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 2009-03-02

00:00:31*amiconn seems to have working automatic overlay placement in the build system :D
00:01:13amiconnNo more, and fairly easy to use for other lowmem targets
00:03:44 Join killan_ [0] (
00:04:34 Quit killan (Read error: 104 (Connection reset by peer))
00:04:40 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
00:04:57 Quit daurn (Remote closed the connection)
00:05:14 Quit daurnimator ("ZNC by prozac -")
00:05:14 Quit bmbl ("Woah!")
00:05:51bluebrotherkugel: why does the table in the r20157 not follow the style in the manual? I.e. use booktabs and why is the caption above the table?
00:06:38kugelthe caption is to explain that this is an example
00:07:00bluebrotheryou're aware of
00:07:31bluebrothercaption should go below. Though we don't use captions on tables usually.
00:07:53bluebrotheralso, the booktabs package is used for lines. Using hline is bad
00:07:54kugeluhm no
00:08:01*bluebrother hmpfs
00:08:02rasherbluebrother: Mind taking a look at FS #9934?
00:08:18kugelI googled latex table, I didn't know of this wiki page
00:09:06bluebrotherwe do have even more wiki pages on the manual. Like: how to build
00:09:30bluebrotherthat would've been ManualHowto (which links LatexGuidelines btw)
00:09:51kugelyea, I know ManualHowto, but I apparently didn't read it to the end
00:10:50bluebrotherbesides, there is no need to add a caption just to indicate the table is an example. Just put it into a sentence ...
00:11:20*kugel doesn't see what's bad about captions
00:12:11 Quit Llorean ("Leaving.")
00:12:12bluebrotherit's out of place. We want a consistent style. Feel free to add captions to all tables ...
00:12:29 Join Llorean [0] (
00:12:33gevaertsrasher: short summary : on my first try the OF booted but rockbox didn't (filesystem and/or disk issues), so I reformatted the disk and now rockbox boots and the OF says "RETRY TO SYNCHRONIZE" (I may not have copied all the right files back).
00:12:34bluebrotherplus, giving a table a number isn't in place if you don't reference it. The only reasoning would be if all tables had a number
00:14:15bluebrotherwhich is not the case for the manual
00:14:28 Quit Seed ("cu, Andre")
00:14:58 Nick fxb__ is now known as fxb (
00:15:18rashergevaerts: Sounds a bit like the bootloader works, but the OF doesn't?
00:15:23gevaertsrasher: yes
00:15:55rasherAnd usb plug in?
00:16:01kugelbluebrother: using the booktabs example, it doesn't quite look like the other tables
00:16:16*gevaerts now removed more files and he gets "NO FONT FILE"
00:16:52rasherSounds like your mrobe isn't doing too good
00:17:33rashergevaerts: if you'd add your findings to FS #9955, that'd be grand, just for completeness
00:18:06 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
00:18:06 Quit pixelma (Nick collision from services.)
00:18:06 Quit amiconn (Nick collision from services.)
00:18:07 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
00:18:18 Join Seed [0] (
00:18:18 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
00:18:24 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
00:20:19 Join DC1 [0] (
00:21:04bluebrotherkugel: I don't know what you're doing but this looks way better to me:
00:22:33rasherI haven't looked at what kugel did, but that does look nice
00:22:47rasherbluebrother: any ideas about my issues with the viewports-in-manual patch?
00:23:02bluebrotherrasher: haven't gotten around looking into it yet.
00:23:12kugelbluebrother: cool, you want to commit?
00:23:15*bluebrother wonders if kugel is going to fix it or if he should
00:23:20bluebrotherok, going to commit
00:23:34kugelwell, I'm playing around with it, but you seem to have it properly already
00:23:59amiconnMeh, all those subtle things which are rather important in Makefiles :\
00:24:21*amiconn ran into the difference between $^ and $+
00:25:25 Quit Actium (Connection reset by peer)
00:27:26pixelmaI wouldn't understand the sorting of the 03-thing in the first column of the example... didn't Llorean say something like this too (respect leading 0s)?
00:28:08rasher03 means "bigger than 2 and smaller than 4"
00:28:13rasherIn natcasesort terms
00:28:52pixelmaI understand where it comes from but looking at that list I don't think that's intuitive
00:29:34rasherI think it makes perfect sense
00:30:47bluebrotherI was wondering how that could be presented better too.
00:31:08linuxstbShouldn't the second "As whole numbers" be "As digits" ? Also, I wouldn't use the word "opposes" - maybe just "demonstrates"
00:31:58kugellinuxstb: oops, yes
00:31:59rasherAlso "this numbers" is bad grammar
00:32:35linuxstbIs our style to say "file names", not "filenames" ?
00:33:29bluebrotherhmm, that setting needs also to get added to the config file options
00:33:54*linuxstb spots an inconsistent use of both
00:34:09rasherlinuxstb: Oh noes.
00:34:25pixelmabluebrother: my remark is not about presentation, it's about what it does... wouldn't that mean that my 01, 02..10. named files are sorted the wrong way now (if there are more than 10)?
00:34:50linuxstbrasher: I know.... We should withdraw the manual from circulation...
00:35:18rasherpixelma: I don't see why?
00:35:50kugelpixelma: no, they still sort the same
00:36:28kugelbluebrother: I will do that
00:36:30***Saving seen data "./dancer.seen"
00:36:34kugellinuxstb: is filenames better?
00:37:06bluebrotherpixelma: why? 01 is still less than 10
00:37:07pixelmaok, I see it now. But I still find the 03 in this example list misplaced
00:37:37linuxstbkugel: Yes, that's what I would say, and is more common in the manual. There are a few instances of "file name" already though.
00:37:42rasherI think it makes sense to have it there
00:37:54 Join midijunkie [0] (
00:37:54bluebrotherwell, I was thinking of a textual description instead of a table −− it's somewhat confusing that both variants are alongside
00:38:27kugelHow that?
00:38:46bluebrothersomething like "If you have the file 01, 02, 3, 04 they will get sorted as 01 02 04 3 / 01 02 3 04"
00:39:00kugelI don't think the table is confusing at all
00:39:00rasherThat's hard to follow
00:39:15bluebrothermaybe choose other numbers and also include some trailing numbers example
00:39:28pixelmarasher: well, I read the 0 first and can't ignore it
00:39:47bluebrotherI'm not sure if the table is easy to follow either ...
00:40:01*rasher stands in defense of the table
00:41:23rasherHaving it inline makes it harder to get the big picture
00:42:33kugelso, change "oppose" to "demonstrate" too?
00:43:06kugelthe table is easy to follow, imo, particularly because it quite matches the list view
00:44:39kugellinuxstb: ?
00:45:02 Quit bs66_ (Read error: 110 (Connection timed out))
00:45:28rasherkugel: Change "this numbers" into something else
00:45:49 Join bs66_ [0] (
00:46:13kugelmaybe just use singular it reference to the number with it?
00:46:16kugel"it", you know
00:46:48kugel"It combines consecutive digits to number and uses it for sorting."
00:46:53kugela number*
00:46:55saratogakugel: yes i've been following that patch, but doesn't jhMikeS's compliant still apply?
00:47:14saratogalinuxstb: I can easily disable that code, I just assumed it was part of the rockbox metadata parsing
00:47:21kugelsaratoga: well, I don't know, I haven't experienced that issue yet
00:47:53bluebrotherrasher: I'll try to look into the viewports patch tomorrow. gtg now, it's quite late already.
00:48:12rasherbluebrother: Great
00:48:23kugelsaratoga: or do you just mean to add lcd_set_enable_hook
00:48:25 Quit bluebrother ("cya")
00:48:51kugelwell, that's not major, since the wps is update very frequently anyway
00:50:43saratogakugel: I'm not sure what he meant, I don't know anything about the WPS
00:51:00kugelrasher: "It combines consecutive digits to a number used for the sorting."
00:51:12rashers/the //
00:52:25kugelsaratoga: I think he was mainly saying, that upon enabling the lcd, the wps might show the old (unupdated) wps.
00:52:59kugeland this is true (for the e200 at least), but only for less than a tick, and I actually think it's a nice indicator that the thing is working
00:53:39Lloreankugel: But it looks like a bug to other people
00:53:44linuxstbsaratoga: No, Rockbox does its own metadata parsing - it's all in apps/metadata/ I had a quick look at Tremor's code, and it's just nasty - using bit-reading functions to read the metadata strings 8 bits at a time...
00:53:50rasherkugel: You don't want to *see* it working.
00:54:26kugelI want, but other probably not, yes
00:54:46linuxstbCan't you tell it's working without that?
00:55:46kugelI'm going to try if that hook even makes a difference. I expect it would still not update until the next tick
00:56:02 Join tajlero [0] (
00:56:28 Quit tajlero (Read error: 104 (Connection reset by peer))
00:56:55 Quit ender` (" Never trust a computer you can't throw out a window. -- Steve Wozniak")
00:59:57saratogalinuxstb: yes the bitstream parsing code in tremor is an absolute mess
01:00:17 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
01:00:19saratogamost of the optimizations added by the tremorlo guy involve rewriting it to be saner
01:01:23linuxstbI've long thought that adding an optimised set of bitstream reading functions to the codec api would be useful - so they read bits directly from the audio buffer.
01:01:48saratogalinuxstb: me too! infact I added it to mrsomeones list
01:02:06linuxstb;) Let's hope he gets round to it...
01:02:10saratogaoptimized huffmand decoding would be nice too
01:02:23saratogarather then 10 different implementations on ten different codecs
01:03:52saratogalinuxstb: i've been meaning to ask what the difference between read_filebuf and request_buffer is
01:03:56linuxstbBut yes, it would be good to get rid of that vorbis comments code in Tremor - it's just wasting RAM.
01:04:21linuxstbread_filebuf does a memcpy, request_buffer returns a pointer into the audio buffer.
01:04:42linuxstbAnd then you use advance_buffer() when you've processed the data returned by request_buffer.
01:05:24*amiconn likes that his overlay auto placement work also fixes the breakage of some plugins for cygwin sim builds
01:05:38linuxstbrequest_buffer() is only guaranteed to return up to 32KB of memory (GUARD_BUFFER_SIZE) - the read function may handle more (I've never used it).
01:05:54linuxstbWhat's Tremor using?
01:06:07amiconnThe problem was that gcc-support.o was being linked explicitly instead of using libplugin.a, causing empty plugins to break
01:06:11saratogai have no idea
01:06:15*linuxstb sees it's using read_filebuf...
01:06:28saratogai was just wondering which i should use to replace fopen/fread in the tta decoder
01:07:03linuxstbIdeally you should use request_buffer(), but that may need you to go deeper into the decoder - i.e. remove the existing API.
01:07:47saratogawhats teh advantage?
01:08:02saratogaoh no memcpy?
01:08:55saratogaout of curiosity, do you think it would be easy to make read_filebuf safe to use on COP if its doing a memcpy anyway?
01:10:58linuxstbI don't know... I can't think of any obvious problems.
01:17:25 Quit n1s ("Lmnar")
01:17:44 Join Rabbit^^ [0] (n=Rabbit^^
01:24:51 Quit markun ("leaving")
01:35:40 Join FlynDice [0] (
01:37:21 Quit flydutch ("/* empty */")
01:39:17amiconnMeh. Why does the build system not pick up the fix?
01:41:44 Join midgey [0] (n=tjross@
01:43:33 Join nuonguy [0] (
01:47:13saratogalinuxstb: is it neat enough if I just make parsing do nothing?
01:47:23saratogai'm not sure how to advance the bitstream other then by reading it
01:47:26 Quit Nico_P (Remote closed the connection)
01:48:45linuxstbThere's oggpack_adv(), but that can only do up to 32 bits at a time (see ogg.h). I can't see any better way though - that bitstream code is incomprehensible to me.
01:49:56saratogaactually I may have a way
01:50:04linuxstbBut I think the important thing is to remove the calls to calloc, I think we can live with still parsing the data....
01:50:37saratogait seems like it doesn't need the bitstream advanced anyway, it just needs some of the header filled out
01:51:21kadobanamiconn: the new overlay stuff shouldn't need any code changes, right?
01:51:27 Join Phletcher [0] (n=18eca8ed@gateway/web/cgi-irc/
01:51:37 Quit Phletcher (Client Quit)
01:51:51kadoban(in other words, the magic "don't overwrite myself" stuff still works if goban steals audio buffer?)
01:52:43amiconnIt just adjusts the overlay start address so that it fits just below the end of audio buffer
01:53:58kadobanahh, i see. looks nice, thanks for doing that
01:55:09amiconnDoes goban have some magic for ensuring that when running as overlay?
01:56:34kadobanyeah, it should be checking when it's passed a buffer to use in sgf_storage.c i believe
01:56:50kadobani believe i copied it mostly from rockboy, but with changes for some reason that i forgot
01:58:00amiconnAh, yes. It's in a different plays than where I looked
01:59:58saratogalinuxstb: I have a version that doesn't parse comments working, but looking at the code i'm not 100% sure it will work for all files
02:00:08saratogasince I don't update the file pointer
02:00:21saratogait doesn't seem to matter for my files, but maybe it does for others
02:00:30saratogashould I commit and see if anyone complains or put it on the tracker?
02:00:57linuxstbPut it on the tracker I think. Hopefully Lear will have time to look at it.
02:01:09linuxstb(I don't have any vorbis files at all...)
02:02:18saratogayeah I don't use vorbis either so its hard to test these things
02:02:58saratogaalso, did anyone ever object to that tremor malloc thing I posted a while ago?
02:03:04saratogai suppose I should commit it one of these days
02:05:07 Quit moos ("Time to sleep a bit!")
02:06:09kugelsaratoga: I've not seen a single complaint
02:08:24saratogahmm anyone object if I pull the trigger tonight then?
02:10:22*amiconn found the reason why doom, midi and reversi need the linker instructions duplicated in their respective .make file
02:10:56amiconnThose plugins don't have an object file named <plugin_name>.o, and that's a prerequisite for the rule in plugin.make
02:11:19amiconnAll other subdir plugins do, and hence it works for them
02:15:10 Nick fxb is now known as fxb__ (
02:21:40 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
02:23:02Unhelpfulmaybe d_main.c could be named doom.c?
02:23:30amiconnThe proper fix would be changing the .make file(s) to not require that
02:24:01amiconnBut understanding 'make' requires a lot of nonlinear thinking...
02:24:02kadobani actually ran into that a while ago, but i didn't notice that others had worked around it...i just changed my main file name
02:25:50Unhelpfulamiconn: i think that specifying the name of the linked output in a make var, and having a generic rule making that var depend on the objects built from the listed sources, might work
02:31:12amiconnIt's more complex than that
02:31:55amiconn$(ROCKS) lists all single file plugins as well as subdir plugins (consisting of multiple objects, not necessarily including <plugin_name>.o)
02:32:40amiconnSo %.rock matches either type - but it must only have %.o as prerequisite if there aren't other prerequisites with the extension .o
02:33:24amiconnI wonder whether it is possible to have a rule defining a variable that is used as a prerequisite for another ruke
02:34:39Unhelpfulcan variables be defined *in* rules?
02:34:56*amiconn doesn't know yet
02:35:53amiconnVariables and function calls are looking (almost) the same
02:36:29Unhelpfulthey pretty much are, aren't they, to make?
02:36:32 Join itcheg [0] (i=62db4767@gateway/web/ajax/
02:36:34***Saving seen data "./dancer.seen"
02:36:55amiconnIt might be possible to have a function call in a prerequisite. The question is whether this function can access the prereqisite list in some way
02:41:42*amiconn found .SECONDEXPANSION
02:42:16 Join zman977 [0] (n=483043c8@gateway/web/cgi-irc/
02:43:40 Quit itcheg (" ajax IRC Client")
02:47:20 Quit zman977 ("CGI:IRC")
02:48:01 Quit CaptainKewl ("( :: NoNameScript 4.02 :: )")
02:49:37 Quit lunix151 ("Leaving.")
02:57:26 Quit DC1 (Read error: 104 (Connection reset by peer))
03:05:23 Quit dberg918 ("Ex-Chat")
03:08:43kugelmidgey: hm, seeing your last commit, why didn't that cause problems?
03:13:11 Quit BHSPitMonkey (
03:13:11 Quit HellDragon (
03:13:11 Quit krazykit (
03:13:11 Quit soap (
03:13:11 Quit lightbulbjim (
03:13:11 Quit keby (
03:13:11 Quit advcomp2019 (
03:13:27NJoinBHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
03:13:27 Join HellDragon [0] (n=jd@Wikipedia/HellDragon)
03:13:27NJoinkrazykit [0] (
03:13:27NJoinadvcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
03:13:27NJoinlightbulbjim [0] (
03:13:27NJoinkeby [0] (
03:13:27NJoinsoap [50] (n=soap@rockbox/staff/soap)
03:17:17 Quit _Auron_ (Read error: 104 (Connection reset by peer))
03:18:15midgeykugel: mpeg.c would only affect HWCODEC targets, and i think the only change only affects creating new directories
03:18:47 Join _Auron_ [0] (
03:19:50 Quit gevaerts (Nick collision from services.)
03:20:02 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
03:22:12midgeykugel: are you sure your last commit is right? I think you actually forgot to add the code changes from that task
03:23:33 Join fyrestorm [0] (
03:27:46kugelmidgey: ah, lol
03:28:24kugelmidgey: I wonder how that happened. the patch command is in my bash history
03:32:09 Join jordoex [0] (
03:34:53 Quit Aurix_Lexico ("Leaving.")
03:45:08 Join CaptainKewl [0] (
03:46:39saratogahmm seems i've made some nasty assumptions about how rockbox codecs init memory in my vorbis malloc patch
03:46:59 Quit Rabbit^^ (Remote closed the connection)
03:52:49saratogamy patch dmalloc will actually fail if two consecutive vorbis tracks are played because the codec will try to reinit the malloc with the memory thats already inited in it
03:53:07saratogai can easily prevent it from calling init every time, however I wonder if this is the right thing to do
03:53:31saratogaout current malloc implementation simply erases the malloc state every track so that memory leaks are impossible
03:54:02 Quit HackyKid (Read error: 110 (Connection timed out))
03:54:18saratogaI'm not sure if I should change this behavior
03:55:18linuxstbWhat's the problem? Can't you just reset the malloc and call the vorbis init() function again, as if the codec has just started?
03:55:48saratogalinuxstb: theres no function that I can see to reset the malloc
03:56:04linuxstbThen I guess you'll need to write one...
03:56:18saratogaeasier said then done
03:56:37linuxstbBut isn't there a function to call before you use it the first time?
03:57:27saratogalinuxstb: yes, however it can be called over and over to add more areas of memory
03:57:40saratogaand it does not initialize, it simply adds more blocks to the pool
03:57:59linuxstbAh, OK.
03:58:32Unhelpfulsurely a newly-added block is initialized? maybe you could call whatever code dmalloc uses to do that?
04:00:06saratogai wonder if I can just null out the "blockHead" variable
04:01:47 Join thegeek_ [0] (
04:02:41 Quit fyrestorm ("ChatZilla 0.9.84 [Firefox 3.0.6/2009011913]")
04:04:28saratogapreventing it from reinitializing the malloc buffer doesn't actually prevent the crash, so something else must be going on too
04:06:06linuxstbHow does it crash?
04:06:19 Join Darksair [0] (n=user@
04:08:34saratogalinuxstb: segmentation fault in the sim, hard freeze on device
04:09:02saratogai found one crash related to calling codec_init again, but theres apparently yet another
04:11:28 Join blkhawk- [0] (
04:13:16 Quit perrikwp (" ajax IRC Client")
04:13:37 Quit thegeek (Read error: 110 (Connection timed out))
04:14:05 Join midijunkie41 [0] (
04:18:21 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
04:20:28 Quit midijunkie41 ("?(???~~)?")
04:20:51 Join toffe82_ [0] (
04:23:07 Quit toffe82 (Read error: 110 (Connection timed out))
04:23:09 Join thegeek [0] (
04:23:19 Nick toffe82_ is now known as toffe82 (
04:23:58Unhelpfulthought i'd not hit the beast restore-after-install issue, despite it having apparently updated flash... but, restarted rockbox, and look, restore sequence.
04:26:04 Quit Darksair ("Use the Force, Luke!")
04:26:28Unhelpfulperrikwp: yet we know that processing with mknkboot produces a working dual-boot, so clearly some of what we know about getting code onto beast still applies :/
04:27:31Unhelpfulwhat happened for you with single-boot? right now, it seems to be hanging with an empty progress bar, for me. results have not been very consistent.
04:27:31 Quit blkhawk (Read error: 110 (Connection timed out))
04:28:27 Nick blkhawk- is now known as blkhawk (
04:29:32 Quit midijunkie (Read error: 110 (Connection timed out))
04:30:32 Quit SirFunk (Read error: 60 (Operation timed out))
04:30:36 Join SirFunk_ [0] (
04:33:23 Quit perrikwp (" ajax IRC Client")
04:33:38 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
04:34:12perrikwpUnhelpful: Sorry, my browser crashed
04:35:39perrikwpwhen I use the single-bootloader, and restart the gigabeat it asks for a firmware restore or upgrade
04:36:00perrikwpbut the dual-bootloader works perfect
04:36:37***Saving seen data "./dancer.seen"
04:37:06perrikwpmy gigabeat S original firmware was version 1.2 if that makes any difference
04:37:15 Quit thegeek (Read error: 60 (Operation timed out))
04:37:23 Join thegeek [0] (
04:37:41Unhelpfulah, i only upgraded to it. :/
04:39:56perrikwpwell when I got the gigabeat it wouldn't start so I restored it with the firmware link available on the wiki, so I guess I upgraded to 1.2 also
04:40:47 Quit thegeek_ (Read error: 110 (Connection timed out))
04:43:35Unhelpfulhrm, so we don't know what yours had before. did you restore using the toshiba upgrader app, or just sendfirm?
04:44:09perrikwptoshiba upgrader app
04:44:57perrikwpI think it originally had 1.1, because thats what Win 7 reported the firmware version was when I looked at the Gigabeat S properties
04:45:21perrikwpbefore I restored
04:51:10Unhelpfuland now single-boot works fine for me. i guess i'm no help diagnosing this. wonder why yours is trouble?
04:52:15perrikwpi have no idea, but I am happy that atleast the dual-bootloader works
04:52:34perrikwpdoes the single-boot load faster?
04:56:46perrikwpthat's the only reason I really do want to be able to use the single-bootloader
04:56:55 Quit miepchen^schlaf (Read error: 101 (Network is unreachable))
04:59:09 Join nothingman [0] (n=chatzill@
04:59:12nothingmanhi, all!
04:59:25nothingmanhi, all!
04:59:27nothingmanmy zune has decided to spring to life, and so I ask you what's new in zune/rockbox development
05:01:19saratogano one has done anything
05:06:12nothingmanOK, good to know
05:08:27 Quit rocko (Remote closed the connection)
05:08:40saratogaalso if you want to keep tabs, just check the wiki and forum thread
05:09:51 Quit HBK ()
05:15:39 Quit _Auron_ ("Infinity repeatedly denies rumours of plotting with zero to bring down the Universe.")
05:21:40 Quit goffa_ (Read error: 54 (Connection reset by peer))
05:23:18 Quit CaptainKewl (Remote closed the connection)
05:24:04 Join CaptainKewl [0] (
05:24:28 Join goffa [0] (n=goffa@
05:34:54nothingmanwell, if nothing in two years, I figure it's not going to happen
05:36:03krazykitwell, if everyone who wants progress does nothing, then it's not going to happen
05:41:22nothingmanI'm not pointing fingers, just saying
05:41:44nothingmanI only just got this one second-hand, so unless there's someone already doing something, I'm a bit late to help
05:42:10nothingman<−−no programming, no hacking, would like to keep a working device rather than meddle
05:44:44nothingmanbut if there's some development somewhere, even old, looking for someone with a zune to test, I'd be willing
05:46:43 Quit Horscht ("Verlassend")
05:57:55 Quit kugel ("ChatZilla 0.9.84-rdmsoft [XULRunner]")
06:04:50 Join rocko [0] (
06:10:59 Part rocko ("Leaving")
06:24:22 Quit saratoga ("CGI:IRC (EOF)")
06:25:00 Part toffe82
06:33:09 Join Darksair [0] (n=user@
06:36:38***Saving seen data "./dancer.seen"
06:40:31 Join proudfoot [0] (
06:40:45proudfootanyone know if rockbox is doing gsoc this year?
06:41:07 Quit CaptainKewl (Read error: 54 (Connection reset by peer))
06:41:08scorchewe are planning on it, or i wouldnt have suggested it ;)
06:41:34proudfoottrue true
07:26:11 Join _lifeless [0] (n=lifeless@
07:39:20 Quit __lifeless (Read error: 110 (Connection timed out))
07:43:08 Join MethoS- [0] (
07:49:57*amiconn wakes up, and noticed that he seems to have caused some weird breakage in plugin building
07:49:58 Quit _lifeless (Read error: 110 (Connection timed out))
07:50:53amiconnIt only happens on, and it happens for a plugin that I did have to take special care of :(
07:53:34midgeyjust have bagder disable that server - problem solved ;)
07:53:52 Join _lifeless [0] (n=lifeless@
07:57:37 Quit _lifeless (Remote closed the connection)
07:57:54 Join _lifeless [0] (n=lifeless@
08:01:56amiconnIt only happens for target builds on fnarfbargle, but consistently, and I am unable to reproduce it here no matter what -j option I use
08:02:41 Join Rob2222 [0] (
08:03:10amiconnI wonder what version of gnu make fnarfbargle is using. Maybe it doesn't understand $+ properly?
08:07:02 Quit _lifeless (Remote closed the connection)
08:07:25 Join _lifeless [0] (n=lifeless@
08:10:35 Quit z35 (Read error: 60 (Operation timed out))
08:12:39 Join __lifeless [0] (n=lifeless@
08:13:35 Join ender` [0] (
08:15:07 Join MethoS-- [0] (
08:16:40amiconnThe order in which things are built on fnarfbargle is correct...
08:17:59*amiconn summons one of the 3 swedes, or optionally whoever operates
08:20:07 Quit Rob2223 (Read error: 110 (Connection timed out))
08:23:42 Join markun [50] (n=markun@rockbox/developer/markun)
08:26:28 Quit _lifeless (Read error: 110 (Connection timed out))
08:30:30 Join nate3000 [0] (
08:33:52 Quit MethoS-- (Read error: 110 (Connection timed out))
08:34:02 Join z35 [0] (
08:34:14 Join MethoS-- [0] (
08:34:24 Quit MethoS- (Read error: 110 (Connection timed out))
08:36:01 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
08:36:42***Saving seen data "./dancer.seen"
08:37:18amiconnhi Zagor
08:37:56amiconnDo you know which version of 'make' uses? Or what odd condition might trigger the build error on it?
08:40:10Zagorno I don't. "funny" how everything works except memset and memcpy
08:40:13amiconnIt seems like the problem is caused by a change of mine in plugin.make, but I can't reproduce it at all nomatter what -j option I use. The link commands for that target are correct, checked with make V=1 and only wav2wv.rock deleted
08:41:12amiconnZagor: It seems to be a library linking order problem, but my plugins.make change has something to specifically dela with that.
08:41:38 Join midijunkie [0] (
08:41:38amiconnBasically, gcc-support.o was being linked explicitly before, which was nasty, and libplugin.a contains it
08:42:22amiconnBut wav2wv.rock also uses libwavpack, and if the plugin code itself doesn't need memset/memcpy but libwavpack does, libplugin must be listed again after libwavpack
08:43:11amiconnThis is dealt with by adding libplugin.a a second time in the special prerequisite rule for wav2wv.rock, and changing the $^ in the command with $+
08:43:37amiconnHence I guess it's a problem with 'make', and that's why I was asking about the version
08:45:46amiconnBut then, gnu 'make' 3.80 and even 3.79 support $+, so either it's not gnu make, or there's an odd bug hitting us
08:47:27 Join bertrik [0] (
08:47:59amiconnZagor: Btw, if you check the log - I have found the reason why doom, midi and reversi don't match the %.rock rule in plugins.make and need their commands duplicated into their respective .make file
08:48:28Zagorcan you point me to a time?
08:48:37 Quit midgey ()
08:48:58 Quit midijunkie (Read error: 104 (Connection reset by peer))
08:49:06amiconnlast night starting 02:10. I don't have a solution yet.
08:51:58 Quit BHSPitMonkey (Remote closed the connection)
08:51:59Zagorah, how simple. good catch.
08:53:33 Join petur [50] (n=petur@rockbox/developer/petur)
08:53:51amiconnA possible workaround would be to make each of those plugins have an object file named <plugin_name>.o
08:54:28amiconnBut it would be better to rewrite the rules so that %.o is only added to the prerequisites if there are no ther .o files added yet
08:54:35amiconn*the rule
08:56:40 Quit perrikwp (" ajax IRC Client")
08:56:47Zagoryeah fixing the rule feels better
08:57:05amiconnYeah but how?
08:57:43*amiconn would also like to know the cause why his change breaks on, but has to go to work soon :\\
08:58:19ZagorI think we'll just have to wait until whoever runs it comes in
08:59:21 Join Rob2223 [0] (
09:00:54 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
09:14:38Zagoramiconn: simply removing %.o from the plugin link prequisite list should work
09:15:57amiconnIt would work for the subdir plugins, but might break single-file plugins
09:16:12 Join Bagderr [241] (n=daniel@rockbox/developer/bagder)
09:16:38 Nick Bagderr is now known as B4gder (n=daniel@rockbox/developer/bagder)
09:17:26amiconnThe .o must be built before the .rock can be built for those
09:17:27 Quit Rob2222 (Read error: 110 (Connection timed out))
09:18:41amiconnUnfortunately a pattern rule without commands overrides another rule, that's why I was thinking about the .SECONDEXPANSION magic
09:18:56amiconnBut there might be a simpler solution - splitting $(ROCKS) in two
09:24:18Zagorthat would duplicate the link lines
09:25:45amiconnIiuc you can have as many rules as you want for each target, as long as only one actually has commands
09:26:02Zagorcorrect. perhaps I misunderstood what you meant.
09:27:02amiconnSo we could have $(ROCKS) and $(ROCKS1), and only have the dependency on %.o for the $(ROCKS1) target, but the linking rule (with commands) for $(ROCKS) $(ROCKS1)
09:28:09 Quit bertrik (Remote closed the connection)
09:28:12amiconnHmm, that won't work as a pattern rule for $(ROCKS1)
09:29:13B4gderoh man there was some fun blame game tonight! ;-)
09:29:34amiconnBut even if we have to duplicate the rule, it's better than having it 4 times (as it is now)
09:29:50B4gderI guess I need to disable that server
09:30:42amiconnI'd really like to find out what's going on there.
09:31:16 Quit z35 ("Leaving")
09:33:57 Join Thundercloud [0] (n=thunderc@
09:40:42 Quit jordoex (Read error: 60 (Operation timed out))
09:42:01 Quit nate3000 ("using sirc version 2.211+KSIRC/1.3.12")
09:45:10 Quit Thundercloud (Remote closed the connection)
10:14:11 Join HackyKid [0] (
10:22:09linuxstbB4gder: There also seems something odd on - it's producing modified r20098 builds..
10:23:04 Quit kachna (Read error: 113 (No route to host))
10:24:16linuxstbHow? You fixed the server, or just removed it?
10:25:39B4gderI fixed it
10:25:47B4gderit was just a locally modified
10:26:01B4gder$ svnversion
10:26:33linuxstbOK, good ;)
10:36:46***Saving seen data "./dancer.seen"
10:41:48 Join daurnimator [0] (n=quae@unaffiliated/daurnimator)
10:45:06pixelmaaren't there some build servers that needed a modification to use the right shell or so?
10:47:39 Quit ender` (Read error: 60 (Operation timed out))
10:48:15B4gderI don't think so
10:48:33B4gderand if there would be any such, we should rather fix the build to not rely on any particular
10:48:54B4gderwe had bashisms in the past but I believe they're (mostly) fixed
10:53:23 Quit HackyKid (Read error: 110 (Connection timed out))
10:54:50 Quit thegeek (Read error: 104 (Connection reset by peer))
10:55:10 Join thegeek [0] (
10:57:53 Join Munkie [0] (n=29b13251@gateway/web/cgi-irc/
10:58:12Munkiewhen is 3.2 coming out
10:59:30B4gderMunkie: end of march presumably
10:59:57Munkieok cool thanks
11:00:18Munkieis there anywhere i can look what is being changed/updated
11:03:09scorchethere is the SVN log...
11:03:21B4gderand the MajorChanges wiki
11:03:29scorches/wiki/wiki page
11:13:15*B4gder stands corrected
11:13:23 Join kachna [0] (
11:14:17 Quit Darksair (Read error: 104 (Connection reset by peer))
11:15:10 Join gartral [0] (
11:21:41gartralok, trying too set up a build environment on ubuntu 8.10 (Intrepid Ibex) and iv'e hit a small snag with the automated method ( the instructions are clear enough, but i think i can just sudo it and make it work, or am i wrong?
11:22:13B4gderyes, if you want that install path you need to sudo
11:22:59gartraloook, i have no problem with the paths it wants, i just needed too make sure im not going to bork the entire thing
11:23:28B4gderwe offer a full money-back guarantee if that would happen!
11:24:43gartralhahaha... perhaps a better phrase is: "make sure whatever hapens is user error, and not a known bug/flaw"
11:25:09B4gderright, it should be ok
11:25:33gartralits working..
11:28:27 Quit daurnimator (Read error: 60 (Operation timed out))
11:29:14 Join midijunkie [0] (
11:29:35 Join PaulJam [0] (
11:29:50 Join daurnimator [0] (
11:31:59 Quit Munkie ("CGI:IRC")
11:37:01 Join robin0800 [0] (
11:43:47 Quit PaulJam (".")
11:44:17 Quit markun (Read error: 104 (Connection reset by peer))
11:45:06 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
11:46:09 Join leftright [0] (n=50bb6915@gateway/web/cgi-irc/
11:47:24 Quit Seed ("cu, Andre")
11:47:48 Join Seed [0] (
11:48:39 Join markun [50] (n=markun@rockbox/developer/markun)
11:49:53 Join numeric_file_ord [0] (n=50bb6915@gateway/web/cgi-irc/
11:51:31 Quit numeric_file_ord (Client Quit)
11:51:37 Quit leftright ("CGI:IRC (Ping timeout)")
11:53:37 Join numric_file_sort [0] (n=50bb6915@gateway/web/cgi-irc/
11:55:20 Quit markun (Read error: 104 (Connection reset by peer))
11:55:38 Quit numric_file_sort (Client Quit)
11:56:22 Join numric_file_sort [0] (n=50bb6915@gateway/web/cgi-irc/
11:56:47 Quit numric_file_sort (Client Quit)
11:57:22 Join numeric_file_sor [0] (n=50bb6915@gateway/web/cgi-irc/
11:58:38 Quit numeric_file_sor (Client Quit)
11:58:51 Join moos [0] (i=Mustapha@rockbox/staff/moos)
12:05:18 Join PJ_B [0] (n=54bef574@gateway/web/cgi-irc/
12:11:14 Quit gibbon_ (Remote closed the connection)
12:12:44 Quit PJ_B ("CGI:IRC (EOF)")
12:13:26gartralits complaining that it cant find arm-elf-gcc or arm-elf-ld paths
12:13:46scorchedefine "it"
12:13:48B4gderand you set your PATH accordingly?
12:14:14gartralthe script set it, i thought
12:14:27scorchethe script explicitely tells you that you need to set it
12:14:37 Join gibbon_ [0] (
12:14:49scorchehence that message that it outputs when it completes
12:15:22gartralok, now im just a retard....
12:15:47B4gderadd /usr/local/arm-elf/bin to your PATH
12:16:06scorche(and the other compilers if you built those as well)
12:16:59gartralexaactly what path, or command do i need to change/execute to add that?
12:17:41gibbon_gartral: use the force of your shell... path refers to the PATH environment variable
12:23:04gartralok, i found some info, but the methods there talking about say i have to do it manually, is there a permanent way of adding to PATH?
12:24:55gartralscorche thats the same link -_-
12:25:24scorcheand what does the very first link that pops up say?
12:26:36gartralright, but it goes ON the say that it will only remain there for the duration of the console window, and will be returned to null...
12:29:30scorchethat is the second link in my results, however in that post as well, it goes on to say a method to add to the path so it does not do that...
12:29:39gartralnvm, somehow i scrolled down i link...
12:29:52gartrala link*
12:30:54 Join XxXSparkXxX [0] (n=7cbb11b6@gateway/web/cgi-irc/
12:31:41XxXSparkXxXhey all, does anyone know anythng about the gigabeatt S port?
12:33:02XxXSparkXxXmainly how to get USB charging to work
12:33:02pixelmawell, it exists ;) you are better off asking a specific question...
12:35:12 Quit gartral ("Leaving.")
12:36:49***Saving seen data "./dancer.seen"
12:37:15XxXSparkXxXdoes anyone know how to get USB charging working on the gigabeat S?
12:37:53 Join gartral [0] (
12:38:43 Quit nuonguy ("This computer has gone to sleep")
12:38:48BigBambiXxXSparkXxX: It already does
12:38:58BigBambiXxXSparkXxX: Although not in the bootloader
12:40:12gartralalright scorche now the tools/configure program is complaining about the very same thing
12:40:25 Join einhirn [0] (
12:40:30 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
12:40:44scorcheand how did you add the binary folders to your path?
12:41:04gartralvia the sessions method
12:41:17scorcheand have you logged out and back in again?
12:41:29*BigBambi bangs his head against the wall and screams
12:41:32gartralthats why i just dissapeared, yes
12:42:10 Join nibbler [0] (
12:42:20gartrali screwed up somewhere
12:42:56scorchewhat did you add to your PATH?
12:43:19gartralwhat you told me too
12:44:00BigBambiAnd what does your PATH say now?
12:44:31gartralone moment
12:48:01*scorche doesnt see /usr/local/arm-elf/bin in there anywhere
12:48:10gartralneither do i
12:48:23BigBambigartral: Then you did it wrong
12:48:35 Join jesse [0] (n=jesse@
12:48:43BigBambigartral: Really, you should be able to check for these things yourself without us prompting
12:49:11BigBambigartral: Also, this isn't a general linux help channel. Go and read how to set your PATH, then when you think you have done it, check if you actually have
12:52:31XxXSparkXxX@Bigbambi : it is plugged into my usb port of my PC, but its not charging.
12:52:56BigBambiXxXSparkXxX: Are you on Windows by any chance?
12:54:04BigBambiAh, then I don't think it will yet as it needs to present itself as a device properly and ask to charge
12:55:49XxXSparkXxXit says its a 2 partitioned hard drive tho, but in the sttings it is set to safe remove mode, not performance mode
12:55:58XxXSparkXxXas that got anything to do with it?
12:56:10BigBambiIt has got two partitions
12:56:25BigBambiand no, the number of partitions is not related
12:58:09XxXSparkXxXlet me clarify what i said, it picks the gigabeat s up as a HDD, but in its properties there is a setting for cache mode, would this have anthing to do with it?
12:58:30BigBambiNo, I don't think so
12:58:45BigBambiIt is to do with how the USB "presents" itself
12:58:59BigBambiBut I'm no expert, gevaerts could explqin it properly :)
12:59:09XxXSparkXxXso there is no fix for it?
12:59:43BigBambiYes, to implement it properly
12:59:57XxXSparkXxXok damn
13:00:23XxXSparkXxXi was going to use it as a external ard drive, but the batery wont lastlong enough
13:01:14BigBambiWhen connected in mass storage it should charge fine
13:01:24scorcheis there not a DC plug on the gigabeat S ?
13:01:34BigBambiIt is just only charging that doesn't currently work in Windows
13:01:38BigBambiscorche: yes there is
13:02:40XxXSparkXxXok, so i have to wait till someone finds how to fix it on windows
13:03:06scorcheor use the DC plug along with it...
13:03:27XxXSparkXxXcarrying that to school is a pain
13:03:55BigBambiXxXSparkXxX: Charging does work when using mass storage
13:04:20BigBambiIt may not be quite enough if the access is going constantly, so it will slowly drain
13:04:34BigBambiBut then there is nothing to be done about that, and you will need the charger anyway
13:05:17 Quit linuxstb (Read error: 113 (No route to host))
13:05:50XxXSparkXxXthank you for your time
13:05:54XxXSparkXxXgood bye
13:05:59 Quit XxXSparkXxX ("CGI:IRC")
13:06:12 Part gartral
13:07:11 Quit J-23 (Read error: 60 (Operation timed out))
13:07:47 Join MethoS- [0] (
13:10:29 Join J-23 [0] (
13:11:07 Quit MethoS-- (Read error: 60 (Operation timed out))
13:17:34B4gdera good read, relevant for us as well
13:25:45*B4gder puts new clip OF firmwares up on his site and is again reminded of the huge size differences between clip v1 and clip v2 firmware images
13:28:10 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
13:32:43 Join t0mas [0] (n=tomas@rockbox/developer/t0mas)
13:49:51 Quit nothingman (Read error: 110 (Connection timed out))
13:52:02 Part LinusN
14:05:43 Join tyfoo [0] (
14:06:25 Quit jesse ("Ex-Chat")
14:11:08 Quit kachna (Remote closed the connection)
14:11:19 Join kachna [0] (
14:15:26 Quit robin0800 (Remote closed the connection)
14:31:04 Join jaykay [0] (
14:36:50***Saving seen data "./dancer.seen"
14:56:42 Join markun [50] (n=markun@rockbox/developer/markun)
14:58:00jaykayis dual boot for the ipod mini 2g working properly?
14:58:54 Quit J-23 ("ZNC -")
14:59:05 Join J-23 [0] (
15:06:10 Join MethoS-- [0] (
15:07:00 Quit midijunkie (Read error: 104 (Connection reset by peer))
15:15:33 Quit GodEater (Client Quit)
15:22:08 Quit MethoS- (Read error: 110 (Connection timed out))
15:26:51 Quit linuxstb (Read error: 110 (Connection timed out))
15:32:07 Join Darksair [0] (n=user@
15:32:58 Join woyciesjes [0] (
15:35:03 Join robin0800 [0] (
15:36:03 Join PaulJam [0] (
15:38:40 Quit gevaerts (Nick collision from services.)
15:38:49 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
15:39:47 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
15:42:11 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
15:49:16 Quit robin0800 (Remote closed the connection)
15:51:31B4gderbtw "I don't know if you're still interested in MPC, but if someone can work on its RockBox codec, SV8 is finished and the SVN has the latest code"
15:51:44B4gder(/me got this in a privmsg)
15:54:48Zagormpc? as in motorola?
15:55:06Zagorah, the codec
15:55:24*Zagor crawls back under his cozy rock
15:55:33LloreanWe got a notification that SV8 was done shortly before they posted it on their site, too, in-channel.
15:55:43 Quit avis ("maybe when we drown, the fish will be our friends")
15:56:11LloreanSaratoga talked with a couple of them about it for a bit.
15:58:52 Join Killerbee [0] (
15:59:45KillerbeeHi.. i tried to install rockbox on my ipod video, but it gives me *PANIC* Unsupported physical sector size: 4096. Is this because it is a 30gb ipod upgraded with a 80gb harddrive?
16:00:24 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
16:02:35 Quit kachna (Read error: 110 (Connection timed out))
16:04:41jaykaykillerbee: which build did you donwload, for the 30gb or for the 80gb?
16:05:26Killerbeei used the 30gb, because it is not the 5gen, its the 5.5gen
16:05:56Killerbeei used the utillity, and it said my ipod was a 30gb
16:06:38 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
16:06:45 Quit perrikwp (" ajax IRC Client")
16:07:34pondlifeHello, all.
16:07:41 Part Llorean
16:08:00pondlifeAny objections to me committing (ttimestretch) ?
16:08:05Killerbeei read on another page, that someone had the same problems, but they couldnt solve them
16:08:23 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
16:08:29pondlifeIt will produce a red delta, but I don't think there any other major downsides
16:09:05Killerbeesry i didnt answer your question pondlife
16:09:05pondlifeI'd be grateful if someone who understands resampling (or dsp.c in general) could sanity check it though.
16:09:42*B4gder looks forward to play with timestretch!
16:10:10pondlifeSo... ok to commit? It's not the nicest-sounding thing (when enabled)... ;)
16:10:12Nico_PB4gder: could you svn up the webserver?
16:10:30B4gderah right
16:11:09pondlifeStill, you can somewhat enjoy listening to stuff with a high speed but low pitch.
16:11:18pondlifeMusic for whales
16:11:20 Part domonoky
16:12:10Killerbeecan you please help me get my ipod in disk mode again.. i cant get it to disk mode after i used the rockbox installer
16:12:33B4gderpondlife: I'm in the camp who's for committing stuff early
16:12:48B4gderespecially when it doesn't hit people who don't use the feature
16:12:53scorchecommit early, commit often? =P
16:13:22pondlifeOK, well I want it in and fixed before 3.2, so here goes. I've no idea exactly how red the delta will be, but hopefully less than the last attempt I made (and reverted).
16:14:41Nico_PI was playing with pitch and timestretch, then went to the file browser and got a corrupted file list
16:14:46Nico_Pthe player then froze
16:14:56pondlifeWas that with the latest patch?
16:15:08Nico_PI confess it was not
16:15:18pondlifeI did have a buffer pointer wrong untill the 27th (or thereabouts)
16:15:51pondlifeI've been running it here for the past 3 days with no lockups or oddness.
16:15:56Nico_Pmy version was compiled on 09 02 28
16:16:19pondlifeHmm, should have been ok then. Do you have the .patch file around?
16:16:37 Join LambdaCalculus37 [0] (n=rmenes@rockbox/staff/LambdaCalculus37)
16:16:42 Quit Killerbee ("Going offline, see ya! (")
16:16:54*pondlife checks tdspeed_090227.patch..
16:17:39pondlifeYes, the 27th includes the fix.
16:17:50Nico_PI was using that one
16:17:52pondlifeSo it might not be good to go after all.
16:18:31Nico_PI also have another patch though
16:18:36pondlifeAny particular type of "playing around"? I've tried the extremes.
16:19:22Nico_PI set high pitch and low speed
16:19:42pondlifeOK, will try that - have been mainly on high speed and high pitch.
16:19:51Nico_Pthen normal pitch, then I wanted to go into the settings to go back to normal speed and selected files instead
16:20:02pondlifeIt's possible to get buffer underruns but nothing worse that occasional short gaps
16:20:20pondlifedircache on?
16:20:30Nico_PI think so. let me check
16:21:16Nico_Pyes, and the disk did spin up when I selected "files"
16:21:25pondlifeAlso, check dithering/eq/crossfeed...
16:21:33Nico_Pnone are on
16:22:55pondlifeOK, just wondered. Will try your recipe now
16:25:00 Quit homielowe ()
16:26:17pondlife50% timestretch and 200% pitch sounds truly awful, and is hogging much CPU (with gapping), but no other nastiness...
16:28:22Nico_Psomething tells me it's not going to be easy to repro
16:28:30woyciesjesKillerbee: Can your iTunes see the iPod?
16:30:33Nico_Ppondlife: got another freeze
16:30:42 Join toffe82 [0] (
16:30:46pondlifeNico_P: Which device. btw?
16:30:58Nico_Psame settings as you, and I pressed select. on a beast
16:31:19pondlifeAh, so should be plenty of CPU...
16:31:25pondlifeSelect to reset pitch to 100%?
16:31:48Nico_Pno, select in the WPS, so it should go to the browser
16:32:03Nico_Pthe status bar appeared on top on the WPS then it froze
16:32:25pondlifeSo probably about to list a directory...
16:32:31Nico_P"data abort at 0004E190"
16:32:49pondlifeNice! Can you check a map against that?
16:33:21Nico_Pthe only problem is I don't have the same source anymore
16:34:16pondlifeMaybe try a current SVN with just #8894 ?
16:34:17 Quit nibbler (Read error: 113 (No route to host))
16:34:40Nico_Pyes, I'll try with the lastest version of the patch, too
16:34:59Nico_PI do have some map files in the build dir though
16:35:25 Join kugel [0] (n=kugel@rockbox/developer/kugel)
16:35:58 Join {phoenix} [0] (
16:36:31Nico_Paccording to those, 0x0004E190 is in strncpy
16:36:52***Saving seen data "./dancer.seen"
16:42:29 Quit jaykay (Read error: 110 (Connection timed out))
16:42:52Nico_Phmm, I'm getting a BUNCH of warning from setting_list.c
16:43:09Nico_Pplus a few errors
16:44:27pondlifewarning? patch failures?
16:44:38pondlifeI synced yesterday...
16:45:40Nico_Pclean svn does it too
16:45:44Nico_P(build errors)
16:45:57pondlifeI'll try a beastybuild
16:46:02Nico_Ptrying make reconf
16:46:21pondlifeYou'll need to do that (and maybe clean too)..
16:47:33 Join Zoxc [0] (
16:48:44evilnickpondlife: If you can send it to me, then I can test the beast build
16:49:01pondlifeHmm, I get errors too
16:49:27pondlifeLots of RTC/alarm problems.
16:50:24Nico_Ppondlife: I deleted my build dir and started over, now it works
16:50:34pondlifeI'm just svn reverting
16:51:18 Quit kugel (Nick collision from services.)
16:51:22 Join kugel [0] (n=kugel@rockbox/developer/kugel)
16:51:26 Quit kugel (Client Quit)
16:51:43 Join kugel [0] (n=kugel@rockbox/developer/kugel)
16:52:59 Join kachna [0] (
16:53:43Unhelpfulpictureflow main thread calls update_scroll_animation, which updates the center slide index (among other things), and posts a message to the loading thread... any idea why the loading thread can fall so far behind on device, but not sim? shouldn't the loading thread get to run, receive its event, and load the slides the next time the main thread yields?
16:53:55 Join MethoS- [0] (
16:54:00 Join nothingman__ [0] (
16:54:32kugelUnhelpful: because the sim is faster?
16:55:06Unhelpfulkugel: but the multitasking is cooperative - shouldn't the next yield cause the render thread to be blocked until the load thread has loaded a slide?
16:55:18Nico_Ppondlife: OK, clean svn build with only the latest version of the patch. I got a data abort at 0004E1F4
16:55:39kugelUnhelpful: is it really blocked?
16:56:17Unhelpfulit is unless something the load thread does yields implicitly... do core open or read do that?
16:56:31pondlifeI still can't get the Beast to build -
16:56:52pondlifeThat's after svn revert -R . and svn up, make reconf, make clean, make
16:57:12Nico_Ppondlife: make reconf wasn't enough for me. I deleted the build dir and made a new one
16:57:50Nico_Pthe data abort still appears to happen in strncpy
16:58:08 Quit Xerion (" ")
16:58:21Nico_Plow speed isn't enough to trigger it, high pitch is also necessary
16:58:33pondlifeHow about high pitch alone?
16:58:38Nico_Plet me try
16:59:02pondlifeMaybe a lower-level issue (high pitch + boosit??)
16:59:09pondlifeOr does the beast not boost?
16:59:17Nico_Pit doesn't
16:59:19Unhelpfulall the load thread does is some buffer management to figure out whether it can load a slide, and which slide to load, then it calls read_pfraw, which calls into core to open and read a file... but there are no explicit yields done anywhere, so unless there's an implicit one on read or open...
17:00:44 Join Xerion [0] (
17:00:55 Quit MethoS-- (Read error: 145 (Connection timed out))
17:01:35Nico_Ppondlife: got another data abort address which appears to be somewhere in dircache.o
17:01:52Nico_PI'll try with dircache disabled, for confirmation
17:02:12 Quit JdGordon (Read error: 110 (Connection timed out))
17:02:15pondlifeNico_P: pitch and timestretch again? Or just one?
17:02:33pondlifeMaybe we're overheating it? ;-)
17:02:43*kugel tries timestretch too
17:02:53pondlifePlease let me know if you smell smoke
17:02:58Nico_PI doubt the beast struggles even with both, there are no audio skips at all
17:03:38 Quit PaulJam (".")
17:04:05Nico_Pthe file browser works fine in the absence of dircache, even with pitch+timestretch
17:04:37Nico_Pmy guess is that pitch+timestretch cause a buffer overflow which corrupts the dircache somehow
17:05:05 Quit Zagor ("Client exiting")
17:05:14 Join ender` [0] (
17:06:06pondlifeNico_P: Can you try just pitch + dircache?
17:06:19pondlifeMaybe have timestretch enabled, but at 100%
17:06:21Nico_PI tried that, it didn't crash
17:06:25pondlifeAh, ok
17:06:49pondlifeWas that with the timestretch buffer allocated (i.e. booted up with the master enable on)
17:07:55Nico_Pit seems that the condition for the crash is clearly timestretch+pitch+dircache
17:08:10pondlifeThat's what I've been using
17:08:11Nico_Pthe values of pitch and timestretch don't even need to be extreme
17:08:16pondlifeI'd add + fast CPU
17:09:11Nico_PI think either one of pitch or timestretch overwrites the dircache buffer that is allocated with buffer_alloc
17:09:56pondlifeTimestretch also calls buffer_alloc... perhaps they are not init-ed in the right sequence?
17:10:00 Join _Auron_ [0] (
17:12:07Nico_Ppondlife: timestretch also appears to be allocated right before dircache is (assuming the order of appearance in settings_list.c is relevant), which would confirm what I said earlier
17:12:49 Join AndyI [0] (i=AndyI@
17:13:02*Nico_P needs to go to class now
17:13:26LambdaCalculus37amiconn: Here?
17:13:29Nico_Ppondlife: can I let you summarize the issue on the tracker?
17:13:37pondlifeNo worries, will do
17:13:50 Join mcuelenaere [0] (n=mcuelena@rockbox/developer/mcuelenaere)
17:13:53Nico_PI'll try to investigate further when I get back
17:14:30 Join midijunkie [0] (
17:14:34kugelpondlife: ha nice
17:14:48pondlifeThat's not the word, I'd use ;)
17:14:57kugelnot good enough for normal music imo, but the audio book people will be definitely happy
17:15:02pondlifeBetter with speech than music
17:15:31pondlifekugel: What device are you using?
17:15:58kugelno data abort so far (I have dircache+timestrech+pitch
17:16:25 Join MethoS-- [0] (
17:16:44pondlifeFor the logs - the timestretch buffer_alloc happens (via settings_apply()) before any init_dircache() call.
17:17:18pondlifeSo a write off the end of it's buffer would probably be the issue. But why pitch-dependent?
17:18:59kugelhaha, 200% pitch and 250% timestrech is indeed funny
17:19:24kugelstill no data abort
17:19:36 Quit midijunkie ("?(???~~)?")
17:19:56 Join midijunkie [0] (
17:21:53 Join CaptainKwel [0] (i=2669ecc2@gateway/web/ajax/
17:23:18 Quit blithe ("Lost terminal")
17:23:30 Join blithe [0] (
17:27:02 Join archivator [0] (n=archivat@
17:30:41 Quit MethoS- (Connection timed out)
17:31:00 Part B4gder
17:33:52 Join flydutch [0] (
17:34:33 Join n1s [0] (n=n1s@rockbox/developer/n1s)
17:36:42 Quit J-23 (Remote closed the connection)
17:37:05 Join J-23 [0] (
17:39:44 Join Mouser_X [0] (i=cf9bb00a@gateway/web/ajax/
17:39:51 Quit Darksair ("Everything that has a beginning has an end.")
17:40:15 Join jaykay [0] (
17:40:40Mouser_XQuestion: I'm attempting to use the VMWare Rockbox image in qemu, but it's not going well.
17:40:47LambdaCalculus37amiconn: I'll read the logs later. I wanted to ask you about FS #8778.
17:40:52 Quit LambdaCalculus37 ("Fwump")
17:41:02Mouser_XSpecifically, it appears that the image is attempting to do something that it's not supposed to.
17:41:40Mouser_XI keep getting an error message about some data being sent illegally or something (I'll have to check to see what it is exactly). I think it's related to networking though.
17:41:46kadobanMouser_X: i don't think that anyone has tried very hard to get the vmware image to work in anything but vmware
17:42:10Mouser_XIf you check the Wiki page about it, there's an entry for qemu there.
17:42:14 Quit nothingman__ (Connection timed out)
17:42:20Unhelpfulhrm, lines seem to be getting chopped in logfdump?
17:42:30Mouser_XAlso, I have a similar problem when running it in VMware. It drives my computer to a complete halt.
17:42:40rasherUnhelpful: logf limits line-length
17:42:42gevaertsUnhelpful: logf limits things to 32 (IIRC) characters per line
17:42:44Mouser_XI can't do anything except CTRL+ALT+DEL the thing.
17:42:48archivatorI just generated a bunch of .talk files and was extremely disappointed when I heard what espeak passes for "voice". Any chance of getting rbutil to work with festival? I have 2 multisyn voices that certainly sound better.
17:43:13Unhelpfuli could swear i managed to get longer ones out via serial logf, but maybe i'm wrong about that
17:43:29rasherarchivator: rbutil does work with festival last I heard
17:43:35rasherarchivator: or you could use the bulidstystem
17:43:59gevaertsarchivator: I don't really know, but have you tried the mbrola voices with espeak? Those apparently sound much better
17:44:06Mouser_XIf someone could tell me how to disable network everything from the VMware image (more specifically from Linux), I think it would probably take care of the problem.
17:44:14mcuelenaereUnhelpful: longer lines should get split over 2 or more lines AFAIK
17:44:38BigBambiMouser_X: This doesn't seem a very Rockbox specific problem
17:44:43Unhelpfulit certainly appears they're simply being truncated
17:44:52archivatorgevaerts: mbrola voices require another executable, right? I haven't bothered to set them up..
17:44:58kugelUnhelpful: there's a task
17:45:04mcuelenaereUnhelpful: what output are you using, USB serial logf?
17:45:04gevaertsarchivator: not sure...
17:45:06archivatorrasher: I only see flite, swift, and espeak in tts.cpp ..
17:45:30Mouser_XBigBambi: I'm using the VMWare image from Rockbox's site. I figured that since the image was built from some Rockbox people, that they'd be more applicable to helping here.
17:45:33Unhelpfulno, logfdump... i have serial on, and it was working before, but now i'm getting nothing. might be a host problem.
17:45:42mcuelenaereoh but there is MAX_LOGF_ONE_LINE_SIZE..
17:46:06mcuelenaerewhich is 200, try increasing that
17:46:16BigBambiMouser_X: It seems to me you have a general VMWare issue - and using it in qemu is ceetainly not a Rockbox issue
17:46:32*Mouser_X is attempting to get FS #7331 (GBS Codec) working again.
17:46:41Unhelpfuli don't think i was getting this before: usbserial_generic: probe of 2-2:1.1 failed with error -5
17:46:46kugelUnhelpful: logfdump always truncated afaik
17:46:51BigBambiThe Rockbox VMWare image is just a linux VM with the cross compilers installed
17:47:04Unhelpfuli'm getting truncated at *much* shorted than 200 :/
17:47:17rasherarchivator: Weird. I'd just use the build system -
17:48:00mcuelenaereUnhelpful: 29 perhaps?
17:48:10kugelUnhelpful: , committed but only applies to display, not dump it seems
17:48:27archivatorrasher: I've never used that, it only builds the menu clips, right? I actually wanted to build the file/dirs .talk clips..
17:48:43mcuelenaerekugel: it does change logfdump AFAICS
17:48:52kugelhm, well
17:49:11kugelit doesn't work though. I've been using logf[dump] recently too
17:49:18Unhelpfulit would certainly appear to be truncated at ~30
17:50:04*mcuelenaere admits he didn't test logfdump, only logf display when testing that patch..
17:51:02rasherarchivator: You can use the tools/ script to generate .talk clips as well. Though I don't think it's seen much use
17:51:09Unhelpfulhrm, no, it has to be something stranger than that... my logfdump has this: "load_new_slide(1215): 59 23 23 23" but also this: "load_new_slide(1215): 60 18"
17:51:14kugelmcuelenaere: fixes it maybe?
17:51:43*mcuelenaere hasn't seen that one yet
17:52:32mcuelenaereit does look like I committed those logf.c changes myself though
17:54:01 Part toffe82
17:54:07 Join nuonguy [0] (
17:55:17 Quit blithe ("Lost terminal")
17:55:28 Join blithe [0] (
17:56:54Unhelpfuli'd rather have serial work properly anyway. maybe i'l ltry it on the beast or reboot the PC :/
17:57:07 Join nibbler [0] (
17:59:47 Join Jason77 [0] (n=c35d1505@gateway/web/cgi-irc/
18:00:55 Join toffe82 [0] (
18:01:45Jason77does rockbox work on the Archos Gmini XS202?
18:02:29krazykitJason77, the list of supported players is on the front page
18:06:47archivatorrasher: when building in Voice mode (or whatever you call it), are there any requirements as to what festival's default settings should be? I get a perfectly normal build with no english.voice file. No errors, too.
18:07:09rasherarchivator: did you run make voice?
18:07:30archivatorrasher: Oh. that explains a lot. Sorry....
18:11:14 Join MethoS- [0] (
18:11:21 Quit MethoS- (Remote closed the connection)
18:13:45 Join MethoS- [0] (
18:13:51 Join przemhb [0] (
18:15:14przemhbHi all
18:16:04kugelUnhelpful: Are you trying this patch?
18:17:09 Quit gromit`` (Read error: 60 (Operation timed out))
18:17:28 Quit petur ("work->home")
18:18:14Unhelpfulkugel: can't right now, but i'll keep it handy. it'd be nice if i could duplicate on sim, really. :|
18:19:05przemhbis there a way to connect bypassing new RB USB mode? (i.e. for charging only and not for file transfer)
18:20:26evilnickprzemhb: Which player do you have?
18:20:40rasherarchivator: you may run "V=1 make voice" to see the exact commands that are executed
18:21:47 Join jgarvey [0] (
18:21:48archivatorrasher: it's not the commands that bother me, I actually think I hit a festival bug just now =/
18:22:13rasherAlright, I was just thinking that maybe your arguments weren't being passed as you expected them
18:22:49kugeljhMikeS: ping
18:23:21evilnickprzemhb: I'm not 100% about the H10 but with my players, there's a button that you hold while inserting USB to force Rockbox to NOT connect and just charge. Try the Menu button.
18:23:41 Quit shadearg ("goodbye.")
18:25:15 Quit MethoS-- (Read error: 110 (Connection timed out))
18:26:20przemhbevilnick: thanks
18:26:35 Join taylor_ [0] (
18:27:15evilnickprzemhb: Did it work?
18:27:47kadobandon't all of the ipods work the same with with charging? (hold <somebutton> while inserting USB to charge only) only some of them have it in the manual in section 3.3
18:28:00przemhbone more question about USB stack - is there a chance download speed will be faster? (currently upload speed is as fast as it is using OF but download speed is 4 times slower)
18:28:32gevaertsprzemhb: first of all, what do you mean by upload and download?
18:29:06przemhbgevaerts: upload = write to player
18:29:33gevaertsWhat sort of speeds do you get?
18:29:58przemhbgevaerts: for H10 upload is about 8MB/s using both RB and OF but download is 4MB/s for RB and 16-18MB/s for OF
18:30:51gevaertshm, which H10 is this? 5GB or 20GB?
18:31:02 Join MethoS-- [0] (
18:31:03przemhbgevaerts: upload speed is almost constant (no matter file location on a disk)
18:31:03gibbon_(H10 manual online version gives a 500 internal server error btw)
18:31:22przemhbgevaerts: H10 20GB (with 60GB disk)
18:31:26*gevaerts gets 7.3 MB/s read speed with the H10 5GB
18:31:30kadobangibbon_: try it again, it seems sporadic
18:31:35 Join shadearg [0] (
18:32:41gibbon_kadoban: ok, then its fine... i just wanted to look if the USB bypass button is documented (since i think i remember it was for the ipods)
18:32:43gevaertsprzemhb: the current situation is that we have a bottleneck at about 9MB/s for USB (I know where it is, it just needs work). Anything lower is the disk driver, probably related to rockbox not using DMA
18:33:36kadobangibbon_: yeah, it doesn't seem to be documented for every player that i thought could do it, but i don't know the players well enough to fix it...the old ipods also don't show it, and the nano
18:33:40przemhbgevaerts: oh I see
18:34:09przemhbgevaerts: so there is hope it will be faster :)
18:34:21 Join gartral [0] (
18:35:11gartralmy patch at fs #9961 seems too work just fine
18:35:35gevaertsprzemhb: it should be possible to make USB faster at least. Whether we get the disk faster depends on people figuring out how to do ATA DMA on PP5020. It might be similar to PP5022, but I don't know about that
18:35:37 Quit MethoS- (Read error: 60 (Operation timed out))
18:36:48gibbon_kadoban: ah, ok
18:36:55***Saving seen data "./dancer.seen"
18:36:55 Join miepchen^schlaf [0] (
18:37:22gibbon_kadoban: are you - lets say - the documentation maintainer? ... or one of them?
18:37:33kadobangibbon_: not at all, i'm just some random guy
18:37:47kadobani'd make a patch if i knew the fix ,that's the only reason i stated it like that. sorry for any confusion
18:38:00gibbon_no... nevermind :)
18:38:08 Quit midijunkie (Read error: 104 (Connection reset by peer))
18:38:50gibbon_i just wanted to offer my help on documentation / translation ... i am not a coder, but 5hrs/day on a train bored to hell... perhaps i could be of any use to the project
18:39:01 Quit Mouser_X (" ajax IRC Client")
18:41:06 Quit nibbler (Connection timed out)
18:41:36kadobangibbon_: that's great, i'm sure rockbox could use you. it seems that there's always documentation and translation work to do. if you have the skills, it's as easy as making patches and posting them to the FlySpray tracker, or there's documents that explain most of it in the wiki if you don't know how
18:42:08gibbon_my patches were of no use ... a morse trainer :P
18:42:37gibbon_but documentation - ok - will read the guidelines.
18:45:02archivatorrasher: are all those .mp3s the script builds supposed to be valid mp3 files (headers and all)? I can't open any of them. They're not 0-sized, though, that's why I'm asking.
18:45:39rasherarchivator: They should be valid mp3 files
18:46:00rasherarchivator: Which target are you building for?
18:46:11archivatorrasher: ipodvideo
18:46:32pixelmalooks like one of the download mirrors has a problem (re. the "500 - internal server error" for the online manuals). I get it too for some manuals, some other show but look as if they miss their CSS
18:47:39rasherarchivator: Weird
18:47:41pixelmapdfs are affected as well
18:48:37woyciesjesJust tried manuals for iPod video, okay here. (CT, USA)
18:49:15gibbon_pixelma: how do you distribute requests among the mirrors?
18:49:28rashergibbon_: round-robin dns
18:50:06archivatorrasher: Well, it's festival's fault, that much is clear. The question is, what the hell is going on in there.. Festival segfaults if I don't explicitly set it a voice. It should be reading the config file and getting a voice from there.. Oh, well, they're Scottish..
18:50:07 Quit Jason77 ("CGI:IRC (EOF)")
18:50:20rasherSeems to be the videolan mirror
18:50:23 Join Riku [0] (
18:50:43Rikuhi i was wonder why the 5g rockbox lags more than the 4g ones
18:50:56gartral500 internal error on e200 manual, ohio USA
18:51:02rasherRiku: Bigger screen
18:51:06pixelmadionoea: any idea about the videolan mirror?
18:51:30woyciesjesrasher: confirmed 500 internal error on that link
18:51:40rasherNo more confirmation needed!
18:51:49rasherYou're loading the same page, of course it's giving the same error.
18:52:22woyciesjesK. Just trying to help. :) Could be locality based tho, couldn't it?
18:52:33woyciesjesNo, wait, I'm thinking of something else. Sorry....
18:52:38kadobanfor a server error? not really, no
18:52:56woyciesjesYeah, I realized that now. Silly me...
18:53:27gartralrasher: i got through on the link you provided
18:53:32gevaertsrasher: this latest H10 bootloader seems to fix things
18:53:48rashergevaerts: Awesome
18:54:05rasherExcept that you still don't get rockbox usb when plugging in while powered off :\
18:54:55rashergartral: Doesn't really matter - the server's giving out 500 errors, that it works once in a while is not really interesting
18:58:40 Join avis [0] (n=ident@pdpc/supporter/student/avis)
18:59:39 Join midijunkie [0] (
19:00:59 Join MethoS- [0] (
19:01:12 Join gromit` [0] (
19:02:32 Quit avis (Client Quit)
19:04:38archivatorrasher: you sure those are valid mp3s? I got festival fixed and VLC still says that there isn't a final MPEG frame... Perhaps the build script strips those before assembling the final file?
19:05:39 Quit FOAD ("I'll be back")
19:05:55rasherWhich mp3 files are these? Doesn't it delete temporary .mp3 files?
19:06:33mcuelenaereanyone opposed to placing the speaker_enable() prototype in audiohw.h? It seems a bit silly to make a separate speaker.h file for just one function..
19:06:43 Quit midijunkie (Read error: 104 (Connection reset by peer))
19:07:15archivatorrasher: It does, I try to open them before it completes the process (it doesn't hold a lock on them - a bug?).
19:08:09rasherI don't see why it should hold a lock on them. If you go around deleting them while it's building, you're getting what's coming for you.
19:08:56rasherWell, try commenting out line 507 of to make sure what's going on
19:10:03archivatorrasher: You're right. The important thing is, I got a decent voice files from a multisyn corpus. However, the intonation at the end of the clips is messed up as festival expects a fullstop at the end of complete phrases. This is definitely a bug :)
19:10:35rasherThat's interesting
19:10:39archivatorrasher: no need to, I just used the mp3s as an indicator and thought the process had failed when it had actually succeeded. has a method for altering the strings before sending to TTS. Adding a full stop to all strings would be easy
19:11:14 Part przemhb ("User is away.")
19:12:03rasherarchivator: try adding $string ~= s/$/./; at line 135 and see if that sounds better
19:12:20rasherarchivator: Do you know if this is true for all festival languages/voices?
19:12:22 Join FOAD [0] (
19:13:00 Join midijunkie [0] (
19:14:01archivatorrasher: no, most of them don't have such detailed corpora to build from. Even if they do, they don't go into that much detail when building the voices. The multisyn voices are > 100 MB each any way, so they can afford 20MB on top of that for the closing intonation.
19:14:52rasherAre these commercial voices?
19:15:01archivatorNo. Just very large :)
19:15:40rasherAre they free to distribute? Specifically, could we use them for the voices on
19:15:51rasherThe current ones are rather pitiful
19:16:37archivatorrasher: I'll check in a minute, going to try your line now.
19:16:46 Quit MethoS-- (Read error: 113 (No route to host))
19:16:52rasherYou need to use =~ rather than ~=
19:16:55*rasher blames perl
19:17:32archivatorI think it's about time we ditch perl for python *ducks*
19:18:02*rasher certainly wouldn't mind
19:19:16rasherThough the does quite a few things where python probably won't be a huge advantage. Lots of text processing and regular expressions, and a great deal of forking and IPC
19:19:34archivatorApparently, I deleted the LICENSE files, will have to redownload the voices to check.
19:20:41archivatorrasher: python can do ipc! and forking! and that other thing you mentioned! On a more serious note, I'm just complaining because I never did learn perl..
19:21:54 Join maddler [0] (
19:22:42 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
19:23:03rasherThey're just a bit bothersome in python. Don't get me wrong, I prefer python over perl, I just don't think it's worth it to port the voice script (again). At least not yet
19:25:29archivatorSomething is seriously messed up here.
19:25:40archivatorRockbox refuses to use the .voice file
19:26:06archivatori.e. I seem to hear *something* in my headphones but it's more like a very faint white noise
19:26:29 Part gartral
19:26:40rasherI suspect it's related to the broken mp3 files in some way
19:26:52 Join gartral [0] (
19:27:02archivatorI say _seem_ cause I don't really hear it, my multimeter tells me something is sent to the headphones..
19:27:24rasherThe mp3 files are not actually mp3 files.
19:27:36rasherThey're containerless speex files
19:27:45rasherThe mp3 filename is simply a legacy thing
19:28:04rasherThat would explain *that* part
19:28:30archivatorYeah, file can't pinpoint its type.
19:28:45archivator*facepalms* should've thought of that earlier..
19:28:47rasherWe still use the same tool for generating the final .voice file, which looks for mp3 files
19:28:48taylor_Hello. Does the RockBox team have an ipod emulator that would emulate or run firmwares of the older ipods?
19:28:57rasherarchivator: It's not exactly obvious..
19:29:25rashertaylor_: no. The closest is this:[0]=&sev[0]=&pri[0]=&due[0]=&reported[0]=&cat[0]=&status[0]=open&percent[0]=&opened=&dev=&closed=&duedatefrom=&duedateto=&changedfrom=&changedto=&openedfrom=&openedto=&closedfrom=&closedto=
19:29:30archivatorrasher: file a bug? Cause it _is_ a bug..
19:29:41rasherarchivator: Where?
19:30:06jaykayif a lang file contains something like *: "somerandomcontent" , is it compiled for all targets?
19:30:10archivatorIn the merger thing. The fact that it looks for *.mp3.. I call that a bug.
19:30:32taylor_Hmmm.. Ok thanks
19:30:45rasherjaykay: That is the meaning of *, yes.
19:30:52rasherjaykay: unless there's a better match, iirc
19:31:18rasherarchivator: I don't see that as a huge issue. It's nothing regular users will see
19:31:47archivatorI'm a regular user and I saw it! Never mind, dl-ed the voices.
19:31:51archivatorNo LICENSE files.
19:32:05archivatorWhich would mean they fall under the festvox licenses.
19:32:11jaykayin mcuelenaeres latest commit he added *: "Enable Speaker" to the lang files, so it would be compiled for all targets, although its only for this onda
19:32:12archivatorIf only I knew what those are..
19:32:26mcuelenaerejaykay: yes, I just noticed that too
19:32:44mcuelenaerebut doesn't this depends on features.txt ?
19:33:04rasherOr you could add the specific target
19:33:05gevaertsrasher: the H10 (at least the 5GB) doesn't actually boot when you plug in USB, so maybe it's not that bad
19:33:17mcuelenaererasher: it is meant to be cross-target compatible
19:33:31*mcuelenaere should probaby add it to features
19:33:37rashermcuelenaere: I take it this is an external speaker?
19:33:49mcuelenaerewell actually no
19:33:53mcuelenaereit is internal to the device
19:34:03rasherYeah, that's what I *meant*
19:34:13rasherSomething like "speaker" in features seems reasonable
19:34:21mcuelenaereyeah, that's what I though
19:35:42archivatorrasher: This is the closest I got to a license covering those voices:
19:36:42archivatorrasher: the voices themselves are on that same server, so I think they fall under the "no claim" heading.
19:36:58rasherX11 license is certainly free enough
19:37:52mcuelenaererasher: something like should do, right?
19:37:53 Join Llorean [0] (
19:37:59archivatorThat's the software. I'm not completely convinced the actual voices have the same license...
19:38:22rasherYeah, doesn't seem obvious
19:39:12 Part pondlife
19:39:20archivatorShould I write them an email?
19:40:07rasherIt would be nice if we could create some better voice files
19:40:43rasherI take it you're talking about the files festvox_cstr_us_jmk_arctic_multisyn-1.0.tar.gz and festvox_cstr_us_awb_arctic_multisyn-1.0.tar.gz?
19:40:45archivatorWould like me to build a .voice for another target for you to see the difference?
19:40:53archivatorThose are the ones.
19:41:23archivatorThere are two or three more but for commercial use.. (i saw their names in the config files)
19:41:38kugelrasher: can I mass convert the scrollwheel feature into another, like wheel_acceleration?
19:41:42 Join nibbler [0] (
19:42:20kugelwith the lang-tools I mean
19:42:27rasherkugel: I don't see why sed wouldn't do
19:42:50rasherBut maybe, I don't remember if I added something like that
19:43:28taylor_does anyone here have an ipod photo?
19:44:08kadobantaylor_: i would think that you can assume that the answer to that is yes and just ask your real question
19:44:38taylor_ok then. can someone try something for me if they do have a photo?
19:44:41rasherkugel: You could try −−help
19:45:03rashertaylor_: Again, there's no reason to wait for an answer to that question
19:45:18rasherThe only possible answer is yes, or no answer.
19:45:52taylor_ok. If someone does have a photo. Does that photo have a notes section?
19:46:03taylor_ok. you know, the place to view tiny txt docs
19:46:57 Join Xithal [0] (
19:47:27rashertaylor_: How is this related to Rockbox?
19:47:59taylor_Its related to linux4nano. And the possible *exploit* we are working on.
19:48:34rasherThen you'd appear to be in the wrong channel
19:49:01 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
19:49:36Lloreanrasher: Well, developing a way to get bast the signature check would be valid for both
19:50:00LloreanThough I'd imagine the DFU mode is a bit more interesting.
19:50:13 Part Xithal
19:50:16taylor_Well, I think thats basically what I asked when I came over here and you said to just ask the question ;)
19:50:49taylor_We aren't sure the DFU mode would lead to anything *yet* we would still have to decrypt the DFU files
19:50:54rasherAnyway, I thought most Ipods have had the Notes thing? I certainly remember a lot of new ipod users asking why the Rockbox function didn't work (while it was actually just showing the *folder* named "NOTES")
19:51:47*rasher finds circumstantial evidence that the multisyn voices are Free
19:52:06taylor_the reason being: I am going to edit the Sansa source just a little bit to run ipod firmware(photo) and test this ug.
19:52:11taylor_the reason being: I am going to edit the Sansa source just a little bit to run ipod firmware(photo) and test this *bug
19:52:33archivatorrasher: they should be, the corpus _is_ free and the tools are free as well. Anyone with enough knowledge could rebuild them..
19:52:41 Join tvelocity [0] (
19:53:13kugelrasher: I'm not familiar with sed, but fortunately text editors have a replace-in-all-files feature
19:53:29 Join Thundercloud [0] (n=thunderc@
19:53:45rasherkugel: sed -i s/scrollwheel/wheel_acceleration/ apps/lang/*.lang
19:54:07kugelhm, that sounds too easy :p
19:55:26 Join bertrik [0] (
19:56:00pixelmain case "scrollwheel" is used as an opt in the manual, it should be replaced too
19:57:38pixelmaand IIRC it is used at least once in the pictureflow.tex which made me investigate the history of HAVE_SCROLLWHEEL
20:00:52 Quit t0mas (Read error: 104 (Connection reset by peer))
20:01:05 Join archivator_ [0] (n=archivat@
20:01:20 Quit archivator_ (Client Quit)
20:06:14*kugel sees this giant \opt and wonders whether this includes all targets (except h10)
20:06:29 Join jaykay_ [0] (
20:13:57pixelmain pictureflow.tex or something else that already excludes some targets?
20:14:45kugelin pictureflow.tex
20:15:09*kugel summons BigBambi
20:15:59kugeluhh, the manual could need a good cleanup w.r.t scrollwheel
20:16:24BigBambiscrollwheel appears to be a silly define
20:16:27kugelBigBambi: I was wondering about the big \opt in pictureflow.tex
20:16:44kugelnot anymore in a few minutes
20:17:28kugelbut still, this one \opt seems to include every target separately
20:17:35BigBambikugel: hmmm, that appears to be everything except H10 :)
20:18:02BigBambiI suspect I started with some different then changed them, then forgot to replace it with nopt h10
20:18:27kugelok, just wanted to make sure
20:18:29gevaertsSo what's different about H10?
20:18:30rasherThere are no archoses there
20:18:57pixelmalooks like it could probably be replaced by a \nopt excluding the H10 pad - would improve readability for now but will give a different result if targets will be added that have no \ButtonSelect
20:19:10*BigBambi needs to look at the whole file
20:19:11pixelmarasher: that's pictureflow.tex
20:19:37rasherpixelma: I don't see any reasons why they couldn't have pictureflow in the future
20:19:45BigBambirasher: no pictureflow for archos
20:20:07BigBambiwell, I don't think it is too important
20:20:25BigBambiIt could use nopt h10, but then someone would maybe have to change it back in the future
20:20:42pixelmayes, you'll need to work on the tex then anyways (as you don't know the button mapping yet) - and yes, they don't have a select button
20:20:53BigBambigevaerts: The H10 doesn't use the "select" button for that option
20:21:03*kugel wonders why
20:21:17BigBambidoes it have one?
20:21:24pixelmathe H10's "select" button isn't called "select" in the code...
20:21:27*BigBambi can't remember
20:21:28kugelprobably some PLA weirdness
20:21:34BigBambiI have to look at the layout everytime :)
20:21:38kugelpixelma: or that
20:21:39 Join stripwax [0] (
20:21:45BigBambiI knew I wasn't going mad!
20:21:53pixelmathey don't really have a "select" button
20:22:19*kugel commits his scrollwheel / wheel_acceleration thing and crosses his fingers
20:23:18 Quit jaykay (Read error: 110 (Connection timed out))
20:23:58BigBambikugel: Do we now have a proper manual define for scrollwheel?
20:24:03 Join anotherfoot [0] (
20:24:35*kugel seems the acceleration isn't covered in the manual
20:24:47kugelBigBambi: yes, scrollwheel, after my commit
20:24:49 Join fml [0] (n=4fd3d3b6@gateway/web/cgi-irc/
20:25:16BigBambikugel: Excellent
20:25:44BigBambikugel: I was bitten by that before, assuming that scrollwheel meant it had a scrollwheel :)
20:26:15kugelhaha, those mass-file commits are thrilling
20:26:27 Join LambdaCalculus37 [0] (n=rmenes@rockbox/staff/LambdaCalculus37)
20:26:28kugelBigBambi: me too, so I decided to change it ;)
20:26:57BigBambikugel: thanks :)
20:26:58kugelit actually said: it may have a scrollwheel, but if, it has some wheel acceleration code
20:27:08kugelnow it says: it has a scrollwheel!
20:27:26BigBambikugel: Yeah, it was mainly me being lazy and not looking into it deeply enough
20:27:58Nico_Pkugel: about FS #8894, which target are you testing on?
20:29:03kugelI would test on my fuze too, but there's not much sense in that
20:29:05Nico_PI didn't get data aborts on the ipod either, do you have a gigageat?
20:29:22kugelno, /me is still sansa-only :(
20:29:23 Nick jaykay_ is now known as jaykay (
20:29:25fmlHello. Am I the only one who gets compile errors in settings_list.c?
20:29:37fml...sim build, e200
20:29:46mcuelenaerefml: since when?
20:29:57 Quit Riku (Read error: 110 (Connection timed out))
20:30:25fmlmcuelenaere: svn update one minute ago
20:30:33mcuelenaerewhat revision?
20:31:14fmlmcuelenaere: r20177. I get: /settings_list.c:796: error: ‘LANG_SORT_INTERPRET_NUMBERS’ undeclared here (not in a function)
20:31:20kugelfml: I probably broke something, let's wait for the build servers
20:31:32kugelno, that seems unrelated
20:31:46pixelmafml: did you try a make clean + reconfigure?
20:32:17kadobani just tried r20177 e200 sim and it built fine on linux
20:33:00fmlI tried make very clean, reconf to no avail
20:33:18kugelBigBambi: svn up :)
20:33:40 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
20:33:40BigBambiyes sir!
20:34:13 Join Strife1989 [0] (n=michael@
20:34:27kugelone could do "s/\opt{IPOD_4G_PAD,IPOD_3G_PAD,SANSA_E200_pad}/\opd{scrollwheel}/" now
20:34:56kugel\o/ no red
20:35:03 Nick Strife1989 is now known as Strife89 (n=michael@
20:35:24LambdaCalculus37kugel: Nice work keeping it green! :)
20:35:37shotofaddsfml: I had the same errors this morning, nothing seemed to fix it except deleting the build folder and starting again (make clean, make reconf etc did not help)
20:35:53BigBambime too
20:35:58*kugel too
20:36:04 Quit proudfoot (Read error: 110 (Connection timed out))
20:36:19BigBambikugel: Get off the bandwagon!
20:36:58***Saving seen data "./dancer.seen"
20:38:29 Join z35 [0] (
20:39:51 Nick anotherfoot is now known as proudfoot (
20:40:56fmlshotofadds: the changes in lang are not picked up!
20:41:43fmlThere are new LANG_xxx symbols, but genlang hasn't been called by the make script
20:43:04fmlpixelma: yes, I did make clean and make veryclean. Now I deleted everything (like shotofadds) and it seems to work
20:47:40fmlkugel: two notes about natsort: (1) The menu entry in the settings should be Title Case (to be like other menu entries)
20:47:58shotofaddsseems like the makefiles need to capture that dependency, then. I'm not entirely sure where to start though.
20:48:17BigBambifml: Not a good time to suggest that
20:48:29fmlkugel: and (2) I have some folders with cyrillic glyphs. They are placed in the beginning of the list when the option is activated
20:48:33BigBambifml: Not all menu entries are, and there is a big debate about that
20:48:34kugelfml: sure, if we didn't already agree on sentence case
20:48:47BigBambifml: Given UK English (which Rockbox is) doesn't tend to use it
20:49:44fmlBigBambi: I think if it's done then everywhere. But for now it's very ugly because not consistent
20:50:02fmlWhat about the (2)?
20:50:07BigBambifml: Which doesn't mean that new onces should be title case
20:50:27n1sshotofadds: it seems to be related to the recent move of the generated lang files
20:50:32kugelfml: about the cyrillic, I suggest (a) you fix it yourself or (b) contact the author of the algorithm
20:50:57n1sdeleting lang.[ch] in the build dir fixed my builds
20:51:00bertriknice to see that some more information about the ams sansa USB was discovered
20:51:19kugelI have no way of testing non-iso-8859-1 codepages
20:52:16kugelbut all that was done, was to replace the call of strcmp with strnatcmp (and IIUC strcmp isn't utf8-aware either)
20:55:01amiconnkugel: The good thing about the plain str* functions is that they don't need to be utf-8 aware in order to sort utf-8 properly, due to the way utf-8 works
20:55:02gevaertsbertrik: now we just need someone to work on it :)
20:55:21pixelmakugel: it's not impossible to rename files to something that includes cyrillic for testinh
20:55:27pixelmaor testing
20:56:19kugelpixelma: if I don't know what I write, show should I know how that's supposed to be sorted?
20:56:39 Join denes [0] (
20:56:44 Join MethoS-- [0] (
20:57:28amiconn(sort properly in a basic sense - proper unicode collation is an entire different matter)
20:58:25LambdaCalculus37amiconn: Regarding FS #8778, kugel had mentioned that you had some sort of concern about it?
20:59:49 Quit Thundercloud (Remote closed the connection)
21:00:08 Quit Nico_P (Remote closed the connection)
21:00:56 Join itcheg [0] (i=41d59de2@gateway/web/ajax/
21:01:21bertrikgevaerts, you have a clip, right? You could start right away! :P
21:01:56pixelmakugel: you would only need one cyrrilic letter - it was about sorting cyrillic letters before numbers - not how the cyrillic characters are sorted in themselves
21:02:13LambdaCalculus37Speaking of Clips, funman is still looking for a new one so he can get back into port work.
21:02:21gevaertsbertrik: I could, but there are these annoying ARC performance problems...
21:03:05orsonjbertrik: last week you asked if I could do some testing on a sansa c240v2. I think I got the dev setup going now.
21:06:37bertrikorsonj, you can help with getting radio to work on the c200v2. It's not really something critical, but nice to have anyway
21:06:48LambdaCalculus37pixelma: Thanks.
21:07:04kugelbertrik: that would require a working display and buttons first I suppose
21:07:50bertrikkugel, yes that would make it a lot easier
21:07:59fmlamiconn: yes, I think it's because of unicode. And maybe signed chars
21:08:08 Join tessarakt [0] (
21:09:10bertrikdual boot support for the c200v2 on the sansaams wiki page says "no/yes", I don't understand
21:10:33kugelbertrik: well, it boots the OF :)
21:11:37 Quit MethoS- (Read error: 110 (Connection timed out))
21:13:09 Join DuneStar [0] (
21:13:18 Join Thundercloud [0] (n=thunderc@
21:13:46orsonjso, I can't boot rockbox to test the radio? Or is the wiki out of date?
21:15:47DuneStarHello. Is it possible to open an 8GB MicroSDHC card in Windows Explorer from an r19569 rockbox'd Sansa e260? Or will I need an external card reader?
21:16:24BigBambiDuneStar: I can't remember what build USB was enabled in, but one after that will let you use the card directly through Rockbox UB
21:16:36*BigBambi goes to check
21:16:56DuneStarthank ya Bambi
21:17:03gevaertsDuneStar: you basically have three choices. (a) use a card reader, (b) use a current development build, or (c) wait for 3.2 (should be in about three weeks I guess)
21:17:06 Join amoe [0] (
21:17:22gevaertsBigBambi: r19569 is 3.1
21:17:30bertrikorsonj, I didn't realise how poorly developed the c200v2 port was. First we need display and buttons to work.
21:17:37BigBambiDuneStar: After r20105
21:17:37BigBambigevaerts: aha
21:17:45DuneStaralrighty. I'll take my chances with the dev. build. Thank you.
21:17:51DuneStarafter r20105, gotcha
21:18:05BigBambiDuneStar: Just get the current
21:18:40BigBambiThat is changed on every source code change, so is always the newest at the point you download it
21:20:01orsonjwell, I'm still happy to help, I have the c240v2, I am able to download svn and compile, but to do more I'll need some directions
21:21:50DuneStarBambi - worked perfectly. Thank you!
21:22:32DuneStarThis build makes it so easy to connect my Sansa to XP - no reloading to the default Sansa firmware :)
21:23:18 Quit blithe ("Lost terminal")
21:23:28kugelorsonj: It needs a skilled hacker to get the lcd and button driver working
21:23:29 Join blithe [0] (
21:23:30 Join MethoS- [0] (
21:23:34BigBambino problem. It should be in 3.2 in about three weeks too
21:24:13kugelhm, speaking of 3.2
21:24:24kugelwe should start to
21:24:28kugelprepare the release
21:24:40gevaertsindeed. Especially the freeze
21:24:51kugelI've not seen any speek of a freeze, nor a Release32 page
21:26:25rasherDon't we freeze for 2 weeks?
21:26:55kugelbut 2 weeks before the release (the second week is after branching)
21:27:02gevaertsThe freeze for 3.1 started about two weeks before the release
21:27:10gevaertshm, plenty of time :)
21:27:49rasherkugel: Release32?
21:27:57fmlI have a patch for strnatcmp: The key is to use int (and not char) in isxxx funcs. Could anyone try it? It works for me.
21:27:59kugelWe need to wake up Bagder once more, I guess
21:28:02DuneStarkeep up the good work fellas
21:28:05amoeGood evening all
21:28:27amoeDoes anyone know, do vorbisgain and mp3gain use different reference levels?
21:28:32 Quit DuneStar ()
21:28:45BigBambiamiconn: Not really a rockbox question that
21:28:56 Join Aurix_Lexico [0] (
21:29:03BigBambiamoe: Not really a Rockbox question that
21:29:14amoeBigBambi: I know... can't think of anywhere else to ask
21:29:26BigBambiStill not here
21:29:34amoeBigBambi: fair
21:29:36kugelrasher: I mean ReleaseNotes32 (
21:30:09BigBambiamoe: There is always #rockbox-community for off-topic
21:32:36 Quit MethoS-- (Read error: 113 (No route to host))
21:33:19amoeBigBambi: thanks
21:38:18 Quit blithe ("Lost terminal")
21:38:29 Join blithe [0] (
21:40:02 Join Rabbit^^ [0] (n=Rabbit^^
21:40:45Rabbit^^Can anyone tell me where to find the Powerdesk article on the Rockbox forums?
21:42:21LambdaCalculus37And what does Powerdesk have to do with Rockbox?
21:42:31 Quit stripwax ("")
21:43:02mcuelenaerefml: shouldn't that be short instead of int?
21:43:09Rabbit^^The writer used it to help get Rockbox on the Sansa View
21:43:24BagderRabbit^^: no
21:43:31Bagderthere is no rockbox for the sansa view
21:44:28*kugel pokes Bagder about 3.2
21:44:31LambdaCalculus37There's not even any work on a port for the Sansa View.
21:44:53Bagdersomeone proved our dummy-signing works on it, that's all
21:45:08LambdaCalculus37Exactly. No actual work has been done, though.
21:45:46*Bagder pokes kugel back, mentioning speaking in whole sentences if he wants to be understood
21:46:03gevaertsBagder: that was a whole sentence!
21:46:16kugelI think so too
21:46:28gartralalright, im stupid.. i don't know how to install the test bootloader from sansa patcher :\
21:46:35BagderI'm not forcing anyone
21:46:36kugelBagder: I just thought we might start preparing 3.2 soon, going into freeze etc
21:47:01kugelThere's no timeplan about 3.2 yet, is there?
21:47:08BagderI think the two-week approach we did before worked fine
21:47:27 Quit MethoS- (Read error: 110 (Connection timed out))
21:47:30Bagderie freeze 14 days before, unfreeze and branch 7 days before, release on day 0
21:47:39kugelday 0 == ?
21:47:49BagderMarch 23rd
21:47:53*kugel suggest 23rd
21:47:56BigBambigartral: There is a command line option - does sansapatcher not give any help output (I think it does)?
21:48:31kugelWe also need to manage the new bootloader releases I think
21:48:36Rabbit^^it is what the poster used to help roll back his firmware version
21:48:58fmlmcuelenaere: all toupper, isxxx etc funcs accept int IIRC. Int is the best type in C
21:49:16mcuelenaereyes, but isn't this supposed to parse UTF8 strings?
21:49:49 Quit itcheg (" ajax IRC Client")
21:49:54LambdaCalculus37Rabbit^^: What you saw was someone testing an e200 build on a View just to see what it would do. That does not mean there is a Rockbox port for the View.
21:50:08 Join bzed [0] (
21:50:30fmlmcuelenaere: I think this will also work correctly with UTF8 They can be sorted as if they were normal strings. The problems in the strnatcmp were because of signedness
21:50:35Rabbit^^Do you remember the poster's ID?
21:50:44 Quit taylor_ ("Leaving")
21:51:23kugelSo, let's freeze in 1 week
21:51:59kugelmarch 9th
21:52:11Bagderyeah, sounds fine to me
21:52:34mcuelenaerefml: ok, I misunderstood the patch. I wonder though why they initially didn't go for this more performanter solution..
21:52:51fmlmcuelenaere: if you use char you'll get negative values because in utf8 encoded strings there are bytes with the highest bit set.
21:53:02 Quit amoe ("erk")
21:53:15*mcuelenaere mixed up UTF8 with UTF16, which uses short's
21:53:17 Join pyro_mania1 [0] (
21:53:21fmlmcuelenaere: I don't know. May be nobody tested this with utf8
21:53:29kugelIsn't there a comment at the top of the .c file, regarding to signedness?
21:53:59fmlkugel: I saw it. But somehow it leaks through
21:54:18 Join barrywardell [0] (n=barrywar@
21:54:27kugelfml: for speed, if you look at strncmp.c, the speed version compares 4bytes at a time too
21:54:32 Part Rabbit^^
21:54:54 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
21:55:56fmlkugel: yes, that's made to compare 4 bytes at a time. But here the problem is not this but the conversion from char (strnat signatures) to isdigit (std c signatures). Using int makes it correct, faster, and smaller bin size
21:56:25kugelcommit then? :)
21:58:03Bagderwhy is doing frequent releases bad?
21:59:00Bagderit allows careful users to keep up easier, it allows us to be slightly less careful/scared about committing
21:59:35Bagderit also makes us remain capable of doing releases
21:59:48 Quit __lifeless (Remote closed the connection)
21:59:51 Join _lifeless [0] (n=lifeless@
22:00:49fmlkugel: I'd rather wait that at least someone tests it. Could you?
22:01:28gevaertsI've been wondering. Has the freeze actually been useful for 3.0 and 3.1? Would just disallowing major overhauls and backporting only fixes not work just as well?
22:01:40gevaertsIt's only been two releases of course...
22:01:58 Quit midijunkie ("?(???~~)?")
22:02:17BagderI don't see a lot of pending big changes coming anyway
22:02:21 Quit moos ("Rockbox rules the DAP world")
22:02:54Bagderand it's hard to tell how it would've been without the freeze
22:02:55*mcuelenaere wonders whether should be considered for our strnatcmp implementation too
22:04:08gevaertsI vote for not changing the release process and cycle until we've had three or four. We can evaluate things properly at devcon and then adjust things if needed
22:04:23BagderI think that's a good idea
22:04:36Bagderit will produce a better evaluation
22:04:52kadobaneven without the forced release cycle, wouldn't now be a good point for a release anyway? (usb seems release worthy)
22:05:08BagderI think so, yes
22:05:26bertrikdo we have a name for 3.2?
22:06:40jaykayhad 3.0 and 3.1 a name?
22:07:05*gartral didn't know we were naming the releases
22:07:09bertrikno I don't think so
22:07:18gevaertsYes :)
22:07:31gevaertsNot official, and maybe not even unique, but there are names!
22:07:32Bagdernah, we once went with some names, but then it took us aaaages to reach there and then we didn't use the names any longer
22:08:01kugelfml: I could. Unfortunately I already deleted my test_sort folder, but I can surely create a new one
22:08:16kugelbut as you noticed, I only tested ascii
22:09:14fmlkugel: try Ä and Ü. The result should be the same
22:10:54fmlmcuelenaere: I don't quite understand the other version, but I think our variant is not worse!
22:11:27mcuelenaerefml: this is the bug report related to that 'fix':
22:11:55*mcuelenaere doesn't really understand the point of casting twice
22:12:04fmlmcuelenaere: in the end, they convert everything to int. Why use two steps for that?
22:12:15mcuelenaereright, that's why I'm wondering..
22:12:36rasherIf you have more than one Festival voice installed, please test - it *should* allow you to pick which one to use when building voicefiles
22:12:48Bagderbecause casting signed char to signed int sign-extend
22:13:01Bagderwhile casting to unsigned char first takes away that
22:13:03kugelfml: ä/ü sorts OK here
22:13:13kugela -> ä -> ü
22:13:24 Quit Strife89 (Read error: 110 (Connection timed out))
22:13:40gevaertskugel: how do ae and ue sort?
22:13:52mcuelenaereBagder: can't you cast to directly to unsigned int then?
22:14:08mcuelenaeres/to directly/directly/
22:14:13rashergevaerts: We don't do any fancy language-dependent collation, so I wouldn't expect anything there
22:14:14kugelok, but ä sorts after b, not so good
22:14:28Bagdermcuelenaere: you can, but that won't give the same results
22:14:34rasherkugel: Why not?
22:14:37rasherThere's no *right* way to sort ä
22:14:49kugelhm, probably
22:14:55Bagderthere is a unique one in every language
22:14:58gevaertsrasher: I still maintain that it's actually not just hard to do right, it's impossible
22:14:58rasherI'd expect it to sort between æ and ø
22:15:12fmlBagder: that's true. But in all char processing code, I just write "int c = *str" (with char *str) and it always worked.
22:15:22rashergevaerts: Oh sure, but we're not even trying. Which I think is perfectly acceptable.
22:15:33kugelasciisort sorts b before ä too, so it's "correct"
22:15:34gevaertsI agree there :)
22:15:47rasherAs long as we don't lose track of that
22:15:53Bagderfml: yes sure that works, but then you didn't need that extra weirdo conversion the double-cast does
22:16:20BagderI'm just saying there can be a point
22:16:40 Quit XavierGr (Nick collision from services.)
22:16:51 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
22:16:51kugelfml: according to you, ä and ü should be before everything else?
22:17:16fmlBagder: I don't like that double cast either. And I understand that there might be a point. But it just works! :-)
22:17:18 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
22:17:28 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
22:17:44 Quit barrywardell ()
22:18:11kugelfml: where's your patch?
22:19:12jaykayis "Battery: Trickle Chg" still used anywhere?
22:19:21gartralfml: did you see fs9961?
22:19:56rasherjaykay: The player, I assume
22:20:05rasherArchos Player, that is
22:20:13kugelgartral: I think he did, he committed it
22:20:30mcuelenaerekugel: his last version seemed to be
22:20:35Bagderwe once had some scripts that checked if all phrases are used in code
22:20:35pixelmajaykay: should be on some Archos (Recorders or Player, don't remember exactly)
22:21:00jaykaytheres also recorder: "Battery: Top-Off Chg"
22:21:22jaykaybut the lang file says only charging: for trickle chg
22:21:59 Quit LambdaCalculus37 ("Fwump")
22:22:36fmlkugel: no, I think umlauts should be after the ASCII
22:22:51 Part gartral
22:23:29 Nick fxb__ is now known as fxb (
22:23:44 Join barrywardell [0] (n=barrywar@
22:24:27kugelfml: they are, with SVN too
22:24:47kugelanyway, your patch changed nothing for me, so I guess it can go in
22:25:19fmlkugel: then try cyrillic. Ä and Ü are not weird enough! :-) The patch is FS #9975
22:25:27 Join petur [50] (n=petur@rockbox/developer/petur)
22:26:15rasherBagder: I'll probably commit a change to configure that'll change voicebuilding soon
22:26:20kugelfml: You test cyrillic enough. just commit it
22:26:29rasherBagder: FS #9976
22:26:31fmlBut it would be good if more people confirmed that
22:26:39BagderI noticed, I'll adapt
22:26:52bluebrotherBagder: speaking of configure: does building the manual still configure as (M)anual?
22:27:08fmlkugel: I only tested in the sim
22:27:19bluebrothersince you need to call the manual target explicitly anyway I don't see much point in having this separately
22:27:20kugelfml: I tested on target. it breaks nothing for me
22:27:30Bagderbluebrother: it does
22:27:45Bagderand no, there's really not much point with it
22:27:46rasherBagder: This should also allow you to use a different festival voice with less hassle (the two "multisyn" ones sound good -
22:27:48 Join low_light [0] (i=c730180b@gateway/web/ajax/
22:31:23 Quit _lifeless (Remote closed the connection)
22:31:37kugelfml: I don't see anything wrong with your patch
22:31:40 Join _lifeless [0] (n=lifeless@
22:33:06 Join Strife1989 [0] (n=michael@
22:33:38 Quit jaykay ("ChatZilla 0.9.84 [Firefox 3.0.6/2009011913]")
22:34:16 Join HBK [0] (
22:35:44fmlkugel: done
22:36:59***Saving seen data "./dancer.seen"
22:37:32rasherBagder: Committed. It'd be great if you could have a go at using one of those higher quality voices for festival
22:37:35 Quit barrywardell ()
22:38:07 Quit fml ("Bye!")
22:38:48 Join Horschti [0] (n=Horscht@xbmc/user/horscht)
22:39:02 Quit Strife1989 ("Huzzah!")
22:43:03Bagderfestvox-rablpc16k - British English male speaker for festival, 16khz sample rate
22:44:19Bagder"festvox-rablpc16k: Depends: festlex-oald (>= 1.3.1) but it is not installable"
22:44:35 Part woyciesjes ("Ex-Chat")
22:44:48 Join woyciesjes [0] (
22:45:12 Part woyciesjes ("Ex-Chat")
22:46:54 Quit Thundercloud (Remote closed the connection)
22:47:08archivatorBagder: it's not so much about the sample rate as it is about the algorithms used to transform the corpus into the voices. There are several voices from the ARCTIC corpora that are way different from each other - some algorithms optimize for size, others for accuracy. The corpora are generally well-balanced..
22:47:14 Join __lifeless [0] (n=lifeless@
22:47:30BagderI still prefer installing from debian packages
22:48:21archivatorWell, the installation is nothing more than an untar. No configuration needed.
22:48:29 Join Rabbit^^ [0] (n=Rabbit^^
22:49:26Rabbit^^Just fyi, the article I was referring to earlier was actually in the View forums on Sansa's site.
22:50:58 Join barrywardell [0] (n=barrywar@
22:51:30kugelgevaerts: :)
22:52:10rasherBagder: These voices aren't available as Debian packages, unfortunately
22:53:18archivatorrasher: I would imagine that's because they're HUGE (compared to other packages).
22:56:13amiconnBagder: Isn't festlex-rablpc from non-free?
22:56:14 Quit __lifeless (Read error: 145 (Connection timed out))
22:56:26amiconn(thinking about licensing issues)
22:56:29BagderI don't know
22:56:29 Quit Horscht (Read error: 110 (Connection timed out))
22:56:48Bagderdoes festival have non-free voices?
22:56:55archivatorI would like to add festival support to rbutil. The only problem is, unlike the other engines, festival has a server-client model. I would need to first start the server and then pass it commands to generate the clips. Any idea how to do that without rewriting all of tts.cpp?
22:57:38archivatorBagder: yes. The University of Edinburgh offer several voices for commercial use - you can hear a demo on their website - they're quite good!
22:58:11 Quit Rabbit^^ ("Off")
22:58:29amiconnIirc supports festival
22:58:59 Join thegeek_ [0] (
22:59:09*amiconn always uses sapi for voice & talk file generation, so he's not sure
22:59:09archivatoramiconn: from what I can tell, the multisyn voices are public domain. Either that, or they fall under the X11 license of the source code.
22:59:47*Llorean really liked the mbrola voices with espeak.
22:59:54LloreanI seem to recall that voices built with them were redistributable.
23:00:41 Quit _lifeless (Connection timed out)
23:00:45archivatorLlorean: I never tried mbrola. It requires a separate executable, right?
23:01:11bluebrotherarchivator: start the server in the constructor, stop it in the destructor and call the client like the executables right now?
23:01:24amiconnAh, festvox-rablpc is from contrib, because it depends on a non-free package (festlex-oald)
23:01:36Lloreanarchivator: Yeah, it does unfortunately.
23:01:39 Quit low_light (" ajax IRC Client")
23:02:00Lloreanarchivator: It sounds really good though, and I tend to feel espeak is much better at choosing the right pronunciation for words (at least in my limited experience)
23:03:01 Part pyro_mania1
23:03:02archivatorbluebrother: The server can take up to 5 sec to start and I don't know of a way to be sure it works. I'll see what I can do, though, thanks for the tip.
23:04:57bluebrotherhmm, that might be problematic then.
23:07:01bluebrotheryou could try to utilitze start() and stop() for the server, and pass a signal with the status
23:07:24bluebrothernot sure if any of those work −− I'm not familiar with that part of the code at all
23:12:02Bagderthe multisyn 98MB contains not a single word about installation
23:12:28Bagderbut 4506 files ;-)
23:14:10 Join _lifeless [0] (n=lifeless@
23:14:25archivatorBagder: right. Do you know where festival keeps its voices in your distro?
23:14:37archivatorBagder: Ubuntu uses /usr/share/festival/
23:14:41Bagderbut I guess I can find out
23:14:56 Quit thegeek (Read error: 110 (Connection timed out))
23:15:02archivatorBagder: do that. You need to find festival/voices/ and create a festival/voices-multisyn
23:15:36Bagderso without a license in it either, how do we know these are fine?
23:15:48archivatorBagder: I don't remember what the tar structure looks like but it needs to be festival/voices-multisyn/english/cstr_...._multisyn
23:15:58 Join thegeek [0] (
23:16:13archivatorBagder: read this
23:16:53archivatorBagder: right. Open up festival from a console (it loads in interactive mode) and type (voice.list) Make sure this voice is there.
23:17:27 Join gartral [0] (n=gareth@
23:17:59 Quit {phoenix} (Remote closed the connection)
23:18:11Bagderamazingly awkward procedure
23:18:52archivatorBagder: tell me about it. The whole thing is in Scheme. Anyway, I guess you saw it there. You need to switch to it with (voice_cstr_...)
23:18:52 Quit FlynDice (Remote closed the connection)
23:18:58archivatorIt has tab-completion
23:19:43 Quit gartral ("Leaving.")
23:20:14 Join gartral [0] (
23:21:05archivatorBagder: You can make it speak to /dev/dsp with (SayText "Text.") Of course, if you have pulseaudio, you need to run it under padsp..
23:21:32Bagderno I can't, since this is on the server
23:21:39 Join FlynDice [0] (
23:21:46 Join Strife89 [0] (n=nds@
23:21:51Bagderwell, I guess in theory I can make pulse play remotely
23:21:57Bagderbut I have no desire
23:22:01 Quit thegeek ("( :: NoNameScript 4.2 :: )")
23:22:03 Join akur [0] (
23:22:47archivatorBagder: I see. If you're running any tests, just make sure you have trailing fullstops - the multisyn models change intonation at fullstops.
23:23:02BagderI do rockbox voices
23:23:07Bagderfor all targets
23:23:19BagderI think that can be seen as a test ;-)
23:23:38Bagdercoming up in ... 7 hours
23:23:48archivatorRight. Didn't know that.
23:23:49 Quit Strife89 ("I need to read.")
23:23:57archivatorWell, then, good luck.
23:24:30jhMikeSkugel: yessir?
23:24:40*Bagder remembers to clean the voice-pool
23:25:10kugeljhMikeS: I was wondering about your comment here:
23:26:02kugelI haven't actually noticed a difference after removing lcd_set_enable_hook() from the e200's lcd_enable
23:26:36 Join rocko [0] (
23:27:10jhMikeSkugel: you mean with old data being displayed?
23:27:23rasherOh, forgot about the full stops
23:27:44kugelthe display updated within a tick (it seems to me at least), no matter of lcd_set_enable_hook()
23:27:50 Quit thegeek_ (Read error: 110 (Connection timed out))
23:29:07 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
23:29:15saratogatheres a datasheet for that AMD SOC with some USB info, I added it to the wiki
23:29:47 Join bs66_1 [0] (
23:29:53BagderAMS surely
23:30:13Bagderuh no
23:30:17LloreanWasn't the USB thing identical to an AMD SOC?
23:30:18*Bagder shuts up and hides
23:30:19jhMikeSI was just thinking that for this application, you might want the data updated before anything is visible and that provides a way to do it in sequence. For mpegplayer, the overlay won't be refreshed in time if paused or in "set start time".
23:30:46BagderI wasn't aware of this
23:31:30 Part akur
23:31:40 Quit Zoxc ()
23:31:56kugeljhMikeS: well, the e200 does lcd_update() and lcd_set_enable_hook(), and the old data is still visible for a split second (nothing that bothers me actually)
23:32:33kugelthis isn't actually visible with default settings though, you need to disable backlight fade and lcd_sleep
23:32:46Bagdera huge document it is
23:32:47jhMikeSI find that sort of thing looks sloppy
23:33:19gevaertssaratoga: nice
23:33:36 Quit maddler (Read error: 110 (Connection timed out))
23:33:44kugeljhMikeS: but adding lcd_set_enable_hook() apparently doesn't do it
23:35:02jhMikeSwhere is the code?
23:35:10jhMikeSin the patch now?
23:35:25kugelthe e200 does it in SVN
23:35:25saratogathe 5536 is a SOC, but since the register maps are reported to match, they probably licensed the same USB controller [whatever that may be]
23:35:42Bagdersaratoga: we know it's a Synopsys
23:36:01jhMikeSkugel: you need to register a function to be called that should only be registered while actually in the screen.
23:36:02saratogaah ok
23:36:35jhMikeSit does nothing at all if you don't have that in place
23:36:37Bagderbut there's no mention of that in the pdf, I checked ;-)
23:37:12Bagderthat linux code is even gplv2 or later
23:37:27kugeljhMikeS: sending a message will still issue the update for the next tick, right (not instant)?
23:37:54jhMikeSit it sleeping on a queue or using sleep()?
23:38:25kugelhuh? there's no sleep involved
23:38:34kugelit just returns if lcd_enable is false
23:38:38kugellcd_enabled() I mean
23:38:52kugelwps_refresh in this case
23:41:35jhMikeSI see get_action(..., HZ/5);
23:41:40Bagderrasher: won't unconditional full stops at the end risk making weird phrases?
23:41:57Bagderserial_port: "19200"
23:42:02kugeljhMikeS: where?
23:42:16Bagderor possibly "Bar (Gradient Colour)"
23:42:20jhMikeSkugel: gwps.c
23:42:35rasherBagder: Maybe :\
23:42:42Bagderwell, I guess we'll find out
23:42:55Bagder"File/directory exists. Overwrite?"
23:43:15rasherSure it'll *look* odd, but will the output be different?
23:43:27Bagderquite possibly not
23:45:17jhMikeSkugel: mpegplayer creates a custom system message and posts it to the button queue when the callback is called. When it gets that it updates the overlay.
23:45:31 Join MethoS- [0] (
23:45:43Unhelpfulthis IF_PRIO macro... do we sim thread priority?
23:45:48 Quit archivator ()
23:45:52 Quit bs66_ (Read error: 110 (Connection timed out))
23:46:50jhMikeSUnhelpful: Not now but who knows, it's not impossible.
23:47:49Unhelpfulhrm, that might be the difference in behavior, the PF loader thread is set to background priority, and it lags behind the display thread on target but not sim, despite *lots* of yielding in the display thread.
23:48:15jhMikeSwhy is it background priority?
23:48:54kugeljhMikeS: well, to me it seems that it can't get instant anyway, can it?
23:49:22kugelI'll look at it, but I don't really thing it's worth it
23:49:34UnhelpfuljhMikeS: i don't know, that's how it was written when i found it :)
23:50:38 Quit bluebrother ("leaving")
23:51:16jhMikeSUnhelpful: hmmm, it does alot of sleeping but really, the UI is dependent upon it and so it should match that.
23:51:57Unhelpfulwell, not *dependent* on it - the blank slide is preloaded, and displayed for anything not cached yet
23:52:04jhMikeSkugel: It can get instant if it's ok to just update on the backlight thread, sure but it should happen in milliseconds through a post.
23:52:24 Quit jgarvey ("Leaving")
23:53:28kugelit's happening within milliseconds as of now
23:54:39jhMikeSkugel: it must lag a bit if it's visible though
23:55:33 Join moos [0] (
23:55:45 Join killan [0] (
23:56:54 Quit CaptainKwel (" ajax IRC Client")
23:57:03kugeljhMikeS: so, I can basically say lcd_set_enable_hook(gui_wps_refresh), and add lcd_call_enable_hook in all the drivers?
23:58:02 Quit flydutch ("/* empty */")
23:58:05kugelhm, no
23:58:10jhMikeSkugel: I think it's in any driver that disables the lcd (or was at the time I added it). x5,H10,e200,etc.
23:58:12 Join crashd_ [0] (
23:58:21 Quit parafin (
23:58:21 Quit MethoS- (
23:58:21 Quit blithe (
23:58:21 Quit denes (
23:58:21 Quit Llorean (
23:58:21 Quit J-23 (
23:58:21 Quit mcuelenaere (
23:58:21 Quit _Auron_ (
23:58:21 Quit balou (
23:58:21 Quit proudfoot (
23:58:21 Quit bertrik (
23:58:21 Quit Seed (
23:58:21 Quit evilnick (
23:58:21 Quit BlakeJohnson86 (
23:58:21 Quit Beta2K (
23:58:21 Quit freqmod_gq (
23:58:21 Quit SUSaiyan (
23:58:21 Quit desowin_ (
23:58:21 Quit rasher (
23:58:32NJoinMethoS- [0] (
23:58:32NJoinblithe [0] (
23:58:32NJoindenes [0] (
23:58:32NJoinproudfoot [0] (
23:58:32NJoinbertrik [0] (
23:58:32 Join Llorean [0] (n=DarkkOne@rockbox/administrator/Llorean)
23:58:32NJoinJ-23 [0] (
23:58:32NJoinmcuelenaere [0] (n=mcuelena@rockbox/developer/mcuelenaere)
23:58:32NJoin_Auron_ [0] (
23:58:32NJoinSeed [0] (
23:58:32NJoinevilnick [0] (i=0c140464@gateway/web/ajax/
23:58:32NJoinBlakeJohnson86 [0] (
23:58:32NJoinBeta2K [0] (
23:58:32NJoinbalou [0] (
23:58:32NJoinparafin [0] (
23:58:32NJoinfreqmod_gq [0] (
23:58:32NJoinSUSaiyan [0] (
23:58:32NJoindesowin_ [0] (n=desowin@atheme/member/desowin)
23:58:32NJoinrasher [50] (n=rasher@rockbox/developer/rasher)
23:58:58jhMikeS(undesired freenode ride)

Previous day | Next day