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 2010-05-13

00:00:35bluebroas for where to put it ... well, working with JdGordon in the skinenginebreak branch sound like a good idea to me.
00:00:55bieberI'm not thinking about integrating it into the theme engine myself, but if I could write my code in such a way that it could be built and used from the rest of Rockbox's code base without modification, I'd imagine that would make integration significantly easier, and make sure that the two files always stay perfectly in sync
00:01:26bluebroif the new skin engine is not dependent on the wps format it would also make sense to simply do it as parser rewrite in trunk.
00:01:43bluebrobut that obviously depends on how flexible that skin parser would become :)
00:02:00bluebroyeah, that would be good.
00:02:01pixelmabertrik: the TEA list is missing the OndioFM (and the Samsung tuner too), both were built into the OndioFM
00:02:08bieberIf I'm not mistaken, the only really significant change I'd have to make over typical coding style is using static buffers instead of malloc()?
00:03:21pixelmaand the Samsung one was also in *some* Archos V2-Recorders
00:04:38bluebrobieber: well, the libwps repository is c++ so there are some more differences. Or are you referring to a different code base?
00:04:49pixelmabertrik: oh, and the Iaudio X5s have an unmarked tuner chip which seems to be a TEA one too (same instructions work on them)
00:04:54bluebroand yes, you would need to use static buffers, Rockbox does not have malloc()
00:05:06 Quit TheSeven ()
00:05:18pixelmaand maybe M3 as well, not a 100% sure though, amiconn?
00:05:32bertrikpixelma, I just searched the wiki when making that page, I apparently missed a few
00:05:50gevaertsbluebro: depends on how you define "static". There's of course the skin buffer to fill up
00:06:15*AlexP just added the H100
00:06:20CIA-5New commit by 03alle (r25974): Make table headers bold; fix a typo
00:06:41bieberA different code base: I'm going to go ahead and build a pure C parser that just disassembles the WPS file into simple structs without checking the semantics, and then another function to check all the tags against a description file to make sure they're valid
00:06:57bluebrogevaerts: well, I guess the parsed skin should not use any dynamically allocated buffer, right?
00:07:18bieberMy goal is to get away from the masses of boilerplate and separate class files that were starting to pile up and use something that's less labor-intensive and easier to modify
00:07:51CIA-5New commit by 03alle (r25975): Use a nice apostroph
00:08:51amiconnThe iAudio M3 and X5 have the D&A DA-201 or DA-202, which is actually a module containing the TEA-5767 (or at least a compatible chip)
00:09:03gevaertsbluebro: again, define "dynamically allocated" :)
00:10:35 Quit robin0800 (Remote host closed the connection)
00:11:07*bluebro rather tired tonight
00:11:53bluebrogevaerts: right ... how dynamic is the skin buffer currently?
00:12:19gevaertsbluebro: the total buffer itself is static, but wps elements are presumably allocated inside it
00:12:38gevaertsThere's no deallocation or anything like that though
00:13:17bluebrowell, my guess would be that we want to keep this. Unless someone has a better idea :)
00:13:43bluebrobut AFAIU this is as dynamic as that buffer would be without a real malloc
00:13:44gevaertsIndeed. I'm not sure if that still falls within the scope of the parser itself though
00:14:30 Join robin0800 [0] (
00:14:37bluebroshouldn't the parser also store the tokens it recognizes? Or should the parser simply output a stream of tokens, leaving it to the next layer to store them somewhere / somehow?
00:15:05 Join robin0800_ [0] (
00:15:29gevaertsI think that depends on how similar the requirements are for the actual skin parser, checkwps, and the theme editor
00:15:30 Join robin0800__ [0] (
00:15:52gevaertsi.e. something for bieber and JdGordon to work out :)
00:16:22bieberHere's the plan I've got for memory management, please tell me if it's horribly wrong:
00:16:24bluebroas the displaying code needs to retrieve the tokens independently it might make sense to separate some token storage layer. OTOH simply defining one and using it all over the place should be sufficient too :)
00:16:46bluebrogevaerts: definitely something for bieber and JdGordon to discuss :)
00:17:09CIA-5New commit by 03alle (r25976): More nice apostrophes
00:17:52 Quit robin0800_ (Remote host closed the connection)
00:17:59 Quit robin0800 (Read error: Connection reset by peer)
00:18:03bieberI'll declare one large buffer of struct skin_element and use them to hold all the skin elements. They're a linked data structure, so elements of the array will be pointing to each other, but I'll keep a count of how many are active, and when a new one is needed just increment that counter and use the next element in the array.
00:18:52*pixelma wished that fml was around more in IRC during or after commits
00:19:11bieberAnd then for strings, when I need them, I have another large array of char's allocated, keep track of the last element used, and whenever I need storage for a string I just use some more space in there and hold on to a pointer to the beginning of it? Or is there another way that you normally do linked data structures without malloc()?
00:19:54gevaertsbieber: what's a "struct skin_element" exactly?
00:20:43gevaertswell, *more* exactly :)
00:20:50*gevaerts doesn't need the real full details
00:20:59bieberIt'll be the basic element of a WPS file. It can hold either a newline, plain text, a standard tag with name and parameters, or a conditional with name, parameter, and struct skin_element pointers for children
00:21:35gevaertsWon't storing them all in an array be wasteful then?
00:22:05bieberIt is, but is there another way to do it without using malloc()?
00:23:04gevaertsThat's basically what I was talking about with bluebro a few minutes ago
00:23:07bluebrowhat's the difference between holding a newline or plain text?
00:23:31gevaertsYou basically use the mechanism you described for strings, or something very similar
00:23:32 Quit robin0800__ (Remote host closed the connection)
00:24:40bluebroalso, without a real malloc a possible way could be to assume the memory to hold a skin element structure with a pointer to the text (which could be NULL) and a pointer to the next element. You could put the text immediately after the structure, and set the pointer to the next element just after that text.
00:24:40gevaertsBut again, talk to JdGordon about it. He knows the details of how it currently works, and he presumably has some sort of idea on how he wants it to work
00:24:49bluebronot sure if it's a good idea though.
00:24:53bieberIt would be the same mechanism, it's just that they have to have pointers to each other as well, since the entire relationship can't be modeled with just a linear array
00:25:08 Quit bertrik (Remote host closed the connection)
00:25:15 Join fml [0] (
00:25:36fmlpixelma: you wanted to talk to me?
00:26:07bluebrobut yeah, JdGordon is your man :) Hopefully I'll manage to read up how it's currently done −− I really don't know how the skin tokens are currently stored
00:26:32bieberbluebro: Just declare, say, a char[] and cast pointers to the struct type when I need to use them?
00:26:52fmlpixelma: if you just say something without telling to whom you are speaking I'll probably miss it.
00:27:18gevaertsbieber: yes. Taking care of alignment of course
00:27:25bluebrobieber: yes
00:28:08gevaertsI don't know if you want a linked list though. A tree might work better, but I haven't thought about it much
00:28:42pixelmafml: I meant this (a) in general and (b) specifically, I don't understand the frequency thing (I don't see the point personally and I can't see anywhere how the new tag, or whatever it is, is used)
00:28:42 Quit liar (Ping timeout: 258 seconds)
00:28:55bieberThe structure I'm thinking about is basically both. For the most part, the structure is linear, progressing from one element to the next. It's only when it gets to a conditional that it branches
00:29:34fmlpixelma: you'd then better ask JdGordon who designed the tokens. I just optimized the code.
00:29:48gevaertsI could see it branch at viewports as well, but I'm definitely not sure
00:30:07fmlpixelma: but IIUC the tokens are used to display the current/min/max frequency
00:30:24pixelma"create a formatting function for FM frequency " is not an addition by you? the custom FM radio screen
00:30:33gevaertsBut I don't think the actual data structures after parsing is what JdGordon desperately wants to get rid of, so looking at the code probably gives a good idea of what to do
00:31:07fmlpixelma: I just factored out duplicated code. Have you looked into the diff?
00:31:26pixelmawell yes, doesn't tell me anything though
00:31:33pixelmain this case
00:32:00pixelmaand the commit message lead me to believe otherwise
00:32:11biebergevaerts: What exactly do you mean by alignment? As long as I increment my next_space or w/e index by sizeof(my struct) every time I insert one, I should be clear, right?
00:33:14gevaertsbieber: that's if the structure is always the same, which I suspect it's not. You basically want to align on multiples of four bytes
00:33:22fmlpixelma: the message maybe should have been "avoid code duplication *through* introducing blah"
00:33:49bieberIs that just for efficiency's sake? And can I assume that a char will be one byte on all of our targets?
00:33:55*gevaerts apologises for dragging details into this that probably aren't relevant yet
00:34:40gevaertsno, it's to make things actually work. On e.g. arm you can't access an int (which would be in your struct) at an address that's not a multiple of 4, and you can't access a short at an odd address
00:35:42bieberOh my. So as long as every struct starts on a 4-byte multiple, the compiler will make sure the struct's memory is laid out so that it works correctly?
00:35:43pixelmafml: that would probably not have helped my understanding, the "introducing function" sounds more like a new functionality for a token. I didn't understand it was something "behind the scenes"
00:36:08gevaertsyes, unless you specify that it's to be packed, which (in this case) would obviously a bad idea
00:36:22fmlpixelma: yes, it was a purely technical commit, no visible functional changes
00:37:11pixelmamaybe that could have been mentioned somewhere :)
00:37:32pixelma"no functional changes" or somesuch
00:37:37fmlpixelma: now it has been mentioned here! :-)
00:37:47bieberGot it. And treating 4 chars as 4 bytes will work fine wrt alignment?
00:38:14gevaertsyes. char==byte on all reasonable systems :)
00:38:57fmlpixelma: if it would bring new functionality, I'd also update the manual. Which is (pity!) not in sync at some places.
00:39:09gevaertsIf you've ever seen a data abort, this is what that is. Some code tried to dereference an unaligned pointer
00:39:25bieberSounds good, thanks for the primer
00:39:33pixelmafml: yeah, I guess especially not in this part
00:40:07bieberI've never really written serious code for a microprocessor :/
00:40:49*pixelma is reminded of her few half-finished manual patches and hopes to find some time tomorrow on national holiday (not related to FM skin though)
00:41:13fmlpixelma: Some WPS tags are not in the manual. VI, Vi, radio... But I have to go now. The radio tokens are described in the wiki, so it should be straight forward to copy them to the manual.
00:41:19CIA-5New commit by 03nls (r25977): Simplify special case function, speedup of about 0.2MHz on both coldfire and pp decoding the files in the test set
00:41:38 Quit fml (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
00:41:57 Quit joeyg (Quit: WeeChat 0.3.0)
00:42:17pixelmafml: the tags might be but..
00:42:33gevaertsbieber: in most cases you won't notice anything about this. It's just when you do things like memory management or handling binary data from external sources that the alignment things can bite
00:45:29 Quit raphi_ (Quit: leaving...)
00:48:04 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
00:49:53 Join josh_ [0] (~josh@
00:52:19AlexPI'm planning to update WPS/FMS tags in the manual tomorrow (today!)
00:52:28 Quit Gump (Ping timeout: 258 seconds)
00:54:43AlexPJdGordon: (for when you are about) - %?tm doesn't seem respected in the menus - e.g. I am in preset mode, so %?tm is false. I exit the fms to menu with radio playing, so my %?mp tag in my sbs calls a specific viewport, which has %?tm<aaa|bbb>. bbb should be displayed as I was in preset mode in the fms, but aaa is displayed. aaa is also displayed if I left the fms in scan mode.
00:54:55pixelmagreat :) I just wondered if the appendix could be restructured a bit (or at least renamed)
00:55:18AlexPyeah, an intro of sorts would also be handy I think
00:56:28 Quit pamaury (Quit: Page closed)
01:01:55 Join apoelstra [0] (
01:02:00 Nick apoelstra is now known as joeyg (
01:02:41 Quit bluebro (Quit: sleep)
01:03:43AlexPJdGordon: The progress bar also doesn't seem to respect the currently playing thing in menus - I exit the fms with radio playing to menu, and the progress bar goes back to zero instead of showing the position of the current frequency on the bar
01:03:51AlexPSorry to only spot these now :)
01:06:08 Join liar [0] (
01:06:12 Join stripwax [0] (
01:06:18 Quit petur (Quit: ZZzzzz)
01:07:36 Quit solexx (Ping timeout: 240 seconds)
01:08:29*n1s wants to point out that accessing a char buffer via a struct pointer is breaking that strict aliasing stuff i.e. it isn't standard c, not that i have thought out a nice way to do it...
01:09:05AlexPJdGordon: I added them as
01:09:28 Join solexx [0] (
01:16:39stripwaxn1s - is r25977 yours?
01:17:01n1sstripwax: yes
01:17:43n1swell, it's les than 1% improvement :)
01:17:47stripwaxsome of that (existing) code is sortof weird. e.g. if read<chunk return -1, but we do that after the for/while loop. seems like we should test before rather than after.
01:19:09stripwaxand read is either < chunk or == chunk, as far as i can tell. so if( read<chunk)return-1 followed by n-=read*dim/2 is also kindof weird. we should already know read*dim/2 = chunk*dim/2 by this point. and we had to do a divide by dim to GET chunk in the first place
01:20:01stripwaxah wait, we only do a divide in one special case.
01:24:06n1sḯm not going to pretend i understand it completely
01:24:54stripwaxwondering if things might be quicker without the divide - if 16*dim>n then just don't use the optimised thing at all
01:24:57stripwaxdivides scare me
01:25:12 Quit stripwax (Quit:
01:25:17n1syes, divides are scary
01:26:22n1si had another patch that i thought should speed this special case thing up by calling the special case function directly instead of through the general case function but it slowed down cf for some reason...
01:35:36n1sI mean, it pops 6 args from the stack, and then pushes them again, just to call the special case, and add to that the 8 cycles of call+return and another pop of the same args, so an overhead of some 30 cycles per call...
01:35:44 Quit kramer3d (Quit: Leaving)
01:45:54 Join Soap-N810 [0] (
01:48:31 Quit S_a_i_n_t_ (Ping timeout: 252 seconds)
01:49:26***Saving seen data "./dancer.seen"
01:54:18 Join S_a_i_n_t [0] (S_a_i_n_t@
01:55:03 Quit DerPapst (Quit: Leaving.)
01:55:35 Join fdinel [0] (
02:05:50 Quit togetic (Quit: WeeChat 0.3.0)
02:12:10 Quit efyx (Remote host closed the connection)
02:12:49 Join togetic [0] (~togetic@unaffiliated/ibuffy)
02:16:38 Part toffe82
02:21:24 Join Rob2223 [0] (
02:25:07 Quit Rob2222 (Ping timeout: 248 seconds)
02:26:25 Join antil33t [0] (
02:26:26 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)
02:27:06 Quit n1s (Quit: Lmnar)
02:29:23 Quit josh_ (Ping timeout: 248 seconds)
02:33:37 Quit Soap-N810 (Ping timeout: 252 seconds)
02:45:03 Quit MethoS- (Read error: Connection reset by peer)
02:51:13S_a_i_n_tHmmmmm, OK. Serious question...possible flamewar. 3.6 is looming, we need a new screen for the "default theme", do we just make a new screen for it, or do we update the whole damn theme staring from scratch?
02:51:36S_a_i_n_tI realise this will be quite a job, but the cabbie code is looking quite elderly.
02:52:22S_a_i_n_tnew release == new face for RB?
02:53:10S_a_i_n_tI'm happy to work with like minded individuals, a small group could get it done in a reasonably short time.
02:53:53LloreanS_a_i_n_t: The default theme contest was marked as ongoing. At any point in time you're welcome to create a new theme and propose it as a replacement default.
02:55:07S_a_i_n_tthat's the thing though, I think a group of people that have a similar idea could get this done a lot better, rathe rthan me sitting on the sidelines working saying "do you like this, nope, what about this? etc."
02:55:11S_a_i_n_tjust an idea.
02:55:40LloreanOne way or another, someone's going to need to make the first proposal (visually)
02:55:41S_a_i_n_tWe *do* need to think about viewports, and progressbar logic for volume/battery whether we change the theme or not.
02:56:05S_a_i_n_talso, not having %C everywhere ;)
02:56:06LloreanAnd no, we don't need to think about viewports or progressbar logic. They don't *need* to be changed. We *can* think about them, though.
02:56:44S_a_i_n_tbut, you're all for marginal decreases in size arent you Llorean?
02:57:05S_a_i_n_tif 10K whets your whistle
02:57:43LloreanS_a_i_n_t: I was against that 10k because the problem could be solved better in other ways.
02:58:05S_a_i_n_tsame here, we can reduce total size by doing things better.
02:58:16LloreanAs it stands, reducing the theme size won't necessarily reduce RAM usage.
02:58:18S_a_i_n_t*but* I *don't* want to argue...
02:58:50LloreanMy Gigabeat currently is using 251K less than the amount allocated for the theme, apparently
02:58:58S_a_i_n_tif dynamic buffers ever go in, then reducing a theme as much as possible makes total sense.
02:59:10LloreanYes, but they aren't in now.
02:59:55LloreanWhile a major overhaul of the default theme is possible for the release, it's something you should get working on first, and try to attract people to once you've got something to show for it.
02:59:58S_a_i_n_tthey will be one day though (fingers crossed), and cabbie should be updated as it is the "face" of RB...if its just me that's interested in doing it, fine.
03:00:03LloreanIf you just want to update the viewport and progress bar logic, just post a patch.
03:00:06S_a_i_n_tbut its a massive job for one person.
03:00:40S_a_i_n_tand I'll need a lot of on DAP testing anyway.
03:00:55LloreanReally? Isn't everything that would be changing in the WPS stuff that would show in the sim?
03:01:14S_a_i_n_tcolours don't look the same etc.
03:01:24S_a_i_n_tI *hate* making things too bright/dark.
03:01:28S_a_i_n_ta pet peeve of mine.
03:01:56LloreanWell, a big part of that will depend on the set backlight brightness
03:02:00LloreanAnd the opinion of the viewer
03:02:05LloreanAnd the age of his player in some cases.
03:02:36LloreanSince it's not going to be one designer with access to all the players, the sim should be more than adequate anyway unless you know someone who has the same eyes as you.
03:03:05LloreanBesides, if you're just updating the existing Cabbie, the colors are already set, no?
03:03:35S_a_i_n_tNew GFX will need to be made in some cases...there is also the FM skin.
03:03:53S_a_i_n_tI'm trying to find out if ZincAlloy has SVGs for the current Cabbie though
03:03:53LloreanIt doesn't change the fact that you can just grab the RGB values from existing images
03:03:58S_a_i_n_tthat would help.
03:05:31 Quit whydoubt (Ping timeout: 252 seconds)
03:06:05 Join ischeriad [0] (
03:09:08 Quit ischeria1 (Ping timeout: 245 seconds)
03:12:43 Join whydoubt [0] (
03:21:13 Join josh_ [0] (~josh@
03:33:14S_a_i_n_tis "button_beep_freq: 5 and button_beep_len: 0" valid config file settings anymore?
03:33:25*S_a_i_n_t admits to never having seen these before.
03:37:53S_a_i_n_t is confusing me, the config is from '06 but he reports only having trouble with the backlight. Unless I'm *really* blind, the settings for backlight appear fine.
03:37:58 Part josh_ ("Leaving")
03:41:21bieberJdGordon: Whenever you come around, I've got some questions about your WPS language changes. Specifically, will ',' now be an escaped character, and will it ever be legal for a tag with no argument list (say %pv) to be followed immediately by plain text without a space in between?
03:42:42JdGordonnice timing
03:42:54bieberIf the latter is possible, then it won't be possible to write a parser that can deconstruct a WPS file just by its syntax, because %xy could be either a two-letter tag or a one-letter tag followed by a letter of plaintext
03:43:13JdGordon, would only need to be escaped inside param lists, and yes to the second
03:44:11JdGordonok, so we put a listing of each tag in the parser?
03:44:48bieberEither that or require an empty () after tags with no params
03:45:24bieberOtherwise, yeah, I'll have to build the syntax parser with some kind of knowledge of the legal tags
03:45:39JdGordonis that a big deal?
03:46:22bieberNot terrible, we'll just have to decide on how to best specify the tags and their formats before I can get the parser together
03:46:56bieberThat's assuming there are single-character tags with no parameters. Are there any of those?
03:47:17bieberOh, escaped characters are
03:47:24JdGordonand ;
03:47:57JdGordonwe can simplify this later when/if we require all tags to be the same length
03:48:56bieberSo right now I'm thinking the easiest way to do this is to include a text-file that specifies the available tags and their number/type of parameters
03:49:29***Saving seen data "./dancer.seen"
03:49:45bieberAnd I can just write the parser to take full advantage of knowing when it encounters a tag exactly what parameters it should be expecting
03:50:05JdGordonmake it simpler... just add a struct at the top struct tagtable { char tag[], char params[] };
03:50:37JdGordonparams would then be something like ddddds (d being digits, s being string)
03:51:14bieberThere's no way to embed a struct constant in C, is there?
03:51:23 Quit adnyxo (Read error: Operation timed out)
03:52:22JdGordon scroll down a bit untill you see the table
03:52:23bieberLike I can declare an array of strings with {"...", "..."}, but there's no way to do the same thing with structs, iirc
03:54:01tmztdon't gnu extensions support that?
03:54:02bieberOh, apparently you can
03:55:51JdGordonbieber: also, stick your code at the top in utils/themeeditor or just themeeditor/ for now
03:55:56JdGordonbut in that branch
03:56:25bieberOkay, so right now I've been fleshing out two files, generic_parser.h and generic_parser.c. Should I put a tag table like that in generic_parser.h, or give it its own header file?
03:56:55 Quit wincent_balin (Ping timeout: 260 seconds)
03:57:02JdGordoncode goes in .c files, not .h
03:57:10S_a_i_n_twow, it seems every time I look at a list of skin tokens I fond something new...
03:57:16S_a_i_n_t*find even
03:57:47JdGordonbieber: so a new .c file probably with a nice function find_tag(char* tag)
03:57:48bieberShould this be considered code, since it's all constant?
03:58:14JdGordonyes, doing that means it is dead simple to replace at a later time (r another project :) )
03:58:30JdGordonand dont worry about not using malloc if you want to
03:58:36*JdGordon is reading last nights log
03:59:18bieberI already threw together some functions for allocating space in a static buffer, why don't I throw up a patch with what I've got so far and see if you like it?
04:00:11JdGordonok, no, if you've done the code already then cool
04:00:14S_a_i_n_tJdGordon: at a rough long has %pE and %pS been there for?
04:00:19JdGordonI was just saying use malloc to make it easier
04:00:24S_a_i_n_ttrack starting/track ending
04:00:33JdGordonS_a_i_n_t: wasnt it your request for them? :)
04:00:42S_a_i_n_tyes, it was....
04:00:49S_a_i_n_tI never knew you did it though!!!
04:01:09JdGordonsince r25239
04:01:35JdGordon2010-03-18 18:08:49 +1100 (Thu, 18 Mar 2010)
04:01:40S_a_i_n_twow...I need to start highlighting commits I believe.
04:02:14S_a_i_n_tI needed those for my cheap-ass AA transitions.
04:03:10 Join Strife89|Laptop [0] (~Strife89|
04:04:38JdGordonbluebrother: the current tokens are stored in a very boring array, sublines/conditionals need special handling to work
04:05:43JdGordonbieber: sublines are s sort of condtional also... the difference is the branch is done based on time and not some tag
04:06:25bieberI wasn't really sure if I should try storing them in a structure when parsing, or just let the viewer code take care of it later
04:06:48bieberI'm thinking the former is probably the best idea, make it like a conditional
04:07:11 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:07:11 Quit pixelma (Disconnected by services)
04:07:31 Quit amiconn (Disconnected by services)
04:07:31 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:07:33 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:07:55 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:10:07JdGordonyeah, do it properly.. the internal displayer is going to probably need to be redone also, so make the parser actually make sense... a tree sounds much better than a array
04:11:07bieberTo submit a patch, I "Add New Task" in Flyspray with the type "Patches" and attach the file?
04:11:20S_a_i_n_tare we still going to have the ability to do "subline(timeout), subline(timeout), subline(timeout)" or is it all moving to "(timeout) subline, subline, subline"?
04:11:28*S_a_i_n_t hopes that is parseable
04:11:58JdGordonbieber: yeah
04:12:19JdGordonS_a_i_n_t: I hope the former
04:12:49S_a_i_n_tsweet, so do I, but the latter won't be *too* much a pain in the ass if it makes everything else a lot easier.
04:13:36bieberNow, my patch file only specifies the file paths within the branch: is there anything special I need to do to specify that it gets applied to your branch+
04:14:33JdGordondo a checkout of that branch and then work as usual
04:16:51bieberOkay, I just submitted FS #11265. If you get a chance to look at it, tell me if I'm doing anything wrong from the beginning
04:20:02JdGordonwps_alloc_string() looks odd
04:20:29JdGordonyeah, I still think #if 0 that code and just use malloc() inside those allocs for the time being
04:20:52JdGordonyou dont want to be stuck with memory issues so early on
04:21:15JdGordonkeep with wps_alloc_string() calls and just use malloc inside that func
04:21:56JdGordonSEMICOLON element should be SUBLINE
04:22:10JdGordonalso replace wps everywhere with skin
04:22:18bieberI just changed that one, and okay
04:22:37bieberIs there anything in particular you don't like about wps_alloc_string()? As near as I can tell, it works correctly
04:23:17JdGordonwhy current_block * 4?
04:23:45biebergevaerts said to keep things lined up on 4-char blocks, so I'm just counting blocks instead of bytes
04:24:47JdGordonchar* 's dont need to be i think, and it looks like you are wasting a.... OH.. right
04:25:03bieberYeah, I'm also packing structs into that same array
04:26:02JdGordonbuffer_front = (void *)(((unsigned long)buffer_front + 3) & ~3); <- simle one liner to keep the pointer aligned
04:26:59bieberOkay, & and ~ are bitwise AND and NOT
04:27:42bieberSo we're adding three, and forcing the last two bits to zero
04:27:44bieberThat's clever
04:28:32bieberSo I'll just count bytes instead of blocks, and throw that in at the end of each malloc function?
04:29:16JdGordonreally, just use malloc for now... the editor has the OS to do memory management for it
04:30:29 Part Strife89|Laptop
04:31:36 Quit joeyg (Quit: WeeChat 0.3.0)
04:33:31 Join joeyg [0] (
04:33:36 Quit joeyg (Client Quit)
04:38:25 Join Boldfilter_ [0] (
04:40:14 Quit Boldfilter (Ping timeout: 240 seconds)
04:40:14 Nick Boldfilter_ is now known as Boldfilter (
04:43:39 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
04:45:07 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
04:49:00bieberJdGordon: Tag parameter types should just be integer, character, or string, right?
04:49:48 Join joeyg [0] (
04:49:54JdGordonsounds about right
04:50:26JdGordonmaybe filename also, but that could be tricky because of the folder not being there
04:51:14bieberWell, the parser will know ahead of time what it's looking for
04:51:23bieberSo being able to tell them apart isn't really an issue
04:52:03JdGordonI was thinking that the parser could complain if the file isnt there, but that probblay isnt good
04:53:08bieberIf you want, I can add a filename type. My parser will just treat it as a string, but if you're integrating it you can have it check, since the file _does_ need to be there when the theme's actually being loaded on the device
04:53:36JdGordonalso, some tags will have (probably at most) one optional arg
04:54:13JdGordonI'm not sure how you want to handle that, it sounds like you want to convert the params into the actual types instead of leaving them as strings?
04:54:32 Quit dys (Ping timeout: 276 seconds)
04:54:42 Join dys [0] (
04:54:53 Join Barahir_ [0] (
04:55:42bieberRight now the format I'm thinking of for the control strings is something where "Ii|S" would mean that you have two integers with an optional string at the end, and if a character is lowercase it means you can select a default with '-'
04:55:43JdGordonwhich I guess means {"Vi", "cddddd" } is different to {"Vi", "ddddd"}
04:56:27JdGordonthat could work
04:56:56bieberThe characters I'm looking at are I, C, S, and F, for integer, character, string, and file name
04:57:04bieberSound reasonable?
04:58:01JdGordonscrap c I think
04:58:17bieberJust have it use strings?
04:58:18 Quit Barahir (Ping timeout: 248 seconds)
04:59:48JdGordonwill this be able to handle the playlist viewer tag which has 2 string params which contain skin code?
05:01:21 Quit mc2739 (Ping timeout: 258 seconds)
05:03:04 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
05:03:26JdGordonUnhelpful: can i bug you about AA?
05:03:40UnhelpfulJdGordon: what's the question?
05:04:17JdGordonwhy is the dim stored seperatly in playback.c instead of using width and height directly from the skin token?
05:04:34JdGordonthe dim vals are static arnt they?
05:06:01Unhelpfuli... have no idea?
05:06:38JdGordonisnt aa your thing?
05:07:18bieberJdGordon: That should be along the same lines as conditionals, it will just recursively parse those parameters as code
05:08:20JdGordonwill that happen automatically?
05:08:26Unhelpfulrockbox had AA long before i got involved. i've looked a bit at the code that actually loads it, and rather a lot at the image loaders, but as to the general design of our implementation? there's a lot i've never seen or touched.
05:08:37JdGordonnuts, ok
05:09:29bieberI just need to include a code character for the parameter control strings, how about using C for that?
05:10:17bieberThen the parser will just run the recursive calls when it encounters those, and store the results in the children array of the skin_element
05:10:40JdGordonsounds good
05:11:02UnhelpfulJdGordon: you're talking about struct albumart_slot in playback.c? i had notihng to do with multi-AA but it looks like what it does is make a unique-ified list of requested AA sizes
05:11:39Unhelpfulin the case that, for some reason, you specify AA twice with the same dimensions, it will give both the same slot.
05:12:14 Quit mc2739 (Ping timeout: 276 seconds)
05:12:18JdGordonok, but still sounds like it could come straught from the skin
05:13:10JdGordongetting radio art working in the current system cleanly is a bit annoying
05:13:30 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
05:13:57Unhelpfulalso the whole "slot" thing doesn't even actually ensure that you don't buffer the same file scaled to the same size twice
05:14:39Unhelpfulsuppose you load a large, square album art, and the WPS asks for 40x40 and 40x80 - the output image for both will be the same size.
05:16:09JdGordonso plenty of room for improvement :)
05:18:22Unhelpfulwell, i want to do something about making the buffer allocations more dynamic, and part of that plan included refcounting and checking for attempts to buffer things already buffered
05:18:55Unhelpfulin the case of AA we could easily make such a scheme work for multi-AA by labeling the file as /path/to/file/<loaded_dimensions>
05:23:26S_a_i_n_tJdGordon: just looked at the FM screen/theme thread again...perhaps I have different ideas on how things could be done?
05:27:01S_a_i_n_tIs it going to be a lot simpler than I thought?
05:27:39biebernoob question I'm sure, but can I change the file attached to a Flyspray task I submitted?
05:29:50JdGordonS_a_i_n_t: this is mc2739's screenshot
05:30:31JdGordonbieber: you should eb able to.. depends on permissions... or beter to just add a new comment
05:31:44S_a_i_n_tI had a *similar* idea. I imagine "is stereo" to be useful as an icon though?
05:31:57bieberOkay, I just attached a comment with a new patch. It has a sample tag table with just a couple entries, and a really simple search function for it (FS #11265)
05:32:09JdGordonS_a_i_n_t: yeah, probably
05:32:59JdGordonbieber: comments for what those characters mean would be nice :)
05:33:03JdGordondo you have commit access yet?
05:33:23bieberThere's a big 'ol comment block in the .h file that explains it, should I copy it over to the .c too?
05:33:35JdGordonah, so there is.. na
05:34:33bieberAnd no, I don't have commit access
05:35:07JdGordoncan you make | only make the next item completly optional? those viewport tags have the first param optional
05:36:29bieberOh, right, the group has to be made optional together
05:36:55bieberIs there ever a case where there's more than one optional group like that?
05:37:30JdGordonI cant tihnk of any
05:38:00bieberI could just let you insert a number before a character to specify the number of times that type must occur
05:38:37bieberSo then it would be IIiii|2s, which would specify that if there's anything there, it has to be two strings together
05:39:23JdGordonthat wont fix %Vi where the first param is the optional one.. all the others are required
05:39:36JdGordonand it is nice having the first param as the label
05:40:16bieberIsn't that first param required?
05:41:16JdGordonI suppose we could change it to required but -'able
05:41:21bieberWouldn't you use V instead of Vi if you weren't specifying an identifier?
05:41:27S_a_i_n_tJdGordon: Are the bitmapstrip GFX already present (+some new ones) being recycled in the FM skin? Or should this be when we get CabbieV2 to use progressbar-type volume/battery also (+more effort)?
05:42:15JdGordonI havnt seen the code for that screenshot.. Ideally it should be as close the the WPS as posible (to make sense)... but it doesnt have to be
05:42:57JdGordonusing the same images would make it all easier
05:44:42 Quit Topy (Ping timeout: 260 seconds)
05:44:49JdGordonand straight away giving it art support would be good... not sure what to put there if the image isnt found though
05:45:08JdGordonmaybe the playlist viewer, or just a massive freq display
05:45:32S_a_i_n_tfor some reason, one thing I'm not a fan of so much is the "now playing" not being something specific to the radio in that screenshot.
05:45:42S_a_i_n_tother than that, it looks nice.
05:45:55JdGordonyeah, me too... so fix it :)
05:46:04 Join Topy44 [0] (
05:46:14S_a_i_n_tI wanna know if ZincAlloy has SVGs
05:46:25S_a_i_n_tor if I should just be butchering bitmaps ;)
05:46:26bieberJdGordon: None of those messages were aimed at me, right?
05:47:08JdGordonno, I apparently missed your question
05:47:31bieberWouldn't you use V instead of Vi if you weren't specifying an identifier?
05:47:59JdGordon%V is a regular viewport, %Vl is a conditional viewport, %Vi is the UI viewport (the one the lists go in), %Vi can have an optional id, kugel was thinking we shuold get rid of %Vl and just use the conditional id on %V also
05:48:25 Part Boldfilter
05:49:31***Saving seen data "./dancer.seen"
05:50:55bieberWhy don't I do both, then? When you have items grouped together that are optional, the number notation will be helpful (as well as just helping to compress definitions), and if you want to have an optional parameter at the beginning the single character will work best
05:51:18bieberSay a ? after a character to make it optional
05:52:13bieberSounds good then
05:52:45bieberTomorrow I should be able to hammer out most if not all of the parser
05:52:51JdGordonawesome :)
05:52:57bieberHeh, talking about it on IRC ended up taking an awful lot longer than I expected
05:53:13JdGordontalk is more fun than actual doing
05:53:23bieberOf course ;)
05:53:47bieberDo I get SVN commit access if you apply my patch, or will I need to submit a public key or something?
05:54:18JdGordonyou need to talk to Bagder to setup a user/pass...
05:54:36 Quit Strife89 (Quit: Rebooting to Windows to try and get this friggin' webcam to do something. XD)
05:54:39bieberIs he/she around often?
05:55:04JdGordonyes, but he's in sweden so the timezones wont line up much
05:55:57bieberIs he usually around in the morning or afternoon his local time?
05:58:27 Join Strife89 [0] (
06:02:57 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
06:09:45 Join shai [0] (
06:12:28CIA-5New commit by 03jethead71 (r25978): Trim down peak calculation a bit.
06:19:10 Quit bmbl (Ping timeout: 258 seconds)
06:23:34 Quit BHSPitMonkey (Remote host closed the connection)
06:26:01S_a_i_n_tthought of something to save skinbuffer with the FM screen, but it involves changing the WPS slightly. I guess, basically, it's cutting the "now playing" out of the Cabbie backdrop, adding a "radio screen" equivalent, and displaying them conditionally with %cs...
06:26:17S_a_i_n_tthat way the FM screen can use the Cabbie backdrop and not look weird.
06:26:47S_a_i_n_tor, the "radio screen" title could just be drawn overtop of the "now playing", but that's kinda hacky.
06:33:34 Quit shai (Max SendQ exceeded)
06:34:20 Join shai [0] (
06:40:16 Quit joeyg (Quit: lions and tigers and bears, oh my!)
06:42:08JdGordonits not hacky
06:42:25JdGordoneither way, using the same backdrop image would be nice
06:42:39JdGordonusing an overlay image in the fms means only 1 extra bmp
06:47:35S_a_i_n_tthat's true...I quess I just have an aversion to drawing overtop of things. But, it is the backdrop so I guess it's not bad.
06:48:20 Join mikroflops_ [0] (
06:51:06JdGordoneither way, using the same backdrop means there should be plenty of room to load a font or two
06:52:12 Quit mikroflops (Ping timeout: 252 seconds)
06:52:44S_a_i_n_tthere should indeed be, I haven't worked out the saving you'd get from chucking out the bitmapstrips for volume/battery...but I suspect it's around ~10-15Kb
06:54:06JdGordoni doubt it
06:54:09JdGordonnot that much
06:54:37JdGordonthe volume bmp is 5KB on e200
06:55:00JdGordonbattery is 6KB.. so yeah ok maybe together
06:56:32S_a_i_n_tI said ~10-15, that fits! ;)
06:57:28S_a_i_n_tlarger screen == bigger saving though I suppose.
07:13:55 Join kifo [0] (
07:15:10 Join kifo_ [0] (
07:15:36kifo_Hi, I might be blind and stupid but I've been looking for 15 minutes and can't find the clip+ manual
07:17:24kifo_and I did try google btw
07:18:28kifo_says its included in the release, however the link to the current release isnt on that page, I think I have the current release, but it isnt in that archive either
07:18:54 Quit kifo (Quit: CGI:IRC (Ping timeout))
07:19:24S_a_i_n_tIt is listed as unusable.
07:19:48kifo_I thought it would still have a manual
07:19:55S_a_i_n_tDo manuals get compiled for unasuable targets? I don't know...
07:20:08kifo_doesnt make sense why its unusable
07:20:23kifo_but I'm only looking at the outside of it I guess not really a dev person so
07:20:38kifo_the release has made great progress in the last what 2 weeks or so anyways
07:20:44kifo_it was listed as unstable at one point
07:20:56kifo_I dont suppose there is a chance that manual still exists maybe?
07:21:21kifo_honestly, I'm just confused about how to rate song is all, although the manual would be great to have.
07:21:35S_a_i_n_t*maybe*, but, whatever exists you'd probably need to compile yourself.
07:21:53S_a_i_n_trate a song by using the context menu in the WPS
07:22:13kifo_It may be that it isn't supported
07:22:28S_a_i_n_tcall the context menu, then select set song rating.
07:22:37kifo_the context menu doesn't seem to have that option
07:22:38S_a_i_n_tthat'll be how to do it, if you can do it.
07:22:49 Quit Horscht (Ping timeout: 248 seconds)
07:23:23S_a_i_n_tI don't have that target, but it should be the same.
07:23:34 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
07:23:39S_a_i_n_tare you calling the context menu in the file tree of the WPS?
07:23:49S_a_i_n_t(they have different selections)
07:24:00S_a_i_n_ts/of the/or the/
07:24:02kifo_File tree of the wps?
07:24:04 Quit Strife89 (Quit: Rebooting.)
07:24:18 Join esperegu [0] (~quassel@
07:24:21S_a_i_n_t"file tree" or "wps" sorry.
07:24:37kifo_ive tried both
07:24:50kifo_when you say it has a different target do you mean it might be under a different name?
07:25:03kifo_I think I've tried all of them but that would be a quick double check
07:25:25 Join n1s [0] (~n1s@rockbox/developer/n1s)
07:25:26S_a_i_n_tno, by "I don;t have that target" I mean, I don't own a Clip+
07:25:37kifo_I guess it just couldn't be supported yet, some short time ago the clip+ didnt have write permissions
07:25:41S_a_i_n_tbut, I thought if it can do it, it is probably done in a similar way.
07:25:42kifo_so it may have been overlooked
07:26:08CIA-5New commit by 03funman (r25979): fuzev2: current estimation assuming same battery capacity than v1
07:26:13kifo_not exactly a critical feature hah
07:26:14CIA-5New commit by 03funman (r25980): as3525: use DMA for recording ...
07:26:24S_a_i_n_tPerhaps, but even so...this is why it's classed as unusable still I guess.
07:26:36S_a_i_n_tamong other things
07:27:54kifo_its so strange because it doesn't seem buggy at all, I mean it hasn't crashed on me and I've been torture testing it trying to see if it's worth installing again
07:28:10kifo_if it isn't unstable soon I'd be suprised just from what I gather but
07:28:20S_a_i_n_tAre you playing files from microSD?
07:28:36kifo_sdhc also too
07:28:39kifo_which is amazing
07:28:45 Join Boldfilter [0] (
07:28:52kifo_It only crashes if you play from sd during database update
07:29:03kifo_thats the only way I've gotten it to crash during sd/hc reading
07:29:09S_a_i_n_tHmmm...then that is a good sign. The port has made some significant jumps lately.
07:30:19S_a_i_n_tI'm not sure exactly how Stable/Unstable/Unusable are defined, but there will be a good reason it's still where it is though.
07:30:34kifo_Yea, no plugins crash either, even clock crashed not even 3 weeks ago
07:30:39mc2739kifo_: You might check the clip manual. It should be fairly close to the clip+
07:31:06kifo_mc2739: yea I'm looking through it now, seems mostly the same, other than a few wheel discrepencies
07:31:26kifo_S_a_i_n_t: incredibly it even follows playlists across internal memory to sdhc
07:31:39kifo_:D so yea ill stop clogging up your chat
07:31:40kifo_thank you!
07:31:47 Quit kifo_ (Quit: snaotheusnatohisatneu,
07:38:41 Join Strife89 [0] (
07:43:13 Join BHSPitMini [0] (
07:45:31 Quit arbingordon (Quit: `)
07:49:35***Saving seen data "./dancer.seen"
07:56:42AlexPS_a_i_n_t: The default theme isn't set in stone by any means, but what you need to do is make a visual proposal - do a mockup, showing how it would adapt to different screen sizes etc., and then if it is decided that it is preferred to Cabbie v2, then you can start working on it.
07:56:50AlexPJdGordon: Yo!
07:57:07AlexPThanks for finally getting fms in :)
07:57:21AlexPDid you see my comment earlier?
07:57:49JdGordonyeah, I'm looking at it now
07:58:06 Join ender` [0] (
07:58:32AlexPSorry to only spot it at this point
07:58:50JdGordonno worries
07:58:54CIA-5New commit by 03funman (r25981): as3525: use a table for mclk dividers, get rid of a division
07:59:04JdGordonI dont mind fixing bugs when they are found :)
07:59:14AlexPfms is looking nice though :)
07:59:36JdGordonand will look even better when radio art is added!
07:59:44AlexPyeah, that'll be ace
08:02:11 Quit mc2739 (Ping timeout: 265 seconds)
08:02:37 Quit JdGordon (Read error: No route to host)
08:03:16 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
08:03:16 Join Ipsi [0] (
08:03:50 Quit hd (Quit: Ω)
08:06:15 Join funman [0] (~fun@rockbox/developer/funman)
08:07:28funmani messed up the fuzev2 bootloader, it shows 'Boot Ver. Ver. 2.0' (2 times Ver.)
08:07:56 Join evilwombat [0] (~marsupial@
08:08:02evilwombatstevenm here
08:08:14evilwombatI see you guys are still maintaining the MIDI plugin.
08:10:29n1sevilwombat: not much has been done to it lately, I'd like to turn it into a codec but haven't even started figuring out how yet
08:10:55evilwombatn1s, I can see this being a bit interesting, with the patchset and all
08:11:08evilwombatbecause they are not accessed in a streamed fashion
08:12:03evilwombatthe "simple" way would be to just blow away the music buffer and fill it entirely with patches
08:12:05n1syeah, we'd need to have a way to alloc space for the patches on the audio buffer
08:12:20 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
08:12:25evilwombati can see just discarding the buffer, using it, and returning it to music when midi is done
08:12:39evilwombatbut, performance hit / having to refill it from the drive again
08:12:43n1sevilwombat: but the whole point of having it as a codec would be so you could have MIDI's in regular playlists
08:12:56evilwombatn1s, yeah, that's true
08:13:02funmanBagder: can you put on please ? (last fix I promise)
08:13:32evilwombatthe elegant way would be to find a way to grant it space in the buffer when a midi file comes up
08:14:10evilwombatbut you probably know a lot more about that architecture than I ever did
08:14:18 Join xavieran [0] (
08:15:19n1sunfortunately i don't know the buffering code well :/
08:15:55evilwombatone thing i can think of is letting midi allocate from the end of the audio buffer, going backwards
08:16:07evilwombatie, wiping out the undecoded songs that are already there
08:16:13evilwombatbut starting with the "last" song first
08:16:23evilwombat(and, properly removing them from the buffer, I mean)
08:17:13evilwombatconceptually it makes sense.. most recent songs (ie, next in the playlist) might be left untouched, but songs that would have been played later in the playlist would have to be reloaded
08:17:32evilwombatbut how to implement this in practice, I am not sure
08:17:50n1smaybe, the whole problem is that we need all the patches for as long as a file is playing, while the buffering system is designed for streaming files that don't need to keep their used data around
08:18:04n1syeah, i don't know either
08:18:48JdGordonwe need a better way of knowing what is controlling playback
08:19:12evilwombatis it possible to introduce a "shim" codec that would call the plugin when a mid file comes up, and then reload the contents of the audio buffer when it's done?
08:20:20n1sprobably, but i don't think such a solution is a good idea nor that it would be accepted into svn
08:20:42evilwombatn1s, yeah, you're probably right
08:20:47evilwombatbut yeah, sleep here also
08:20:59evilwombatthanks again for the maintenance!
08:21:53JdGordonAlexP: for the progressbar... I have it so it will show the freq in the menus as long as the radio isnt off.. is that good enough?
08:22:12AlexPI think so, yes
08:22:14JdGordonin the sim at leats you can be in the radio screen with the radio off
08:22:53AlexPyes, I noticed that - when you first enter the fms it doesn't show anything until you press "play", but on target it is fine
08:23:35JdGordonnone of my radio targets are working atm so its hard to test :(
08:26:00 Join jd [0] (
08:26:00 Quit jd (Changing host)
08:26:00 Join jd [0] (~jd@Wikipedia/HellDragon)
08:26:15AlexPI'm happy to if desired :)
08:28:52JdGordonit looks like the mode is set back to scan mode when the fm screen exits (although i cant see where)
08:28:58JdGordonwhich causes the other problem
08:29:23JdGordonradio.c really needs to be redone
08:30:58JdGordonAH, presets are cleared on exit for some reason!
08:31:28JdGordonany ideas why that happens? that doesnt make much sense
08:32:21 Quit jd (Read error: Connection reset by peer)
08:32:40 Join jd [0] (
08:32:40 Quit jd (Changing host)
08:32:40 Join jd [0] (~jd@Wikipedia/HellDragon)
08:32:57JdGordonno, thats not right
08:33:49JdGordon:O yes it is
08:34:43evilwombatanyways, sleep time is now. good night all
08:34:45 Quit evilwombat (Quit: ^D)
08:35:13 Quit BHSPitMini (Quit: Ex-Chat)
08:36:26 Join godeater [0] (~godeater@rockbox/staff/GodEater)
08:37:12 Part jd
08:38:07JdGordonAlexP: ok yeah can you test a quick patch?
08:40:06JdGordonthe big thing to test there is that presets work on exit and reenter
08:41:13JdGordonfunman: moving them from unusable or unstable?
08:41:30funmanyes, did i make a typo?
08:41:42JdGordonthe last 2 sentances make no sense
08:41:55JdGordonThe reason for removing them from unstable category has been removed in r25924, so I see no blocker.
08:41:55JdGordonOK for moving them to unstable category?
08:42:32funmanthey had been removed from unstable for a reason, and this reason isn't valid anymore since r25924
08:43:04 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
08:44:20 Join stoffel [0] (
08:44:23JdGordonmy e200v1 can boot into rockbox but not OF, it is useing a really old bootloader which tries booting to OF on usb connect.. how do i fix the OF? i cant get to recovery mode
08:44:29JdGordonwith rbutil be able to fix that?
08:44:34S_a_i_n_tJdGordon: Had a thought, not sure if it could be done though...and it'd be pure eye-candy, is a "%boosting?<y|n>" tag possible?
08:44:52S_a_i_n_t-spelling mistakes
08:44:52 Join flydutch [0] (
08:44:55AlexPpointless IMO
08:45:24S_a_i_n_tnot entirely pointless, but pure eye-candy.
08:45:44JdGordonAlexP: it is no less pointless than disk access...
08:45:59AlexPI disagree there :)
08:46:44AlexPI don't actually care though :)
08:46:50S_a_i_n_tIt's something I wouldn't mind seeing on an .sbs, I'm not sure *why* exactly...but I like to animate everything.
08:47:14JdGordonit would be stupidly trivial to add
08:47:35S_a_i_n_tsounds promising...
08:48:13JdGordonall glory to rbutil!
08:48:41 Quit YPSY (Ping timeout: 248 seconds)
08:48:48S_a_i_n_term, OK...
08:49:00*S_a_i_n_t bows to his new deity.
08:49:17AlexPJdGordon: Patch works fine - exit radio in preset mode and sbs shows me preset name and the pb is the frequency, exit in scan mode and the sbs shows me frequency and the pb is frequency too :)
08:49:48JdGordonAlexP: we need zstarsales04 on the forums banned please
08:49:54 Quit leavittx (Ping timeout: 252 seconds)
08:50:14JdGordonhe's adding posts faster than i can delete them
08:51:20 Join Ypsy [0] (
08:51:25 Nick Ypsy is now known as YPSY (
08:54:06JdGordonAlexP: and presets work fine if you are in preset mode, exit to menu, then go back in?
08:55:26JdGordonaaw, rbutil cant install the nightly build? LAME!
08:55:28AlexPseem to - preset is selected, exit to menu same preset fine, go back in, still on same preset and can then skip around them as normal
08:55:55AlexPhuh, progress bar isn't working now
08:55:59AlexPWhat did I do?
08:56:07 Quit bieber (Quit: No Ping reply in 180 seconds.)
08:56:23CIA-5New commit by 03jdgordon (r25982): fix FS #11264 - frequency bar and presets not working in the sbs
08:57:01*godeater is puzzled by the sansa email to the dev list
08:57:07godeaterwho keeps moving them out of unstable?
08:57:18AlexPJdGordon: Right, fm progress bar doesn't work after playing music
08:57:20JdGordonAlexP: "now" being where?
08:58:02AlexPSo boot, use FM - all fine - play a file, works fine there - go back into fm and the pb is stuck wherever it was for the file when you enter fm screen
08:58:32 Join efyx [0] (
08:59:18CIA-5New commit by 03jdgordon (r25983): fix the freq bar after music plays
08:59:23JdGordon^ not the best fix, but good enough for now
08:59:23AlexPta :)
09:00:12ranmafunman: Haven't tested your patch yet, but I remeasured the currents on a better multimeter and got RB idle 28mA, RB mp3 38mA VS OF idle 26mA, OF mp3 28mA (didn't change of course :))
09:00:54ranmaThe cheap multimeter I have here probably has problems with non-constant currents.
09:01:01funmanranma: i tested it with battery bench it has no effect
09:01:02 Join bieber [0] (
09:01:02 Quit bieber (Read error: Connection reset by peer)
09:01:30ranmaThe THS multimeter is a better one, probably TRMS (Hmm, forgot to note that, should check that next tuesday...)
09:01:30funmanyou have a digital one?
09:03:39 Join leavittx [0] (~leavittx@
09:03:53ranmafunman: The multimeters are both digital
09:05:36funmancan you measure an infinite loop reading and/or writing memory in OF & rockbox?
09:05:54ranmaSpeaking of port stability, how about moving C200v2 to unstable?
09:05:57 Quit preglow (Ping timeout: 260 seconds)
09:06:30funmandoes it crash during playback like the clipv1 ?
09:06:54 Part Boldfilter
09:07:06funmanmine doesn't afaict but according to it happens for some people
09:08:31 Join preglow [0] (
09:10:01ranmaIt doesn't crash, but there are still issues.
09:10:45ranmaFor example during battery bench runs I have had it stop recording the runtime, but it still played fine.
09:12:01funmanhm but well the clipv1 is unstable too so i guess it's ok
09:12:05AlexPIs it true that wps files etc need to be 24 characters of less to be remembered?
09:12:47godeateras in the filename?
09:12:52funmanranma: there just needs to be a bootloader for c200v2
09:12:54godeaterI'm sure they can contain more than that ;)
09:12:59 Quit preglow (Ping timeout: 240 seconds)
09:13:12AlexPThere is a note on the manual to that effect, and I think it is probably obselete
09:13:33funmanranma: does changing DMA_S1 to DMA_S4 in pcm-as3525.c change the current usage?
09:15:45 Quit krazykit (Ping timeout: 260 seconds)
09:15:57ranmaI kind of doubt that going from 1 (byte?) to 4 will make a measurable difference :)
09:16:35ranmaMeasuring a memory read/write loop shouldn't be too difficult.
09:17:01funmanranma: it's 1 word (32bits) to 4 but why wouldn't it ?
09:17:05ranmaAnd given that'll mean pretty much constant current maybe my cheap multimeter will even give useful values in that case.
09:17:14funmanfifo is 128 words, i'll try with higher bursts
09:17:28 Join krazykit [0] (~kkit@
09:17:28ranmaThen I'd rather try comparing 1 agains 128 :)
09:17:44ranmaOr rather 64
09:17:47funmanusually the burst is half the fifo size i think
09:18:10funmanDMA_S64 gives too fast playback
09:19:17*ranma is currently working on the usb stuff a bit
09:20:14ranmaI'll go to Akihabara later to buy some parts to make a USB cable where I can measure the D+/D- level to see if the pullup is active or not and also insert a pullup manually.
09:20:57ranmaI think on some models GPIO_A1 controls the D+ pullup, but on my HW variant that code just skips that part completely...
09:22:17funmanis it just fuze and clip keymaps, or it's not possible to leave recording menu without stopping recording?
09:22:42funmanaccording to apps/recorder/recording.c:1393 it seems it's not possible at all
09:22:47 Quit whydoubt (Ping timeout: 264 seconds)
09:23:09funmanranma: DMA_S16 works fine
09:23:20ranmaFWIW 0x30035A5C is the 'usb_otg_reset' code from otg_functio
09:25:43ranmaIt's called from sub_10b2c, which also calls toggle_A1 at the end (but toggle_A1 is a noop on my HW)
09:27:01ranmaThe problem is, so far I haven't gotten any USB interrupts and no reaction from my PC, so my current guess is I'm having trouble with the D+-pullup
09:28:16ranmafunman: IIRC you got your PC to see the device sometimes? Can you send me your patch again, I can't find it anymore. ^^;
09:30:24funman (march 26th)
09:31:40 Join bertrik [0] (
09:32:54bertrikhi funman, I see you committed the AMS DMA recording patch. Does it fully work now?
09:32:57 Join jfc^2 [0] (
09:33:19funmanbertrik: yes, i wouldn't have committed if it didn't ;)
09:33:40bertrikcool, what was the problem?
09:33:58 Quit bertrik (Changing host)
09:33:58 Join bertrik [0] (~57d33175@rockbox/developer/bertrik)
09:34:12funmani'm benchmarking recording now, although I don't know where time left estimation can be shown during recording
09:34:14jhMikeSfunman: rec_dma_start_addr isn't updated during the transfer? isn't there a way of getting the current address?
09:34:40funmanbertrik: well there were several, i think i have written them on FS
09:35:03funmanjhMikeS: it is updated (line 225)
09:35:24 Join whydoubt [0] (
09:35:37funmanthis function is called by pcm_rec_dma_start() and at the end of each DMA transfer (by the dma callback)
09:36:38 Quit jfc (Ping timeout: 246 seconds)
09:37:07funmanbertrik: ah no i didn't, bad quality came from bit 2 of I2SIN_CONTROL
09:37:41funmanwe must set it to 1 = "data valid at positive edge of SCLK"
09:37:46funmanbtw this bit is reset to 0 on read
09:38:26funmanah no, this one stays written, but bit 5 (14bits/24bits sample depth) is reset on read
09:38:37jhMikeSthat's sort of coarse I think. really it's supposed to be the address the hardware is at.
09:39:08 Join preglow [0] (
09:39:11 Quit bertrik (Quit: CGI:IRC (Ping timeout))
09:39:33funmanjhMikeS: what's wrong then?
09:40:46 Join fml [0] (
09:40:47funmanrec_dma_start_addr = X, transfer N bytes, rec_dma_start_addr = X+N, transfer N bytes, rec_dma_start_addr = X+N+N, ...
09:40:48jhMikeSisn't there a status reg that gives the *current* position of the current transfer in hardware? (or bytes remaining)
09:40:57funmanbytes granularity ?
09:41:48jhMikeSit should have word granularity, not the block setup by hardware, but the readout of the pointer
09:42:23funmani think it's possible
09:42:25jhMikeSdon't use portal player as a model, it has special handicaps
09:42:53fmlHello JdGordon. Isn't there a bug in skin_engine/skin_tokens.c:401-402: what if the offset is not 0? The the value of preset will be not -1 and... What is this preset_offset parameter for?
09:43:03jhMikeSeverything else has a size remaining or current position status register (normal things, not pp)
09:43:55 Quit preglow (Ping timeout: 276 seconds)
09:43:56 Join vaguerant_ [0] (
09:45:21 Join preglow [0] (
09:47:06jhMikeSwell, pp has it, but reading it also clears the interrupt which makes it stop working when it reads just before it needs to do fiq
09:47:07 Quit vaguerant_ (Read error: Connection reset by peer)
09:47:10 Join vaguerant [0] (
09:47:10 Quit vaguerant (Changing host)
09:47:10 Join vaguerant [0] (~vague@wikipedia/vague-rant)
09:47:50vaguerantSorry, my client sucks, I'm going to upgrade to a web-based java client, back shortly.
09:48:39 Part vaguerant
09:49:39***Saving seen data "./dancer.seen"
09:50:00 Quit preglow (Ping timeout: 248 seconds)
09:51:34funmanjhMikeS: would it explain why the pitchmeter doesn't work ?
09:51:49funmanpeakmeter* and pitchdetector*
09:54:03jhMikeScould be, it would also be rough. it updates the SrcAddr and DestAddr as the transfer progresses according to the DS. Those need physical addresses and to be word-aligned to the current packed 16-bit sample pair
09:55:07funmani can read the current DMA destination address but for some reason it's below rec_dma_start_addr?
09:55:27fmlJdGordon: I'd propose to change the logic of the %tm tag (radio scan/preset mode) from boolean to enum. For the themers the usage would be the same, but those are equaly values
09:55:31 Quit Forsaken_Boy (Ping timeout: 264 seconds)
09:55:41n1sjhMikeS: since you seem to be the resident pcm guru, coudl you look at FS #10635 ?
09:55:45jhMikeSfunman: it also says TransferSize also counts down during the transfer
09:56:00jhMikeSn1s: sure, why not :)
09:56:09fmlJdGordon: err... accidental ENTER press. They should be treated equally, like e.g. playback mode
09:56:24jhMikeSfunman: below?
09:57:26 Join Forsaken_Boy [0] (~chatzilla@
09:57:26 Join preglow [0] (
09:57:45jhMikeSfunman: because the address is virtual and the DMA put a physical address there perhaps?
09:57:55JdGordonfml: ?
09:58:08 Join petur [0] (~petur@rockbox/developer/petur)
09:58:23fmlJdGordon: what do you not understand?
09:58:42JdGordonnothing, I get it.. ok, if you want to :)
09:59:16JdGordonand the preset_offset is so those tags work in the playlist viewer
09:59:20JdGordonusually they will be 0
09:59:41fmlJdGordon: usually? But not always?
09:59:50funmanjhMikeS: no addresses are flat mapped
09:59:56 Join bieber [0] (
09:59:59fmlJdGordon: then we have a bug
10:00:08JdGordonline 402 was a bugfix from umm... someone..
10:00:16JdGordonfrom mc2739
10:00:26 Quit antil33t (Ping timeout: 246 seconds)
10:00:34fmlJdGordon: what does the offset mean? When is it not 0?
10:01:12jhMikeSfunman: yeah, I just looked. it's cached though? problem there is how do read/write memory as *not* cached? it seems you can't, without an uncached mirroring of it elsewhere.
10:01:17JdGordonit means how far away from the current is the preset we want the info from
10:01:38 Quit preglow (Ping timeout: 240 seconds)
10:01:41funmanjhMikeS: well there is UNCACHED_ADDR() macro for that?
10:02:05JdGordonhmm, it wont work on negative offsets if the current preset+offset == -1
10:03:07fmlJdGordon: yes. My proposal: if preset < 0 return NULL
10:03:26JdGordonnegative offsets should work
10:03:29jhMikeSfunman: if cached address is the same address as physical address and there's no uncached mapping elsewhere, how would that work?
10:03:38fmlJdGordon: the following logic (squeezing preset into the range) is screwed too IMO
10:03:43CIA-5New commit by 03nls (r25984): Some asm for mdct on coldfire, speeds up vorbis decoding by about 0.3MHz
10:03:58 Quit Ipsi (Quit: Leaving)
10:04:04funmanjhMikeS: virtual cached address is the same as physical address, but virtual uncached address is different
10:04:09fmlJdGordon: why should they? Negative offsets should but not negative presets
10:04:41JdGordonthats what the preset += count() is doing
10:04:45jhMikeSfunman: ok, so they mapped it afterall. I'd think it'd be easier the other way.
10:04:51JdGordonif (radio_preset_count() == 0 || radio_current_preset() == -1) is the fix
10:05:06JdGordonif (radio_preset_count() == 0 || radio_current_preset() < 0)
10:05:21 Join notlistening [0] (
10:05:25JdGordonreturn NULL
10:05:28funmanso pcm_rec_dma_get_peak_buffer() should return the uncached address?
10:06:08fmlJdGordon: no. Just replace == -1 with < 0
10:06:29JdGordonno, because preset can legally be <0 at that line
10:06:45JdGordonint preset = radio_current_preset() + preset_offset; <−−−−- preset_offset can be negative
10:07:03fmlJdGordon: yes, but not preset
10:07:16JdGordon0 + negative == <0
10:07:25jhMikeSfunman: if using DMA, yes. both peak buffer functions should return uncached addresses. it's also importanbt for the keyclick to get written into the buffer properly.
10:08:11fmlint preset = radio_current_preset() + preset_offset; <−−- preset_offset can be < 0, that's ok. But if preset is < 0 then we should return NULL
10:08:12funmanok it works: peakmeter is back (only on 1 channel though)
10:08:16jhMikeSn1s: I did play with the pitch detect and had it blow up every time I playback was going. other times it seemed to not work at all. Then, I forgot it even existed. :)
10:09:07 Quit logiclost (Read error: Connection reset by peer)
10:09:23 Join lostlogic [0] (~lostlogic@rockbox/developer/lostlogic)
10:09:43funmanhm no microphone recording is mono, so no problem
10:09:55n1sheh, the funky thing is that playback and recording seems to work at the same time on h300, i dunno if that is even intended? :)
10:10:10fmlJdGordon: i.e. your current preset position is 3 and you want to see the preset with offset -6. What should it be? IMO NULL.
10:10:24 Join komputes [0] (~komputes@ubuntu/member/komputes)
10:10:58funmanjhMikeS: return UNCACHED_ADDR((void*)DMAC_CH_DST_ADDR(1)); << looks correct
10:11:04 Join vaguerant [0] (
10:11:04 Quit vaguerant (Changing host)
10:11:04 Join vaguerant [0] (~vague@wikipedia/vague-rant)
10:11:04 Part vaguerant
10:11:25jhMikeSfunman: I think that's probably correct.
10:11:46funmanthe dma destination register is automatically incremented. The datasheet mentions it's byte aligned but we are transferring 32 bits words so it should be word aligned
10:11:51jhMikeSfunman: the as3525 codec can't put one channel to both? if not, you need to duplicate into the other one.
10:12:01 Join vaguerant [0] (
10:12:19 Quit vaguerant (Changing host)
10:12:19 Join vaguerant [0] (~vaguerant@wikipedia/vague-rant)
10:12:20funmani need to check i thought it was faking stereo already
10:12:20JdGordonfml: hmm, ok I see what you are saying, but if you dont think of the presets as a straight playlist then it does make some sense
10:12:25vaguerantLet's try that again.
10:12:26jhMikeSfunman: if it increments by byte and not transfer size, it should probably add "& ~3"
10:12:37jhMikeSs/transfer size/word size
10:12:38JdGordonthink of it like the playlist has repeat all on
10:12:57funmanjhMikeS: it's byte aligned when transferring bytes, but we don't
10:13:04jhMikeSfunman: I asked because as3514 can't fake stereo (pita)
10:13:09vaguerantDoes mkamsboot work with firmware versions other than those linked in the wiki, or is it limited to the older firmware version linked on the wiki?
10:13:20funmanjhMikeS: hm i probably remember wrong then
10:13:27CIA-5New commit by 03alex (r25985): Fix FS #11196 - update %pv in the manual.
10:13:32CIA-5New commit by 03alex (r25986): Update the discussion of themeing in the manual, and put a note in the wps tags appendix that the actual discussion is elsewhere.
10:13:37fmlJdGordon: yes, the playlist cycles
10:13:39CIA-5New commit by 03alex (r25987): FS #11243 - The %VI (capital i) WPS tag is not described in the manual.
10:13:43CIA-5New commit by 03alex (r25988): Add FMS tags to the manual.
10:14:13JdGordonfml: yes, so a negative prset number just needs to be normalised and it works
10:14:26funmanhm so e200/c200 just duplicate the channel ?
10:16:36 Join preglow [0] (
10:18:04jhMikeSn1s: that "stop recording when buffer is full" stuff is nasty, nasty
10:18:18AlexPJdGordon: As in, I'm at the first preset, I ask for preset -1, I should see the last preset
10:18:25fmlJdGordon: but the we should not compare preset to -1 but rather radio_current_preset (or something)
10:18:26jhMikeSthough it shouldn't cause any difficulty with blowups I'd think (or hope)
10:18:42JdGordonAlexP: yes
10:18:47vaguerantI have a Clip+ that's already Rockboxed but wanted to update to the new bootloader and figured I'd update the OF at the same time, but I'm not sure if it's safe to patch the current firmware or if mkamsboot is only compatible with the specific version linked.
10:18:52JdGordonfml: yes, thats the code i pasted before
10:19:06JdGordonif (radio_preset_count() == 0 || radio_current_preset() < 0) return nULL
10:19:52fmlJdGordon: OK. Then we should move this line one up, before preset is computed
10:19:54funmanvaguerant: mkamsboot will tell you which versions it knows
10:20:01 Quit Forsaken_Boy (Read error: Connection reset by peer)
10:20:05*jhMikeS sees no reason not to use a circular buffer there
10:20:06JdGordonwhy? it makes no actual difference
10:20:22fmlJdGordon: and maybe introduce a local var to avoid many calls of radio_preset_xxx()
10:20:57JdGordongo ahead if you really want to
10:20:59JdGordontucker time
10:21:06CIA-5New commit by 03funman (r25989): as3525: use the correct uncached address for rec peak buffer ...
10:21:28 Quit preglow (Ping timeout: 248 seconds)
10:21:35jhMikeSn1s: what's so bad there is that someone was using it as an example of what to do
10:22:38 Join bluefoxx [0] (
10:23:03bluefoxxAnybody up?
10:23:09 Join preglow [0] (
10:23:54n1sjhMikeS: i haven't looked at the plugin code very closely but yes, it sounds bad...
10:24:01bluefoxxWondering how far along the Gigabeat S port is, if anybody knows. How much can I expect to *not* work?
10:24:07S_a_i_n_tbluefoxx: 126 people in channel ;)
10:24:20jhMikeSn1s: it also doesn't stop playback first :\
10:24:27bluefoxxJust aquired one from a friend as "Broken". Drive was plugged in backwards was all.
10:24:40fmlJdGordon: the patch is here: Any objections?
10:24:41jhMikeSbluefoxx: it's pretty much fully along
10:24:42n1sjhMikeS: i know, on h300 you can have music playing while using it even :)
10:24:48funmandoes pitch_detector even work ?
10:24:57bluefoxxjhMikeS: Is there a walkthrough on how to rockbox it?
10:25:00jhMikeSfunman: not when I tried it
10:25:03n1sbluefoxx: yeah, it's working well
10:25:09bluefoxxBe nice to have another 30Gb alongside my e280
10:25:17funmanit used to not too long ago
10:25:20n1sbluefoxx: there is a manual
10:25:33jhMikeSbluefoxx: try clicking the gigs link on the front page. it's not automatic. you may encounter some issues too.
10:25:37bluefoxxn1s: 'Kay, I'll check
10:25:40bluefoxxjhMikeS: Oh?
10:25:57bluefoxxTBH, I've never used the automagic installs.
10:26:01funman2 months ago it did :)
10:26:12AlexPbluefoxx: Just follow the instructions in the manual
10:26:25funmanbtw the 3.6 release planning just steps over DevCon, doesn't it ?
10:26:41 Quit Kitar|st (Ping timeout: 246 seconds)
10:26:48AlexPBattery life is quite poor, and sometimes the single boot bootloader has trouble finding Rockbox I think is it
10:26:55jhMikeSbluefoxx: there's a problem where immediately after install, the bootloader doesn't find the rockbox bin on the first run. sometimes single boot doesn't work and a restore is forced so you must use dual.
10:27:24AlexPActually, I haven't retested battery life since jhMikeS introduced frequency and voltage scaling
10:27:33AlexPI'm hoping it is a little better now :)
10:27:42jhMikeSAlexP: batt life should have gotten somewhat better. I got two more hours out of mine than before, last bench
10:28:26jhMikeSthat was just running at 264MHz with the CPU at 1.35V
10:28:45AlexPI'm just charging now then I'll test too
10:28:51jhMikeSI'm not sure if 132 helps at all
10:29:10bluefoxxjhMikeS: I'll keep that in mind, thanks :)
10:29:53AlexPjhMikeS: Do you/we have any idea what the single boot issue might be caused by?
10:29:57jhMikeSAlexP: voltage seems more important, and the core sleeps alot and the clock gating is quite efficient. I did a test at 528 and got exactly the same run time as 264. go figure.
10:30:08AlexPheh, odd
10:30:20*fml is waiting for an amen from JdGordon to
10:30:52 Join merbanan [0] (
10:31:05jhMikeSAlexP: I tried some thing. It seems to happen after applying a Toshiba update, so I never did it myself since I don't want the problem.
10:31:18AlexPyes, 1.3 apparently
10:31:23AlexPI also haven't updated :)
10:31:31jhMikeSAlexP: *some things, like padding the bootloader. not much else
10:31:34 Join Kitar|st [0] (
10:31:50AlexPDid anyone with the issue test after your tests?
10:32:37jhMikeSI gave it to them to test. Noone with the issue has worked on it. I'm not sure if you can downgrade it.
10:32:54AlexPYeah, I don't want to upgrade just in case
10:33:20AlexPAnyway, even without single boot always working, is there anything really stopping it being stable? RBUtil maybe?
10:33:22jhMikeSI don't use the OF at all, so personally I don't care.
10:33:28AlexPme neither
10:34:00 Quit preglow (Ping timeout: 240 seconds)
10:34:34vaguerantThanks funman, appreciate the help.
10:34:37 Quit vaguerant (Quit: Java user signed off)
10:34:48jhMikeSAlexP: I think it's the install. I'm not sure where everything left off on that front. I've spent almost all of my time on the firmware. I only did MTP stuff at the beginning, for myself before the USB stack was enabled.
10:35:25AlexPI think there are versions of beastpatcher for all OS now, it "just" needs integrating with rbutil
10:35:38jhMikeSI used OF before charging was working. After that, there was no reason at all.
10:35:40AlexPbluebrother: Is this correct?
10:35:52AlexPjhMikeS: Yes, that was my only use too
10:36:00 Join preglow [0] (
10:36:40CIA-5New commit by 03alle (r25990): Correct detection of the absence of the current preset
10:40:35 Quit preglow (Ping timeout: 268 seconds)
10:41:03CIA-5New commit by 03alle (r25991): Use a nice apostroph
10:41:10jhMikeSn1s: the pitch detector is only if you have mic and/or line?
10:41:15funmanarchosrecorderv2 and iriverh1x0 have spdif input : do they record in 24 bits ?
10:41:21AlexPfml: apostrophe :)
10:41:42*jhMikeS wonders why not SPDIF at least too (it should probably be selectable)
10:42:13bluefoxxI appear to be having some difficulty installing the bootloader for this thing (The gigabeat S). I'm running beastpatcher, and it's telling me everything went fine and dandy, so I reset the thing's USB connection, but it only connects in MTP mode.
10:42:37AlexPWhat OS?
10:42:41n1sjhMikeS: dunno really
10:43:08bluefoxxNo 64bit weirdness here either.
10:43:09fmlAlexP: I see them in the FMS tags as well. Commit will follow shortly
10:43:33 Join Boldfilter [0] (
10:43:43AlexPyeah, could well be - I just copy and pasted
10:44:17AlexPbluefoxx: Did you try single or dual boot?
10:44:21amiconnfunman: We don't have 24 bit recording on any target
10:44:37bluefoxxSingle initially, though dual has also failed to produce results
10:44:51amiconnBtw, archos recorder v1 also has s/pdif in, and it also has s/pdif out, unlike recorder fm/v2
10:44:51bluefoxxI'm getting nothing that lets me access the thing as a plain drive.
10:45:23jhMikeSn1s: actually, it technically could work with playback *if* the target supports full duplex. pp doesn't.
10:45:24CIA-5New commit by 03alle (r25992): Use nice quotation marks
10:45:25AlexPIf the bootloader installed properly you should get the Rockbos bootloader appear which will give you MSC USB
10:45:32amiconnThe archoses recording in mp3 anyway (well, in the core - there is als a wavrecord.rock, which is rather preliminary)
10:45:34AlexPfml: cheat!
10:46:03fmlAlexP: yo!
10:46:30AlexPbluefoxx: Could you paste command line you enter + output etc. to a pastebin?
10:46:58 Join preglow [0] (
10:47:31bluefoxxDo I need any specific settings on the player?
10:47:43bluefoxxOr a specific f/w revision?
10:47:57AlexPshouldn't so
10:48:02amiconnfunman: Iirc the h1x0 could record and play in 24 bit (actually all coldfire targets), but then we couldn't use dma
10:48:03AlexPAnd that looks fine
10:48:15n1sjhMikeS: it doesn't make much sense for the pitch detector thought
10:48:18AlexPbluefoxx: So after it says complete, press enter, you disconnect then reconnect USB?
10:48:26bluefoxxDoes nothing
10:48:32AlexPi doesn't even reboot?
10:48:32JdGordonfml: yeah fine
10:48:42bluefoxxHard reboot produces nothing either.
10:48:51fmlJdGordon: too late :-P
10:49:08AlexPand you press enter before disconnecting?
10:49:16AlexPvery odd
10:49:27 Join Forsaken_Boy [0] (~chatzilla@
10:49:32AlexPI don't suppose you have a linux live CD around to test on?
10:50:15bluefoxxGive me about five minutes to shutdown and dock my laptop, and scavange an optical drive off of a dead laptop
10:50:28bluefoxxI should have an 8.04 disc someplace
10:50:41AlexPSorry :)
10:50:43bluefoxx(Thinkpad's slimline drive went bust)
10:50:50bluefoxxNo problem. Needed to do this anyhow
10:51:00 Quit preglow (Ping timeout: 240 seconds)
10:52:01 Quit fml (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
10:55:57jhMikeSn1s: SPDIF *could*. anyway, it looks like it's DMA-ing the wrong size.
10:56:57n1syeah, i meant regular playback while detecting pitch on input doesn't seem very useful to me :)
10:57:17jhMikeSn1s: oh. thought you were talking about SPDIF. hehe
10:58:28jhMikeSwell, I'll see if this works. I gave it two buffers too so it can record and process in parallel.
10:59:53 Quit Boldfilter (Ping timeout: 246 seconds)
11:01:00bluefoxxBeen needing a good reason to go dig out a few old/dead laptops from my closet anyhow.
11:01:36funmando the iriver record spdif in 16bits? (can't find recording code for recorderv2)
11:02:47amiconn[10:44:23] <amiconn> funman: We don't have 24 bit recording on any target
11:02:54amiconnSo yes
11:03:02amiconns/pdif isn't 24 bit anyway - it's 20 bit
11:03:26 Quit rvvs89 (Read error: Connection reset by peer)
11:04:12 Join preglow [0] (
11:04:40funmanah the format is 20bits and optionally 24
11:06:49 Join rvvs89 [0] (
11:07:46amiconnThe hwcodec recording code is in mpeg.c - it's one big single loop for both playback and recording
11:09:58 Quit preglow (Ping timeout: 240 seconds)
11:10:14jhMikeSn1s: you sure have to scream into the mic for it to work :\
11:10:18 Quit Forsaken_Boy (Ping timeout: 240 seconds)
11:10:18 Quit liar (Ping timeout: 240 seconds)
11:10:52bluefoxxOh bloody hell. The drive I've just pulled is fark'd.
11:11:08 Join liar [0] (
11:11:17n1sjhMikeS: hmm, i rememmber it working at some point
11:13:15*n1s whistles at the c200, seems to work, but requires quite a strong sound, yes
11:13:45 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
11:14:41 Quit rvvs89 (Changing host)
11:14:41 Join rvvs89 [0] (~ivo@pdpc/supporter/base/rvvs89)
11:14:49 Join dfkt [0] (dfkt@unaffiliated/dfkt)
11:22:37 Quit notlistening (Quit: Ex-Chat)
11:24:22 Join preglow [0] (
11:25:58bluefoxxjhMikeS: No, I don't have access to a linux box. Not without far, Far too much effort.
11:26:07 Quit yawny (Read error: Connection reset by peer)
11:26:22bluefoxxThinkpad is misbehaving and for some odd reason I can't change boot settings.
11:26:43bluefoxxClaims I've not entered my supervisor password, though one was never set :|
11:27:27 Join elcan [0] (
11:29:18 Quit preglow (Ping timeout: 268 seconds)
11:29:27pixelmathe fact that presets are discarded when exiting the radio screen was the problem why my simple fms hung for me in earlier versions of the patch until AlexP discovered that the general problem was a hang with no presets loaded but the fms wants to display them. My fms should just show them and I had presets loaded but not so on exiting the radio screen...
11:29:34pixelmaor were
11:31:49pixelmaS_a_i_n_t: there was a cabbiev2 graphics package out there (IIRC available in the wiki) with the original graphics. I think no SVG but bigger bitmaps (in resolution)
11:32:10bluefoxxHang on a minute...If I hook an external drive up to the scsi card in the laptop's dock, I can boot off of that.
11:32:36bluefoxxStill need to go hijack my friend's imbpass setup :\
11:33:18 Join preglow [0] (
11:36:33AlexPbluefoxx: It is still odd that the windows one doesn't seem to work
11:36:43AlexPI *think* bluebrother was the one playing with that
11:37:41 Join DerPapst [0] (
11:40:41 Join antil33t [0] (
11:40:45 Quit preglow (Ping timeout: 258 seconds)
11:41:11bluefoxxIs there any recovery modes that I might try? Something similar to the Sansa e200 series unbricking methods and whatnot?
11:42:13bluefoxxFor that matter, could the fact that I don't have stuff like WMP on my machine affect this at all?
11:43:24mtHow do I add /usr/include to the codec search path ? .. I tried adding it to CODECFLAGS, and included math.h, but it still didn't declare the expected prototypes. (i.e, not actually included)
11:43:42bluefoxxShould I attempt ?
11:43:49mts/included math.h/included math.h in a codec/
11:44:07CIA-5New commit by 03nls (r25993): Two tiny tweaks to some arm asm
11:44:16linuxstbmt: Rockbox has its own math.h - maybe that's being used instead.
11:44:41mtlinuxstb : Yeah that's probably why
11:45:06linuxstbmt: You could try explicitly include "/usr/include/math.h", as a test.
11:45:15mtyes that works
11:45:26mtthat's why I wanted to add it to the search path
11:46:09n1sbluefoxx: can your windows see the player when it's connected?
11:46:15mtI'll just included explicitly and get going, I'll have to get rid of it later anyway.
11:47:17bluefoxxn1s: Windows sees it, and offers to fire up winamp to manage the thing, but past that I don't really have access to it.
11:47:57bluefoxxWell, windows offered to until I told it to stop suggesting things to me every time I connected the thing ( I swear I turned the group policy for autorun *off*)
11:48:30 Join preglow [0] (
11:48:46n1sso your windows can talk to it over mtp, and beastpatcher should be able to too...
11:49:42***Saving seen data "./dancer.seen"
11:49:54funmanjhMikeS: which model did you use to test pitch_detector? (where you screamed in the mic)
11:49:57bluefoxxDoes beastpatcher rely on any services to be running?
11:50:35n1sbluefoxx: i don't know, but i imagine it would print an error if something it needs is missing
11:51:21bluefoxxThat's what I would figure, but I've had some odd programs before that say they're working fine when they're clearly not.
11:55:44S_a_i_n_tpixelma: Thanks for the tip on the GFX pack, I found it, thanks.
12:00:22pixelmaS_a_i_n_t: btw. I'm interested in improving cabbiev2 too, codewise and thinks like displaying volume in dB during adjusting (which means different amount of work depending on whether the port already uses viewports in a good way or not). I don't use cabbiev2 myself but think that if it is the default then it should be as good as possible - that's also the reason I made the Archoses port and improved the greyscale graphics
12:01:31 Join watto [0] (~watto@
12:01:38pixelmaalthough not using it myself means the motivation isn't too high but if some people could work together there that could help
12:04:01jhMikeSfunman: H120
12:04:47 Join JohannesSM64 [0] (
12:04:54 Quit JohannesSM64 (Client Quit)
12:05:27 Join JohannesSM64 [0] (
12:11:00jhMikeSsome of the calcs look dodgey
12:11:46bluefoxxAnyone else have a rockbox'd gigabeat s that they got going on windows than?
12:14:23jhMikeSI don't know the state of sendfirm for windows (we had that, right?)
12:14:28bluefoxxCan you perhaphs send me your copy of beastpatcher.exe ?
12:15:21S_a_i_n_tpixelma: I'm totally with you thoughts, and motivation.
12:15:22jhMikeSI had some homebrew deal using libmtp and some libusb drivers
12:15:47S_a_i_n_tI *want* it to be as good as it can be, but I don;t use it, and it is a lot of work for one person.
12:15:58S_a_i_n_tPerhaps we could share the load somewhat...
12:16:14Tornebluefoxx: what is the problem you are actually having?
12:16:47 Quit komputes (Ping timeout: 258 seconds)
12:16:52*jhMikeS never used beast patcher and had it work
12:17:04jhMikeSbut that was sooooo long ago
12:17:08bluefoxxTorne: Beastpatcher.exe runs, claims it worked, but the gigabeat doesn't go into mtp mode.
12:17:10pixelmaS_a_i_n_t: yes, that's why I told you. I'm more interested in the small screens and monochrome or greyscale
12:17:29S_a_i_n_twow, ok...glad of that ;)
12:17:42S_a_i_n_tI'm not so interested in mono/greyscale :D
12:17:46S_a_i_n_twe'll make a good team.
12:18:14Tornebluefoxx: you are following the instrictions from the manual install section of hte beast manual, yes?
12:18:15*jhMikeS thought it should patch and then the rb bootloader should show up on next boot
12:18:21S_a_i_n_tI can't start it right away, but I'll have a look at everything that will need to be done, make a TODO list, and get back to you tomorrow.
12:18:27Tornewhere did you get the patcher from?
12:18:34bluefoxx would be what I'm using, yes.
12:18:35pixelmaS_a_i_n_t: I thought you weren't :P
12:18:40bluefoxxThe windows section
12:18:50bluefoxxGot my patcher from the link in the artical
12:18:58Torneare you trying to install the dual boot bootloader, or the single boot one?
12:18:58pixelmaS_a_i_n_t: and sounds like a plan :)
12:18:59S_a_i_n_tpixelma: Interested in greyscale/mono?
12:19:05bluefoxxEither one at this point
12:19:07Tornesome beasts have huge issues running the single-boot loader
12:19:18S_a_i_n_tI'm kinda scared of it to be I don't really understand it yet ;)
12:19:33jhMikeSTorne: but then, it should be coming back wanting to restore it, correct?
12:19:44TornejhMikeS: not if it didn't accept it to begin with
12:19:50pixelmaS_a_i_n_t: "it"?
12:20:03 Join kugel [0] (~kugel@rockbox/developer/kugel)
12:20:04jhMikeSTorne: well, I expect that means nk.bin never got written if nothing changed.
12:20:18TornejhMikeS: yeah, it won't overwrite nk.bin unless it thinks the binary is valid, afaik
12:20:24 Quit preglow (Ping timeout: 276 seconds)
12:20:27kugelmt: I also faced a math.h problem yesterday
12:20:29S_a_i_n_tgreyscale/mono...I haven't learnt *anything* about them, so, to me they seem daunting.
12:20:34kugelI will hopefully fix it
12:20:39bluefoxxHolding the player up to my ear as I run the patcher, I clearly hear the drive spin up, clack away as I watch the progress bar run to 100 and that stop when it says complete
12:20:49bluefoxxBut reconnecting gains nothing.
12:20:52kugelthe problem on the sim is that we include *our* headers even if we call OS routines
12:20:59Tornebluefoxx: but then when you disconnect, it reboots to the beast firmware again?
12:21:06Torneor does it not even reboot?
12:21:22bluefoxxDoesn't even reboot
12:21:29kugeland our math.h is really a bad math.h. it doesn't declare any functions but only constants which are not declared in the standard math.h
12:21:34bluefoxxSimply reconnects to the machine in OF
12:21:37Torneright.. that suggests it didn't accept the firmware at all then
12:21:47Torneif it replaced nk.bin with *anything* it would reboot itself
12:21:51kugelmt: math.h was the reason for the wrong results?
12:22:18Tornebluefoxx: so, did you already try giving it nk.bin from the beast firmware update?
12:22:36JdGordonkugel: did you do the original AA code in playback.c?
12:22:49kugelno I only added multi aa
12:22:57bluefoxxI would try a linux install, but I've long forgotten the password for the half-broken install of 8.04 on my thinkpad, which is refusing to allow me to boot the dock's cdrom, and all my other linux boxen lack USB (lolSun)
12:23:17kugelmob-aa was Nico_P's work IIRC
12:23:18JdGordonkugel: does that mean you know about the struct dim which is stored for each one? or no?
12:23:38Tornebluefoxx: well, i'm not sure what you cna do then, i'm afraid :(
12:23:52Torneanyone else? :)
12:24:06bluefoxxHang on
12:24:08JdGordonis there any reason it isnt just used directly from the skin?
12:24:22JdGordonarg, bbs
12:24:25bluefoxxI might have an old Toshiba here that could potentially run linux.
12:24:26kugeldefine directly
12:24:34mtkugel: yeah it was .. I'm not including /usr/include/math.h instead to get the proper prototypes
12:24:39bluefoxxDon't know the condition of it, just that the backlight is had
12:24:55mtkugel ^
12:25:07kugelplayback stores the width&height to re-use slots which have the same width&height
12:25:28Tornebluefoxx: trying it on linux might indeed work, MTP is a huge pain :)
12:25:35pixelmaS_a_i_n_t: well, I don't know what there is to "understand". The big big plus to me is that they are readable without backlight (except in complete darkness) and I like the challenge to make graphics for them - my first work on Rockbox was doing graphics for plugins etc. for the Archos port... but discussing this further might be something for -community
12:26:24*bluefoxx sighs
12:26:29kugelmt: try renaming our math.h, it doesn't deserve the name anywa
12:26:31bluefoxxA fine right mess I'm creating of my room
12:27:12bluefoxxThree and a half laptops strewn about, a box of MP3 players, a pile of big filter caps, and the bag of power bricks/wallworts has just torn...
12:27:55pixelmakugel: I recently tried something with multi-AA that came up in the forums - displaying the same AA in different places using conditional viewports and it didn't work for me. Didn't investigate further though
12:29:07pixelmano, different spots in the WPS depending on some condition (like hold on or off etc.)
12:29:14kugelmy multi aa work didn't touch drawing, it just changed the buffering side to allow more than 1 aa per trac
12:30:12kugelpixelma: that's a different "multi aa" then
12:30:34pixelma%Cl is actually only there once, with the %C in in two viewports where only one is visible at a time (this works for me with bitmaps)
12:30:58pixelmaok, so it is not supposed to work?
12:31:03kugelno idea
12:32:16bluefoxxWhat's a nice lightweight, but usable live distro? The toshiba is only a 900mhz P3 with god only knows how much RAM
12:32:58funmanbluefoxx: linux is offtopic here but feel free to ask on #rockbox-community
12:33:12pixelmakugel: need to investigate further then
12:33:37bluefoxxfunman: 'Kay
12:38:40JdGordonkugel: yes, but from my reading it looks like they could come directly from the aa struct in the skin? or am i missing something?
12:38:57JdGordonnot really suggesting that change, just makign sure doing it that way in the radio wont breakk things
12:39:31kugelI thought playback shouldn't need to know about these structs
12:39:36JdGordonwidth and height are copied into a seoerate dim
12:39:46JdGordonplayback shouldnt know anything about aa
12:40:21JdGordonpixelma: I'm a bit surprised that didnt work
12:40:23kugelthere was some trouble with storing a pointer to skin objects I can't remember now
12:41:43pixelmaJdGordon: yeah, wanted to try again and maybe a few more possibilities (like two smaller viewports displayed at the same time etc.) but forgot about it :\
12:43:54pixelmawhile some people knowing about AA internals are around - is the AA size limited to the smallest dimension of the display? E.g. if I have s square image and display it on a 160x128 screen, the maximum size would be 128x128?
12:45:26kugelI think yes but I don't know for sure
12:45:42funman <- copy one channel into the other on sansa AMS (for microphone recording)
12:45:50 Join Jaykay [0] (
12:46:01funmani didn't test so i'm not sure if it's duplicating the correct channel or the silent one ;)
12:46:36S_a_i_n_tI would think that 160X160 would be possible. But displayed using the alignment tags to center it, so it would appear cropped.
12:46:43pixelmathe reason why I ask is this: in my c200 SBS I load the AA at the same size as on the WPS and let a viewport do the clipping and it works nicely. I thought of a WPS with fullscreen AA (but not actually having different AA images) for smaller displays using this prinviple
12:46:55S_a_i_n_tor aligned to a corner, etc.
12:47:26pixelmait didn't work, AA displayed looked only as big as 128x128
12:47:48S_a_i_n_tah, oh well...just a thought :/
12:47:59S_a_i_n_tat least now *I* know ;)
12:48:07S_a_i_n_tI'd always meant to try that...
12:48:50pixelmawell, maybe I did something wrong but AA was displayed so I don't know - which is why I asked here
12:49:07 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
12:49:54 Join adnyxo [0] (
12:50:52 Quit petur (Ping timeout: 246 seconds)
12:51:02S_a_i_n_tpixelma: I think you could do it with 160X161? IIRC (unless it was fixed since) I found out in a forum discussion with a user that the center align tgs for aa only seemed to crop the image properly if it was *not* square.
12:51:35S_a_i_n_tI never did any further testing on that though, it was a guy trying to make a widescreen effect.
12:51:48pixelmaI think I saw a version in the forums which used scaling tags but that's not what I wanted, I'd prefer cropped images over stretched ones
12:51:58S_a_i_n_tcropping the top/bottom of the AA only.
12:52:23S_a_i_n_tthat's the thing, it doesn;t stretch it, it actually does crop it.
12:52:53S_a_i_n_tbut, it only seems to do it if the image is not square, even by 1px.
12:53:03S_a_i_n_tthis was a while ago now though...
12:53:31pixelmadoes the AA loading actually do cropping? For some reason I thought it only does scaling
12:54:11pixelmain my SBS, I do "cropping" with the viewport
12:54:39S_a_i_n_tyeah, you can (or could?) do cropping if them image isn't square and you use the center alignment tags I'm pretty sure.
12:55:12S_a_i_n_tI don;t *quite* remember the recipie I used now, but the user was happy with the worked, but may have been a bit of a hack.
12:55:35S_a_i_n_ts/if them/if the/
12:59:12n1sany ideas why mounting a beast takes so much longer than other targets?
12:59:43pixelmaout of this night's discussion, Unhelpful said that loading a square album art once at 40x40 and 40x80 will result in the same size on the screen makes me believe that AA will be resized to the smallest dimension
13:00:01S_a_i_n_tpixelma: This may have been before the inclusion of %Cl even.
13:00:05 Join moos [0] (moos@rockbox/staff/moos)
13:00:13JdGordonnew radio art parch up to store more than one image in the buffer
13:00:20pixelmaeh? %Cl has always been there
13:01:19pixelmaat least in SVN album art
13:01:47 Quit kugel (Ping timeout: 252 seconds)
13:04:03 Join Schmogel [0] (
13:04:08 Quit stoffel (Ping timeout: 246 seconds)
13:06:11*n1s tries to "make install" on a sim and gets "Installing your build in your '' dir
13:06:11n1sERROR: No PREFIX given"
13:06:45n1sah, it's not a sim build, doh!
13:13:21n1s...and test_codec segfaults on the sim...
13:16:04 Join stoffel [0] (
13:25:01 Join wincent [0] (
13:25:26 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:25:38 Join wincent_balin [0] (
13:28:30 Quit S_a_i_n_t (Ping timeout: 264 seconds)
13:29:20 Quit wincent (Ping timeout: 240 seconds)
13:29:26 Quit JohannesSM64 (Quit: WeeChat 0.3.3-dev)
13:30:03 Quit wincent_balin (Ping timeout: 260 seconds)
13:30:35 Quit funman (Quit: free(random());)
13:34:01 Join S_a_i_n_t [0] (S_a_i_n_t@
13:36:09 Join tourist08 [0] (
13:36:58 Quit tourist08 (Client Quit)
13:41:43 Join dfkt_ [0] (dfkt@unaffiliated/dfkt)
13:43:04 Quit dfkt (Disconnected by services)
13:43:07 Nick dfkt_ is now known as dfkt (dfkt@unaffiliated/dfkt)
13:44:07 Join dfkt_ [0] (~dfkt@unaffiliated/dfkt)
13:44:54 Quit dfkt (Disconnected by services)
13:44:57 Nick dfkt_ is now known as dfkt (~dfkt@unaffiliated/dfkt)
13:45:18 Join ucchan [0] (
13:49:45***Saving seen data "./dancer.seen"
13:50:04ucchanFor TTA codec (FS #11256) Could you check the decoding speed because I send a new patch ?
13:50:11 Join JohannesSM64 [0] (
13:51:30ucchanplease test for non-ARM CPU players.
13:53:27n1succhan: i can try it in a little while
13:55:28 Quit bluefoxx (Ping timeout: 265 seconds)
13:56:02 Join bluefoxx [0] (
13:56:03ucchann1s, thanks.
13:58:20n1succhan: where can i find a file to test with?
13:59:41n1succhan: there is a bunch of hunks that fail when applying the patch, is it in sync with recent svn?
14:00:18n1sah, a git patch
14:00:45n1sapplied fine
14:01:14*n1s kicks flyspray for stripping the "a/" from beginning of the path, tricking me...
14:04:45ucchantest file:
14:04:46n1sah, there's a file in test_files
14:05:53 Quit moos (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
14:06:15amiconnucchan: As I already stated in the fs task, I think the codec should be committed as is. It will make it easier for others to work on it
14:06:25*n1s agrees
14:06:41S_a_i_n_tJdGordon: about?
14:06:58 Join anewuser [0] (anewuser@unaffiliated/anewuser)
14:08:20 Join bluefoxx__ [0] (
14:08:25S_a_i_n_tI'm having some trouble with the progressbar logic and volume, if you could take a look at something in the sim for me I'd really appreciate it.
14:08:56ucchanamiconn, n1s: I will prepare the commit to TTA codec.
14:09:35amiconnThere is nothing that demands that a codec needs to be realtime before it is committed. In fact most codecs weren't (and some still aren't, at least for some compression levels)
14:09:56amiconnSpeaking about compression levels - does tta only have a single level?
14:10:29kugeldoes anyone else experience a huge delay with firefox when reading today's log?
14:10:44kugelit appears the html5_enable setting causes that
14:11:04 Quit bluefoxx (Ping timeout: 252 seconds)
14:11:38ucchanamiconn: TTA is single level.
14:12:44pixelmaamiconn: I would like if viewports on monochrome screens could use "inverted colours" just like you can use foreground/background shade or colours on greyscale and colour screens
14:13:29 Quit bluefoxx__ (Ping timeout: 276 seconds)
14:14:17n1succhan: 55.09% realtime on my h300
14:16:00n1succhan: have you benchmarked witht different compiler optimizations, i notice you use the pretty agressive "-O3 -funroll-loops" settings
14:24:16ucchann1s: oh! too slow
14:38:36 Quit bmbl (Ping timeout: 258 seconds)
14:40:13CIA-5New commit by 03uchida (r25994): add True Audio (TTA) codec ...
14:40:51ucchanI commit TTA codec now.
14:46:52n1sstripwax: for the logs could you review FS #11268 ?
14:47:37 Join halmi [0] (
14:48:53n1sI wonder why the beast requires more than 10Mz more to decode vorbis than c200
14:51:03 Quit parafin (Quit: So long and thanks for all the fish)
14:51:39n1smight the arm asm be to blame?
14:53:08ucchanremains to optimize more.... speed up more than 60% Is it possible? sigh...
14:54:12n1succhan: the emac unit makes a *big* difference on coldfire for multiplication heavy code, and iram makes a big difference for any hot data or hot code, so many possibilities remain
14:56:12 Join bluefoxx [0] (
14:56:33bluefoxxOh cool, both my machine has decided to boot, and my internet is back \o/
14:57:15bluefoxxSo, while my internet was conked out, I decided that I wouldn't be losing anything to boot a liveCD on my desktop here. Lo and behold, I got rockbox on the gigabeat.
14:57:57bluefoxx(I'm normally quite adverse to shutting down my desktop anymore than I have to. Habit of obsessively logging chatrooms and getting caught up in a billion little tasks)
14:58:14bluefoxxAnyhow, I rebooted it to get some settings to take effect
14:58:27bluefoxxAnd now it's whinging about 'error three'
14:58:45n1sit == the gigabeat?
14:58:46AlexPDid you install the single boot?
14:59:26bluefoxxAnd yes, it==gigabeat
14:59:48AlexPThe toshiba flash loader is *stupidly* picky about the state of the disk
15:00:04bluefoxxIt's currently sitting on my desk, glaring this error message at me.
15:00:15bluefoxxDo I turn it off? Force it off? Plug it in?
15:00:21bluefoxxNuke it and restart?
15:00:31n1sbluefoxx: is it rockbox' bootloader that displays the error?
15:00:40bluefoxxNo, the gigabeat stuff.
15:01:33bluefoxx(That giddy feeling I had when I found out the last idiot who paw'd the thing plugged the drive in backwards? Gone. The euphoria from getting rockbox on something other than *another* e200? Gone. D= )
15:01:53AlexPYou can send the bootloader again
15:02:04bluefoxxWill I wind up with another fancy hard drive based brick if I force a shutdown of it?
15:02:19AlexPThe error codes aren't fatal (usually), they unfortunatly just need a PC and wipe the hard drive
15:02:45bluefoxxRight. So I can safely reboot and nuke it than?
15:02:48AlexPbluefoxx: I *think* you should always be able to get it into its recovery thing (where I think it is now)
15:02:50AlexPWait a mo
15:03:04AlexPI need to check, it has been a long time since I went through this
15:03:39 Join parafin [0] (
15:03:51n1sit's weird, some people have troubles and for others it just works perfectly
15:03:58AlexPThere is some info here:
15:04:10AlexPn1s: Yeah, really odd - mine has never given me any issues
15:04:14 Join evilnick_B [0] (~0c140464@rockbox/staff/evilnick)
15:04:22n1smine has behaved too
15:05:24bluefoxxGive me a moment to switch over to my laptop for IRC - Gotta reboot the liveCD again...
15:07:24bluefoxxFor better future reference, which partition do I toss .rockbox in?
15:07:34bluefoxxThe main bit, or the firmware partition?
15:07:59AlexPThe main one
15:08:23n1syeah, leave the firmware partition alone
15:08:43AlexPOne of the big reasons it has a benny is doing anything at all to the firmware partition
15:09:11bluefoxxRight. That might be why it's borked now. I wasn't sure, and my 'net was still down, so I tossed it in both.
15:09:15TorneIndeed. Don't touch that partition
15:09:16bluefoxxShort term it worked \o/
15:09:29Torneand use a dual-boot bootloader, even if you don't want to dual boot, unless you really can't for some reason
15:09:51AlexPOr the single boot happens to work and you want the much quicker start up time :)
15:10:58 Join joeyg [0] (
15:11:44 Quit kugel (Read error: Connection reset by peer)
15:12:44 Quit joeyg (Client Quit)
15:12:53pixelmaamiconn: or an "inverse" parameter for the viewports in general could be a nice thing to have too. This would allow working with the users colour setting too
15:13:58bluefoxxYeah, windows beastpatcher dun wanna work still
15:14:35AlexPA good reason to keep it in the unstable catagory it seems
15:15:09bluefoxxI'm sure that most other people would have been lost, or frustrated into giving up by now :b
15:16:39 Nick bluefoxx is now known as bluefoxx| (
15:16:45AlexPTrouble is, most people here don't use Windows :)
15:17:02bluefoxx|That too
15:17:12bluefoxx|I do too much photowork to give up CS4 and DPP
15:17:17AlexPand Windows USB MTP stuff is complicated (so I'm lead to believe)
15:17:27 Join bluefoxx [0] (
15:17:50bluefoxx|I've also become rather adept at beating windows into shape within a couple hours, and keeping it there.
15:18:36bluefoxxBecame rather frustrated with spending days on end simply trying to get something in *nix to work only to find it broken after an update or after something else was installed.
15:18:53AlexPWe are getting off-topic :)
15:18:56bluefoxx(There was little escape from dependancy hell)
15:19:18bluefoxxJust waiting on my desktop to shut down now :\
15:20:04JdGordonS_a_i_n_t: am now
15:21:26bluefoxxUgh. 1600x1200 across a single of my CRTs, iintended to do 1920x1200. Not even at at least 70hz either :(
15:21:59AlexPbluefoxx: Stay on topic please chap :)
15:22:17bluefoxxSorry. Have a bad habit of rambling to myself :\
15:22:38bluefoxxSo chmod +x ./beastpatcher64 > ./beastpatcher64 -d ./NK.bin and don't copy .rockbox to the smaller partition
15:23:13AlexPHopefully, yes - you might have to change partition types for linux
15:23:39AlexPah, no - beastpatcher should do that
15:24:31gevaertsThe USB code live-fixes it IIRC
15:24:41AlexPyeah, I think so
15:25:18bluefoxxCommands run, killed rhythmbox with a vengance, copying .rockbox to the 27Gb partition...
15:25:36bluefoxxAppears to be booting...
15:25:59CIA-5New commit by 03uchida (r25995): add the true audio codec.
15:26:06n1si wonder if we will ever se areliable installer on windows for the beast...
15:26:58S_a_i_n_tJdGordon: If/when you get the change, could you run through a Nano sim for me (1G or 2G shouldn;t matter) and see what's wrong with the volume progressbar? I mainly want you to see it in the sim as it's easier to see what it's doing then describe it.
15:27:01S_a_i_n_tI'm wanting to do "%?pv<|%pv-0dB|%xdStatic image for 0dB|%pv+0dB>", the code in the wps seems right, and it used to work (when I first tested it), but it doesn't seem to want to now :[
15:27:35bluefoxxI have to say, the screen on the gigabeat is *much* more usable than on the e200
15:27:38S_a_i_n_ts/get a change/get a chance/ lol
15:27:52AlexPbluefoxx: Yeah, it is quite a decent screen
15:28:10 Join bluefoxx|dtop [0] (
15:28:34bluefoxx|dtopI'm going to be *very* tempted to buy a larger drive for this thing...
15:28:48bluefoxx|dtopHas anybody put a larger zif drive into a gigabeat?
15:29:27AlexPyeah, I have a 120 gb in mine
15:29:39AlexPBut that is dual layer (was originally a S60)
15:30:00AlexPThe toshiba bootloader is rubbish though - it won't go above 137 GB
15:30:26AlexPbut up to 120 is fine
15:30:32AlexPIf you can find a single platter one
15:30:51bluefoxx|dtopYou have to do any magic voodoo to get the new drive to work, or just clone the device and than DD the new drive
15:31:10AlexPThere is something IIRC
15:32:28S_a_i_n_tJdGordon: I think the effect I want should be obvious looking at my code. Why it doesn't work now is a mystery...first time I tested this was before it was fixed to cope with the bug pixelma found, it (what I'm trying to do now) worked when I first tested it out.
15:32:42AlexPbluefoxx|dtop: You should be fine actually - stick the blank disk in, and it'll go into recovery mode - send it the firmware and it'll partition and stuff for you
15:33:29bluefoxx|dtopGoing to have to save a few bucks to contribute to my friend upgrading his U820's hard drive - That thing's rocking a 120 :3
15:35:50 Join robin0800 [0] (
15:36:02TorneI guess if we can't work out why the windows one doesn't work, we could always make a bootable linux that does it :)
15:36:17bluefoxx|dtopSettings layout is about what I'm familliar with :) Couple new things, some stuff missing that's probably platform specific....Anyone able to tell me what "Caption Backlight" is?
15:36:24TorneThe manual is, yes
15:36:36AlexPbluefoxx|dtop: The manual should have all that
15:36:49bluefoxx|dtopBoth my e200 and the gigabeat have it, but I can't really see it affecting anything :\
15:37:05bluefoxx|dtopOr I can spelunk through :>
15:43:28markunbluefoxx|dtop: I think the backlight turns on on track change
15:43:36 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
15:44:26bluefoxx|dtopAh. Interesting...
15:46:21bluefoxxrebooting the gigabeat...Here's to hoping it doesn't error3 on me again
15:47:56bluefoxxSo anybody got a clue what's up here? Because I've not touched the main f/w partition this time
15:48:22bluefoxxAnd yet I've wound up with another error three on rebooting the dratted thing :(
15:49:06gevaertsbluefoxx: welcome to the Gigabeat S!
15:49:14gevaertsOnce it works, it works well :)
15:49:47***Saving seen data "./dancer.seen"
15:50:10Tornebluefoxx: we don't know why it does this
15:50:16Tornethere are several things that seem to make it more or less common
15:50:28Tornebut ultimately it does it a lot for some peple and never for others
15:52:13 Join gerbil [0] (
15:55:11 Quit antil33t (Read error: Connection reset by peer)
15:55:18 Join antil33t [0] (
15:55:28ucchann1s: I confirm FS #11268. iPod video: 128 kbits: +1.8 %, 256 kbits: +1.0%
15:57:01ucchanI'm very tired and fungry. Then I will quit. very thanks amiconn and n1s.
15:57:12Tornebluefoxx: i've painfully picked over some of hte beast's bootcode to investigate several options but nothing has come up so far
15:57:13 Quit ucchan (Quit: Leaving...)
15:57:14n1succhan: are you comparing to a recent, svn? i committed some other optimizations yesterday but FS #11268 shouldn't change anything for ipod video
15:57:16gevaertsbluefoxx: is this a single-boot or dual-boot install?
15:58:18 Join joeyg [0] (
15:58:20bluefoxxDual is what I've been attempting
15:59:17bluefoxxWould I be able to run the *nix beastpatcher under cygwin or a VM installed under windows?
16:00:20Torneunder a VM, yes
16:00:22Tornenot in cygwin
16:01:12gerbilHi, during the work for my Keybox patch I found a small bug in Keybox that happens when you reset a data base and, without leaving the plugin, create a new data base. In function "reset" the number of items in kb_list has to be resetted too. I included this into last version of my patch. Now I ask myself: should I wait if the patch gets applied or should I make an additional bug report?
16:02:18bluefoxx|dtopWhat of ?
16:02:38bluefoxx|dtop(Trying to find options that don't have me rebooting so frequently - kind of annoying :\ )
16:03:14n1sgerbil: we prefer such changes in seperate patches
16:04:19gerbilas a patch or bug report with attached patch?
16:05:29linuxstbgerbil: I think as a patch - it is more likely to be committed that way.
16:05:34bluefoxx|dtopCan't belive I'm about to install *shudder* windows media player...Just to see if it might make any kind of difference in this...
16:05:56linuxstbI thought the Windows version of beastpatcher required WMP?
16:06:10n1sit does?
16:06:12linuxstbOr at least, the Windows MTP library that comes with WMP...
16:06:39*linuxstb looks towards bluebrother
16:06:49 Quit Jaykay (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
16:06:54gerbillinuxstb: thanks, will do it.
16:08:01CIA-5New commit by 03ranma (r25996): Correct register defines. I had to look at the disassembly to spot this bug. orz
16:08:07n1slinuxstb: it would be nice if it displayed some error if that is missing then
16:09:10linuxstbn1s: I don't know the details, I just recall some talk here about beastpatcher and different versions of WMP...
16:09:13*gevaerts wonders if some of those new gcc 4.5 features such as link-time optimisation might provide binsize gains
16:09:32n1sgevaerts: i think they have the potential to
16:10:04n1si want to try it at some point
16:10:17n1sor would like to se the results if someone else did ;)
16:11:00 Quit bmbl (Quit: Bye!)
16:11:10bluefoxx|dtopI've never liked having windows media player on my machines. Ever. I let it get installed when I initially build a box, and I'll toss the updates for it on, but I always remove it from the components page...
16:13:36bluefoxx|dtopOh that's nice.
16:14:45bluefoxx|dtop"The software you are installing has not passed Windows Logo testing yadda yadda will not be installed. Contact your systems administrator" Well for one, I /am/ my administrator. More annoyingly however, that's supposed to be the XP version I just downloaded.
16:15:06bluefoxx|dtopThanks a lot microsoft. Top quality software as usual :|
16:15:47bluefoxx|dtopSo much for the idea of installing WMP to try and get the gigabeat to work.
16:20:21 Quit gerbil (Quit: CGI:IRC (EOF))
16:26:19JdGordonS_a_i_n_t: a full theme .cfg would be nicer
16:27:01S_a_i_n_tblargh...did I only send the wps?
16:27:06JdGordonthe issue is the jumpiness at the ~0Db area?
16:27:40JdGordonfile a bug
16:27:46JdGordonlow priority :)
16:27:52S_a_i_n_tand where it jumps from *some value* to 0dB
16:28:02 Join komputes [0] (~komputes@ubuntu/member/komputes)
16:28:26S_a_i_n_tthe 0+dB area doesn't scroll at all, it just gets displayed.
16:29:08S_a_i_n_tand the 0-dB area stops scrolling before the limits its defined it.
16:29:14S_a_i_n_tit's pretty weird now.
16:29:23S_a_i_n_tit *did* work, once.
16:29:38JdGordonthats nice :p file a bug
16:29:52 Quit stoffel (Remote host closed the connection)
16:31:46 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
16:34:46 Join Boldfilter [0] (
16:35:06JdGordonradio art is pretty much finished
16:35:14JdGordonneeds testers on target
16:36:10mc2739JdGordon: tested latest patch - looks good
16:36:10 Quit bluefoxx|dtop (Read error: Connection reset by peer)
16:36:16 Join bluefoxx|dtop [0] (
16:36:28JdGordonno more delay swapping between stations?
16:36:36mc2739JdGordon: nope
16:37:04mc2739is conditional support something to be added or a bug?
16:37:05JdGordonsweet. I'm just fiishing the change to dump out an old image if there isnt enough room
16:37:16JdGordon%?C doesnt work?
16:37:47mc2739unless I coded it wrong
16:38:30JdGordonI forgot to make it work :p
16:38:32JdGordonadding now
16:38:40mc2739hehe - ok
16:43:40JdGordonhave we got a file_exists() or similar?
16:52:57JdGordonwould help if I uploaded the right patch :p
16:53:08JdGordonthis one is pretty much done, test it and leave comments :)
16:54:02n1sJdGordon: we do have file_exists()
16:54:13*n1s hands JdGordon grep
16:54:18JdGordonyep, too slow! search beat you to it
16:54:24 Join halmi_ [0] (
16:54:43n1si was busy with the buns!
16:55:10 Join BeFalou [0] (~mamutoi@unaffiliated/befalou)
16:56:26 Join toffe82 [0] (~chatzilla@
16:56:52 Quit halmi (Ping timeout: 246 seconds)
16:59:04 Join halmi [0] (
16:59:06 Quit halmi_ (Ping timeout: 265 seconds)
17:11:11 Quit JohannesSM64 (Ping timeout: 265 seconds)
17:14:46 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
17:15:16 Join stoffel [0] (
17:17:12 Join funman [0] (~fun@rockbox/developer/funman)
17:17:12 Join webguest42 [0] (
17:19:39 Quit webguest42 (Client Quit)
17:19:53 Quit robin0800 (Remote host closed the connection)
17:21:44 Join JohannesSM64 [0] (
17:28:35 Join nograsswillgrow [0] (
17:29:07 Quit nograsswillgrow (Client Quit)
17:33:09 Quit bluefoxx|dtop (Ping timeout: 240 seconds)
17:44:36 Join fyrestorm [0] (
17:49:51***Saving seen data "./dancer.seen"
17:52:58 Quit bertrik (Ping timeout: 240 seconds)
17:53:30 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
17:55:35 Join Jaykay [0] (
17:59:50 Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude)
18:02:05Blue_DudeSo far I've been pretty slavishly following directions about setting up build environments. I'm currently using cygwin, and it works, but building is slow. I'm new to cross-compiling, Linux, etc, so this may be a naive question.
18:02:28Blue_DudeBut has anyone considered using something besides Cygwin on Windows boxes? I mean besides virtual machines and the like. What about MinGW?
18:06:33TorneMinGW is not what you think it is
18:06:38Tornethe MSYS environment.. that's cygwin
18:07:01TorneMinGW is an altenative to using MS's tools to build win32 binaries
18:07:07Torneit is not a way to run POSIX code (like gcc) on Windows
18:07:33Tornethe msys tools are all linked against cygwin1.dll same as everything else :)
18:08:07funmani heard of interix but i guess all in all the fastest & easiest is to use a VM
18:08:23Torneinterix is slow as well
18:08:59 Join halmi_ [0] (
18:09:11Torneinterix is just someone else's go at cygwin :)
18:10:23 Join Schmo [0] (
18:11:26 Join Xerion [0] (
18:12:01 Quit halmi (Ping timeout: 276 seconds)
18:12:24bluefoxxOk. I found what was barfing up the gigabeat on me
18:12:31 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-nycwzpjodpvbjoxi)
18:12:47saratogaBlue_Dude: why not just use a VM?
18:12:52saratogathey're very fast and quite easy
18:13:16Blue_Dudesaratoga: no admin rights. :(
18:13:22bluefoxxHad to properly dismount the data partition of it, but not let ubuntu touch the f/w partition with it's unmount.
18:13:25saratogaIMO unless you're extremely low on disk and ram, a VM is the best way to do things from Windows
18:13:32 Quit Schmogel (Ping timeout: 265 seconds)
18:13:38Blue_DudeTorne: oh.
18:13:38bluefoxxCue one tired geek who's now more of a music nerd :D
18:13:50TorneBlue_Dude: MinGW's gcc and cygwin's gcc are the same gcc, in fact
18:13:59saratogaBlue_Dude: theres no way to install a VM without admin?
18:14:05TorneBlue_Dude: you can use -mno-cygwin to compile with mingw headers/libraries instead of cygwin ones, from cygwin
18:14:13saratogai'd think without the advanced network features VB might work
18:14:48Blue_DudeTorne: I just want to cross-compile without having to take a lunch break.
18:14:51TorneBlue_Dude: but yeah, mingw is about using a unixlike OSS environment (based on cygwin) to compile normal win32 binaries
18:15:05Torneas such it is just as slow as cygwin
18:15:27Blue_DudeTorne: well, that's why I asked. I really dont know.
18:15:43Tornemingw is a target, not an environment, and our target is always arm-elf or something
18:15:51Blue_Dudesaratoga: I think you need admin rights just to install VB.
18:15:53bluefoxxOh man the gigabeat is a sick little player.
18:16:19*Blue_Dude is mightily confused about the whole thing, but is trying to keep up.
18:16:26 Quit Xerion (Ping timeout: 260 seconds)
18:16:46bluefoxxI'm going to faintly miss the scrollwheel for flying through long lists of music, but I think that if I use the gigabeat for fewer artists of whom I have more music from, I'll be good.
18:16:55TorneBlue_Dude: the msys/mingw people really don't explain this sensibly at all
18:17:03Torneso it's normal ot be confused :)
18:17:17Blue_DudeI just type in the magic words (make bin, etc.) and it works. I don't know how it works though.
18:17:26bluefoxxThanks to everyone who helped me, and tolerated my confusion. And of course the awesome people who got rockbox going on the gigabeat.
18:18:13Blue_DudeTorne: it took me a day and a half just to get the right packages for mingw/msys. There are all these options and nobody lists the essentials.
18:18:28TorneBlue_Dude: msys is useless, ignore it
18:18:39Torneif you want to compile windows binaries, cygwin can install a mingw toolchain
18:18:51Torneyou can switch with just -mcygwin -mno-cygwin on the gcc command line :)
18:24:37 Quit liar (Quit: Verlassend)
18:24:44saratogalooking on line it does look like all the fast VMs require a kernel mode driver of some kind
18:25:36Blue_DudeI'm getting the manual for VirtualBox right now. I'm going to curl up on the couch and get reading.
18:25:42Tornesaratoga: yup
18:25:53Blue_DudeI'll figure out a way to bribe an IT type for an install.
18:26:17 Quit S_a_i_n_t (Ping timeout: 258 seconds)
18:27:59 Join S_a_i_n_t [0] (S_a_i_n_t@
18:30:03 Quit parafin (*.net *.split)
18:32:29 Join Xerion [0] (
18:34:37 Join parafin [0] (
18:36:54 Quit komputes (Ping timeout: 248 seconds)
18:39:28 Quit Schmo (Quit: Miranda IM! Smaller, Faster, Easier.
18:44:35 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
18:46:24 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
18:46:44 Quit stoffel (Read error: Connection reset by peer)
18:49:17 Join komputes [0] (~komputes@ubuntu/member/komputes)
18:49:29 Quit funman (Quit: free(random());)
18:50:06 Join Schmogel [0] (
18:54:34 Quit bertrik (Remote host closed the connection)
19:02:00 Join mikroflops [0] (
19:05:48AlexPbluefoxx: So you had to unmount the data partition, but just unplug and not touch the firmware one?
19:06:22 Quit mikroflops_ (Ping timeout: 260 seconds)
19:10:36bluefoxxAlexP, More or less
19:11:28bluefoxxMight be different with a newer release, I simply grabbed an old 7.10 64bit edition disc that's been sitting in a drawer
19:11:58bluefoxxBut rockbox is very happily chugging along now
19:12:17bluefoxxIt's considerably snappier than the e200 too :D
19:16:22 Quit TheSeven (Ping timeout: 276 seconds)
19:16:51 Quit JohannesSM64 (Ping timeout: 252 seconds)
19:19:48 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
19:20:00 Join S_a_i_n_t_ [0] (S_a_i_n_t@
19:20:21 Quit S_a_i_n_t (Ping timeout: 240 seconds)
19:24:30 Join S_a_i_n_t [0] (S_a_i_n_t@
19:24:58 Quit S_a_i_n_t_ (Ping timeout: 276 seconds)
19:25:34CIA-5New commit by 03alex (r25997): Correct English.
19:28:29 Quit komputes (Ping timeout: 264 seconds)
19:29:02 Quit saratoga (Quit: Page closed)
19:30:27 Quit ender` (Quit: The word 'politics' is derived from the word 'poly', meaning 'many', and the word 'ticks', meaning 'blood sucking parasites'.-- Larry Hardiman)
19:32:34 Join JohannesSM64 [0] (
19:34:44 Join funman [0] (~fun@rockbox/developer/funman)
19:35:00 Join ender` [0] (
19:35:23funmanincreasing pcm dma burst size doesn't seem to increase battery life :/
19:37:25funmanranma: i think it's interesting that rockbox needs 10mA more for playback while OF needs only 2, compared to idle
19:38:09funmani think since our codecs are very efficient the difference might be in pcm (we ruled out as3514 settings already i think)
19:41:02 Quit CGL (Quit: Soy spammero ---- \m/ d(>.<)b \m/)
19:41:08 Join notlistening [0] (
19:41:46ranmafunman: Hey, I just got my first usb 'ctrl in' interrupt :)
19:41:55funmannice :)
19:42:00ranmaAnd dma to the buffer seems to have worked
19:42:15funmanis it consistent or just happens one every X plug-in ?
19:42:31notlisteningHeya if I am using a dated bootloader with a clip+ has there been a recent release of a new one that i should use?
19:42:34ranmaI expect it to be consistent.
19:43:09ranmaI found a stupid bug in usb-drv-as3525.c, all register defines were wrong (missing braces)
19:43:24notlisteningfunman will know with his finger on the pulse of everything Sansa
19:44:19funmanranma: braces?
19:44:26notlisteningCheers funman
19:44:47ranma"(volatile unsigned long*) USB_BASE + 0x2000" VS "(volatile unsigned long*) (USB_BASE + 0x2000)"
19:45:59funmani wonder what i have done to the gods for this code to have any effect
19:46:22funmanthx for spotting that
19:46:34ranmaI spotted that in the disassembly... ^^;
19:46:37FlynDicefunman: just pasing thru now but saw your pcm comment above. Will using test & clean( clean_dcache() ) instead of clean_dcache_range() help us there?
19:46:41funmanbtw, braces are {}, i think () are parenthesis
19:46:47 Quit ender` (Quit: Remember: A secretary isn't permanent until she's been screwed on the desk...)
19:46:48ranmaAh, right.
19:47:01funmanFlynDice: i think test & clean can't use ranges
19:47:26*ranma wanted to say brackets
19:47:46FlynDicethat's the point you don't need to give it a range, it figures it out on it'd own...
19:47:55FlynDiceon it's..
19:48:18ranmaAnyway, I think I'll go sleep now and continue tomorrow :) (2:48 am now)
19:48:19funmanFlynDice: but we could clean some entries which we don't need to, so it would be slower?
19:49:09 Quit funman (Quit: free(random());)
19:49:19FlynDiceI see, well I'll go finish mowing....
19:49:52***Saving seen data "./dancer.seen"
19:50:36 Join ender` [0] (
19:51:34notlisteningThe updated bootloader .sansa file i just downloaded replaces the rockbox.sansa file in my build dir right?
19:57:16 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-aurujinaicbuddjm)
19:57:27saratogano, you have to install it
19:57:43saratogafollow the directions on the wiki
19:57:48notlisteningwill do
20:00:53saratogain the TTA decoder, I wonder if we could just put the whole tta[MAX_NCH] array in IRAM
20:00:56saratogaits not that big
20:01:53saratogaanyone want to try on Coldfire?
20:01:56saratogai can make a build
20:04:52*gevaerts checks if his X5 is operational today
20:07:14gevaertssaratoga: I can test. Is there a tta file somewhere?
20:08:37 Join arbingordon [0] (~w@unaffiliated/arbingordon)
20:09:13gevaertsOK. Do you have a patch or a build ready?
20:10:04saratogasvn on the H300 gets 55.09%
20:10:10saratogaso lets see if thats any faster
20:11:52gevaertsFile too large :\
20:12:31saratogahow much RAM does the X5 have?
20:12:47saratogawell i guess 16MB
20:13:11saratogai really need to find time to fix that stupid test_codec problem
20:13:26saratogacan you truncate the tta file at 13MB or so?
20:13:30gevaertsI'll see if my h300 is charged enough
20:14:09gevaerts(or truncate it, if that doesn't break things)
20:14:50saratogatruncating probably works but i've never tried
20:15:33gevaertsWe'll know soon
20:17:43 Quit flydutch (Quit: /* empty */)
20:23:52gevaertssaratoga: 120.75%, 102.84MHz on h300
20:24:06saratogathats like a 2x difference
20:24:13saratogafor 8KB of IRAM
20:24:17saratogakind of incredible
20:24:27saratogacan you double check that against stock SVN?
20:27:25 Part parafin ("If you don't like rock'n'roll | It's too late now")
20:27:43 Join parafin [0] (
20:27:45 Quit bieber (Read error: Connection reset by peer)
20:29:37 Join stripwax [0] (
20:35:34gevaertssaratoga: plain svn indeed does 55%. Your version also sounds correct, so there's no bug there
20:36:05 Join komputes [0] (~komputes@ubuntu/member/komputes)
20:36:24saratogagevaerts: 218MHz?
20:36:29saratoga(for the commit log)
20:36:36gevaertsyes, I think so
20:36:41 Join gbl08ma [0] (
20:36:59saratogasounds good
20:37:07CIA-5New commit by 03saratoga (r25998): Put decoder array into IRAM. Improves Coldfire (h300) performance 218MHz ->102.84 MHz. Should be realtime on all CF targets.
20:38:11 Join archivator [0] (
20:38:37gevaerts16, yes
20:40:10 Join Zagor_ [0] (
20:42:30 Part watto
20:42:41 Quit parafin (Quit: So long and thanks for all the fish)
20:42:44 Join parafin [0] (
20:44:12 Quit gbl08ma (Quit: CGI:IRC (EOF))
20:44:53 Quit Jaykay (Ping timeout: 248 seconds)
20:46:43 Quit antil33t (Read error: Connection reset by peer)
20:46:50 Join antil33t [0] (
20:49:05mc2739funman: for the logs - it looks like r25989 is causing problems on my e200v2 - tested with a clean build and a clean .rockbox directory - boot up and change any setting and shutdown - the shutdown splash hangs for 20 seconds and the config.cfg file does not get saved
20:49:08 Quit S_a_i_n_t (Ping timeout: 264 seconds)
20:50:35mc2739funman: also, connecting the usb cable while in rockbox causes it to hang, probably due to trying to save the settings before reboot
20:50:48 Quit joeyg (Quit: lions and tigers and bears, oh my!)
20:51:08mc2739funman: no problems with the r25988 build
20:52:07 Join S_a_i_n_t [0] (S_a_i_n_t@
20:52:55 Quit parafin (Quit: So long and thanks for all the fish)
20:53:31 Join parafin|away [0] (
20:54:24 Nick godeater is now known as GodEater (~godeater@rockbox/staff/GodEater)
20:55:09Zagor_pamaury: did you ever make the wps converter?
20:56:31 Quit parafin|away (Client Quit)
20:56:33 Join parafin|away [0] (
20:57:03 Quit parafin|away (Client Quit)
20:57:05 Join parafin|away [0] (
20:57:09 Join fml [0] (
20:57:54 Quit pamaury (Ping timeout: 252 seconds)
20:58:49 Quit S_a_i_n_t (Ping timeout: 260 seconds)
20:59:07fmlIs font loading broken somehow in the sim? I created a BDF file, converted it to .fnt and also to .c to verify the glyphs. In the .c all looks OK. But when I load the font in the sim (for sansa e200 v1), only garbage is displayed. Other fonts work, e.g. Helvetica 12.
20:59:11 Quit parafin|away (Client Quit)
20:59:12 Join parafin [0] (
20:59:48fmlWhat can be the reason?
20:59:58 Quit parafin (Client Quit)
21:00:15 Join parafin|away [0] (
21:00:16 Join S_a_i_n_t [0] (S_a_i_n_t@
21:00:18 Nick parafin|away is now known as parafin (
21:05:09 Join robin0800 [0] (
21:05:20 Join hebz0rl [0] (
21:05:35Zagor_fml: does it work on target?
21:06:14fml Zagor_ : I haven't checked yet. The fnt format doesn't depend on the platform, does it?
21:08:25fmlZagor_: No, it does not work on target
21:08:38Zagor_them I think we can conclude the sim is not broken :)
21:10:39fmlHrm... This is good and bad!
21:10:49 Join webguest21 [0] (
21:14:57 Quit webguest21 (Client Quit)
21:16:39 Join Lombax [0] (
21:19:25 Quit bluefoxx (Ping timeout: 268 seconds)
21:20:03 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)
21:21:17fmlHow can I build fonts via make?
21:21:41fmlI.e. what target should I call?
21:21:58saratogamake fonts i think?
21:22:06saratogathe make file can list its options
21:23:27Zagor_"make help"
21:24:48 Quit robin0800 (Remote host closed the connection)
21:28:02fmlWeird. If I convert the fonts via make, all is OK. But if I do it manually (via convbdf), it produces garbage
21:29:32gevaertsmake fonts V=1 might be next then
21:41:29 Quit stripwax (Quit:
21:46:55 Join petur [0] (~petur@rockbox/developer/petur)
21:49:55***Saving seen data "./dancer.seen"
22:01:15 Quit esperegu (Remote host closed the connection)
22:04:11 Join funman [0] (~fun@rockbox/developer/funman)
22:19:07 Join chrissavery [0] (
22:24:42chrissaveryI have created a patch to provide WPS - PictureFlow integration. I can now flip between the album flow screen and WPS screen and back with one button. This makes a very nice interface for album based selecting. I've add the patch to the tracker now and would like to hear if anyone tests it out and whether I did this the right way. First attempt at patch for rockbox. It adds a couple new settings that enable this functionality.
22:25:33 Quit JohannesSM64 (Quit: WeeChat 0.3.3-dev)
22:26:14 Quit amiconn (Ping timeout: 258 seconds)
22:26:14 Quit pixelma (Ping timeout: 258 seconds)
22:28:59 Join anewuser [0] (anewuser@unaffiliated/anewuser)
22:33:53 Join kugel [0] (~kugel@rockbox/developer/kugel)
22:34:54funmanpetur: I see you added line apps/recorder/recording.c line 1129 in r18202, do you remember why? it's not described in the commit log and i think it should be removed : the recording interface shouldn't be uninitialized before it's initialized
22:35:02 Quit bluebrother (Disconnected by services)
22:35:04 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
22:35:24peturjust a sec
22:35:50kugelZagor_: ping
22:36:02Zagor_kugel: yes?
22:36:16 Quit powell14ski_ (Ping timeout: 264 seconds)
22:36:38kugelZagor_: I think there's a problem with today's irc log and firefox 3.6 with html5_enabled set to true
22:36:46kugelthere's a big delay before the log is generated
22:36:49funmanfor as3525 we must make sure calls to pcm_init_recording()/pcm_close_recording() are ordered and called the same number of times (like boost/unboost)
22:37:48funmanZagor_: can you put on the download server?
22:37:51Zagor_kugel: ok
22:38:08kugeldisabling html5 support fixes it
22:38:32kugelisn't the server push thing part of html5?
22:38:47funmani think it's just a mozilla thing
22:39:04CIA-5New commit by 03funman (r25999): fix a typo in comment
22:39:16funmanuh oh, epic commit approaching!
22:40:41Zagor_no, the push thing is some very mozilla thing. not html5
22:40:54saratogaafaik theres no standard way to do that, which is why it won't work in chrome
22:40:59Zagor_*very old
22:41:37 Join pixelma [0] (quassel@rockbox/staff/pixelma)
22:42:21 Join amiconn [0] (quassel@rockbox/developer/amiconn)
22:42:35 Join mirak_ [0] (
22:42:39peturfunman: probably, since recording should be closed when leaving that function. Did you find an error with that? I mean, does it fail somehow?
22:43:43 Quit Lombax (Ping timeout: 264 seconds)
22:44:15funmanpetur: this line (to uninitialize) is at the start of the function just before the call to initialize it
22:44:48peturif you look further, you'll see more of those, so it is probably copy-paste ;)
22:44:52funmanit doesn't fail right now but it bother me a lot
22:44:59Zagor_funman: done
22:45:01petura lot?
22:45:15funmanZagor_: thanks
22:45:34peturfunman: I can show you code running on production systems that bothers me a lot and would probably kill you ;)
22:45:48funmanwell ok it bothers me less than that ;)
22:46:28 Join jd [0] (
22:46:29 Quit jd (Changing host)
22:46:29 Join jd [0] (~jd@Wikipedia/HellDragon)
22:46:31peturfeel free to test without that line, I'll try to remember to look at it if you don't
22:46:40kugelZagor_: right, but google says html5 is also going to have some server push feature
22:46:48 Quit bmbl (Quit: Bye!)
22:46:52 Quit Zarggg (Quit: Zarggg)
22:47:17 Join Zarggg [0] (~zarggg@2001:0:4137:9e74:0:829c:beb1:ba3d)
22:49:27Zagor_kugel: ok. the reader uses Content-type: multipart/mixed
22:50:19scorche|shevilnick_B: that char seems right, except mine doesnt get paid in money
22:51:08saratoga"Chromium supports multipart/x-mixed-replace"
22:51:24saratogais there a check for FF? maybe it will work on chrome now
22:52:03Zagor_replace is no good. it resends the complete page.
22:53:16fmlThere is a bug in the convbdf utility. If I convert a BDF file only with the -f option, the .fnt file is different than the one produced with both -f and -c options!
22:54:12 Quit anewuser (Ping timeout: 240 seconds)
22:55:19 Quit notlistening (Quit: Leaving)
22:59:28 Quit fyrestorm (Read error: Connection reset by peer)
23:03:12 Quit chrissavery (Ping timeout: 276 seconds)
23:05:20CIA-5New commit by 03funman (r26000): clipv2: current usage estimation ...
23:06:15 Quit Schmogel (Ping timeout: 245 seconds)
23:07:47funmanin fact i think i should rather insert my code for enabling/disabling dma in pcm_record_data/pcm_stop_recording functions
23:08:41 Join chrissavery [0] (
23:10:19 Part chrissavery
23:10:51funmanpetur: recording works fine with the line removed so i guess it should go
23:11:18peturdid you also go to its settings?
23:11:30funmanits settings?
23:11:34peturrec screen -> rec settings -> rec screen
23:11:44peturpress the menu button
23:12:12 Quit shai (Ping timeout: 265 seconds)
23:12:32funmanjust did that: fine
23:12:37*petur suddenly isn't sure if that actually leaves the function
23:12:56peturwell, it can go, I'll maybe test later
23:13:19funmanthen the call in misc.c is unconditional if recording has been started or not
23:15:14fmlOh! Generating .c file from .bdf seems to screw the internal data structure so that generated .fnt (which is generated after .c) is garbage. One should not specify both -c and -f options! Should we make them mutually exclusive?
23:16:18 Join fdinel [0] (
23:16:59 Join anewuser [0] (anewuser@unaffiliated/anewuser)
23:18:26 Join powell14ski [0] (
23:21:07 Quit Zagor_ (Quit: Leaving)
23:23:56 Quit ender` (Quit: But there, everything has its drawbacks, as the man said when his mother-in-law died, and they came down upon him for the funeral expenses.)
23:24:15 Join elinenbe [0] (
23:27:49CIA-5New commit by 03alle (r26001): Generating .c file make internal data structures unusable so that .fnt file can not be produced after that. Make sure that both options are not used ...
23:32:08CIA-5New commit by 03funman (r26002): fix pitch_detector: recording will stop itself if the callback returns <0 ...
23:32:13 Join jfc^3 [0] (
23:32:13CIA-5New commit by 03funman (r26003): as3525: retain & release DMA engine when starting and stopping recording ...
23:35:16funmann1s: jhMikeS: does r26002 fix the problems you had with pitch_detector?
23:35:41 Quit evilnick_B (Quit: Page closed)
23:35:56 Quit jfc^2 (Ping timeout: 240 seconds)
23:37:41 Join lpereira [0] (
23:41:45 Join bluebro [0] (~4d2c4024@gateway/web/freenode/x-aodzicesfrjvcqrf)
23:41:59 Quit bluebro (Changing host)
23:41:59 Join bluebro [0] (~4d2c4024@rockbox/developer/bluebrother)
23:43:34 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
23:44:13 Join anewuser [0] (anewuser@unaffiliated/anewuser)
23:46:13funmancan i see the current uptime somewhere? running time only tells the time ran since last charger plugin, not since last boot
23:49:59***Saving seen data "./dancer.seen"
23:52:48 Quit fml (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
23:53:20 Quit lpereira (Quit: Leaving.)
23:53:20 Quit xavieran (Quit: ➤➤➤➤➤➤ Unicode shall reign forever! ➤➤➤➤➤➤)
23:53:53 Join xavieran [0] (
23:54:08 Join liar [0] (
23:57:11 Quit DerPapst (Quit: Leaving.)

Previous day | Next day