset by peer)
22:50:12gevaertshm, it doesn't work properly :(
22:50:33*kugel whistles "not worth it" :P
22:50:45kugelbut thanks for looking into it
22:50:59gevaertsI'll fix that bit!
22:51:03 Join Dreamxtreme [0] (~Dre@
22:55:49jhMikeSBuschel: Can't see how your MEM change could impact it. (sometimes I just stick "#error" in some pre-processor block to see if the block is being compiled)
22:56:02 Join Stummi [0] (~Stummi@rockbox/developer/Stummi)
22:56:34BuscheljhMikeS: my MEM change should not affect anything. that was just removing an unneeded thing
22:56:36kugeljhMikeS: I still maintain that's the main purpose of #error :)
22:57:27jhMikeSkugel: lol
22:57:30gevaertsBuschel: did you see the mail on the ML about MEM in gzip?
22:58:16jhMikeSkugel: erm...ok...I put it there where it normally wouldn't be an error, just a sanity check
22:58:22 Quit w0m (Ping timeout: 246 seconds)
22:58:45 Join GeekShadow [0] (
22:58:46 Quit GeekShadow (Changing host)
22:58:46 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
23:01:15Buschelgevaerts: no, thanks for the heads up
23:01:33*jhMikeS just wants to dump that stupid hack forever but has no access to the particular target to investigate
23:01:34Buschelwill undo the change in this file
23:03:01CIA-7New commit by Buschel (r29195): Revert nonsense change.
23:03:59 Join evilnick_B [0] (0c140464@rockbox/staff/evilnick)
23:04:31 Join w0m [0] (~wom@
23:04:49jhMikeS(hint, hint: who can mail me such a device? :) I thought there was one making the rounds
23:04:49gevaertskugel: success!
23:05:07gevaerts(checking for deleted class files)
23:07:38 Quit TheLemonMan (Quit: free(me))
23:07:52CIA-7r29195 build result: All green
23:08:24 Join TheLemonMan [0] (~lem0n@
23:08:46gevaertskugel: do you want to commit this, or shall I?
23:09:50kugelyour work, your commit :)
23:14:05CIA-7New commit by gevaerts (r29196): Change the way java files are compiled (do them all at once) in order to speed up builds. There is now an intermediate jar file to help (although ...
23:14:17gevaertskugel: can you test if it is indeed faster for you?
23:16:19kugelvery much, great!
23:17:24gevaertsNow write 100 times: "Build systems are easy if you concentrate a bit!" :)
23:17:48CIA-7r29196 build result: All green
23:18:26kugel40s vs 3.3
23:23:56pixelmajhMikeS: the developer 80GB Video vanished with Nico_P it seems
23:25:00 Quit TheLemonMan (Quit: free(me))
23:27:16kugelsorry, 23.5s vs 3.3s :)
23:27:57 Join quem_ [0] (
23:32:35 Quit quem (Ping timeout: 276 seconds)
23:32:35 Nick quem_ is now known as quem (
23:35:54 Quit parafin (Remote host closed the connection)
23:36:05 Join parafin [0] (
23:45:00 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)
23:47:47***Saving seen data "./dancer.seen"
23:48:54 Join bluefoxx [0] (
23:49:40 Quit bluefoxx (Client Quit)
23:49:59 Join bluefoxx [0] (
23:52:30 Quit bluefoxx (Client Quit)
23:52:38 Quit bluefoxx_ (Ping timeout: 255 seconds)
23:53:05 Quit Horscht (Quit: Verlassend)
23:58:59 Join bluefoxx [0] (

Previous day | Next day