Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2009-05-31

00:00:28 Part BryanJacobs ("Leaving.")
00:01:24CIA-38New commit by 03amiconn (r21138): Fix shutdown splash. It is meant to be shown on an empty screen.
00:04:32 Quit AndrewRB ("wheeeeeeeeeeeeeeeeeeeeeeeee!")
00:04:39MarcGuaySmarty error:
00:05:58kugelamiconn: can the fore/background pattern for greyscale be made so that accessing viewport.fg_pattern and .bg_pattern is the same as calling lcd_set_foreground and _background?
00:06:29 Quit AlexP (Read error: 60 (Operation timed out))
00:06:32amiconnThat's why these functions exist.
00:06:40 Join AlexP [0] (n=alex@rockbox/staff/AlexP)
00:08:01amiconnGreyscale foreground/background is an integer with a range of 0..(num_greylevels-1). The patterns are derived from that, however, the way they're derived differs between greyscale formats
00:08:49amiconnEven the number of bits is different. For horizontally and vertically packed greyscale, they're 8 bit, but for vertically interleaved greyscale they're 16 bit
00:11:10 Quit MarcGuay_ (Read error: 54 (Connection reset by peer))
00:11:30kugelthe patterns are an abstracted already for the different formats, so they wouldn't be the problem
00:11:56 Join BdN3504 [0] (n=55b2079f@gateway/web/cgi-irc/
00:12:22 Quit stoffel ("leaving")
00:12:27 Quit calman_ (Read error: 110 (Connection timed out))
00:13:17 Quit amiconn (Nick collision from services.)
00:13:20 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
00:13:39 Join moos [0] (i=mustapha@rockbox/staff/moos)
00:13:39 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
00:13:46amiconnfg_pattern and bg_pattern are an internal optimisation and are not meant for direct access from outside the lcd drawing code
00:14:02 Quit pixelma (Nick collision from services.)
00:14:04 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
00:14:22 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
00:15:15amiconnThey will stay correct if you copy them from one viewport to another, but you shouldn't make any assumption about the relation between the argument for lcd_set_XXXground() and the pattern value
00:15:15 Quit {phoenix} (Remote closed the connection)
00:15:22 Quit advcomp2019 ("Nice Scotty, now beam my clothes up too!")
00:15:31kugelyes, I noticed that
00:15:38kugelthat's why I'm asking if that can be changed
00:16:38 Quit ender` (" Today is the tomorrow you worried about yesterday.")
00:17:05BdN3504commit r21138 is womewhat funny, it displays the wps background file. i think this should be the backdrop.
00:17:18 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
00:17:30amiconnHmm, actually the pattern values in the viewport struct are equal to the argument, however, the driver needs to update its internal patterns accordingly
00:18:56*amiconn wonders what kugel is trying to do
00:19:19kugelnothing, I was just confused by the fact that setting the viewport members doesn't work
00:19:35kugelsince that works for mono and color, but not for greyscale
00:19:59Mikachuif i don't rb->close a file from a plugin, does it leak?
00:20:24 Quit Zarggg (Read error: 60 (Operation timed out))
00:20:56amiconnkugel: Imo colour should do the same, for speed reasons
00:21:28amiconnRight now each and every drawing function in lcd-16bit.c does the current_vp->XX_pattern lookup, over and over
00:22:02amiconncurrent_vp is in iram, but the next data fetch might even hit dram (depending on whether the viewport is in iram or dram)
00:22:22amiconnMono has no problem here because there are no "colours"
00:22:47 Quit BdN3504 ("CGI:IRC (EOF)")
00:23:58 Quit calman__ ()
00:24:07 Quit advcomp2019 ("I was raided by the FBI and all I got to keep was this lousy quit message!")
00:25:16 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
00:25:52amiconnHmm. Drawing functions are reading other viewport members as well. Performance tests would show whether caching them in iram variables makes sense.
00:26:37 Quit advcomp2019 (Client Quit)
00:27:14 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
00:28:06kugelsaratoga: going to wps turned out quite easy
00:28:38kugelI just needed to add a new return type to the enum and adapt the lists to catch that
00:30:36 Join Zarggg [0] (
00:34:59 Quit bluebrother ("Leaving")
00:36:25soapLlorean, outside the fact battery_bench is failing to write records on AC charge...going ok.
00:36:51soapI was out of town for two nights there, so there was a pause. Doing the USB charge with #8802 as we speak.
00:36:54 Quit intrados (Read error: 60 (Operation timed out))
00:37:32soapI'm less concerned about the AC situation, as can confirm stock is charging (at least when playback is inactive).
00:39:19 Quit Tristan (Remote closed the connection)
00:41:56 Quit advcomp2019 ("+++ OK ATH OK")
00:43:31 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
00:46:46 Quit pat_mulchrone (Read error: 104 (Connection reset by peer))
00:47:00kugelUnhelpful: ping
00:47:01 Join pat_mulchrone [0] (
00:47:55 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
00:52:34 Quit Beta2K (Read error: 60 (Operation timed out))
00:55:29 Join lee321987 [0] (
00:56:32lee321987Does lua script support RB "Copy" and "Paste" commands?
01:00:13 Quit n1s (Remote closed the connection)
01:05:08 Quit robin0800 (Remote closed the connection)
01:05:25 Join perfectdrug [0] (
01:06:15lee321987Does Maurus Cuelenaere ever hang out here? (I thin his nick starts like "MCu−−−−−−")
01:07:03perfectdrugpixelma: I just added the last SVG for the AMS sansas FS #10265 - sansa m200 :)
01:07:27 Quit bertrik (Read error: 60 (Operation timed out))
01:08:26Mikachulee321987: yes
01:09:00 Quit toffe82 (Remote closed the connection)
01:16:47 Quit lee321987 ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
01:35:49CIA-38New commit by 03amiconn (r21139): Two tiny optimisations for mono bitmap drawing on greyscale displays: (1) H1x0, M5: Manipulate destination masks directly for the aligned case - ~0.7% ...
01:38:20***Saving seen data "./dancer.seen"
01:42:29 Quit DarkDefender ("Leaving")
01:57:13 Join kperri [0] (i=4aa794a0@gateway/web/ajax/
02:02:55 Join Tristan [0] (
02:03:05 Quit Nico_P (Remote closed the connection)
02:06:44Unhelpfulkugel: yes?
02:08:13 Quit pat_mulchrone (Read error: 104 (Connection reset by peer))
02:09:56 Quit matsl_ (Read error: 110 (Connection timed out))
02:09:59kugelUnhelpful: I'm having problems with the keymap in pf
02:10:25Unhelpfulkugel: if it's about the button-map business you mentioned in the other other channel, i'd suggest that you use the tracklist view, which is the same as the core standard context except for the dedicated quit key.
02:10:39kugelI was trying to chain in CONTEXT_TREE, but it didn't work at all
02:11:05Unhelpfulno, those contexts are designed to chain to CONTEXT_STD. use TREE by itself, maybe?
02:11:23kugelas a third context?
02:12:08kugeland why should pf context > CONTEXT_TREE > CONTEXT_LIST > CONTEXT_STD not work (context list actually works, but that doesn't give the WPS button)
02:20:13 Join pat_mulchrone [0] (
02:20:24Unhelpfulbecause tree and list mask things in standard that i want, sometimes. also, the "core" list or tree context is sometimes really CONTEXT_LIST|CONTEXT_CUSTOM, etc - and that won't work with get_custom_action.
02:21:19Unhelpfulwhat exactly do you want that's not in PF's tracklist context? it has select, context, cancel, quit, and scroll...
02:23:34kugelUnhelpful: CONTEXT_TREE is in the context enum
02:23:50Unhelpfulamiconn: if i have a non-aligned int array on targets, i need to read it as characters and use shift/or to build whole ints, don't i? or if it's short-aligned, i can read half an int at a time...
02:23:56Unhelpfulkugel: ok?
02:24:16kugelgo to wps. That's what I want
02:24:57Unhelpfuldo you want that from anywhere in pictureflow?
02:25:29 Join KBH- [0] (
02:25:55kugelif possible, of course
02:26:41kugelit's a list button, so it will automagically work in the tracklist and album view I guess
02:27:19kugelI think I know what the problem is
02:27:41Unhelpfulthat's not a good assumption. tracklist is really just standard with the quit button added. album is tracklist with horizontal scroll added, and remapping for things that that masks.
02:28:24 Part perfectdrug
02:28:27Unhelpfulif you want to add a new button mapping for everywhere in PF, add it to pf_context_buttons
02:28:43Unhelpfulanything in there appears in tracklist and album list
02:30:25Unhelpfulif it's started from WPS, won't core go back there when it's quit?
02:30:55kugelthere's no way to go directly to the wps in SVN
02:32:06Unhelpfulthere's nothing like starting wps from PF in svn. :)
02:32:19UnhelpfulPF will need to clean up its thread, and exit, i'd think?
02:32:38kugelI added a PLUGIN_GOTO_WPS exit code
02:33:21Unhelpfulwell, one thing you could do is pass a parameter when starting it from WPS, so that it will "know" to return PLUGIN_GOTO_WPS on quit.
02:34:31kugelI'm not starting it from wps yet
02:35:32Unhelpfuli'm not really sure we need a dedicate "go to WPS" button, but if you're looking to add a button to both views in PF, pf_context_buttons is the place.
02:36:44kugelI'm not looking into adding the buttons manually
02:36:52kugelI would really like to use the standard core button for that
02:38:12 Quit KBH (Read error: 104 (Connection reset by peer))
02:38:13 Nick KBH- is now known as KBH (
02:38:20Unhelpfuli tried to use context_tree. the "proper" context_tree is completely incompatible with custom actions on some targets, and besides that it masks some things in context_std.
02:39:53 Join intrados [0] (
02:41:26Unhelpfulaside from that, PF's contexts were designed *by hand* to go on top of a specific core context. just changing that context stands a good chance of breaking the whole thing.
02:44:31kugelso, you think I'm better off adding them manually?
02:45:09 Join KBH- [0] (
02:45:12kugelTREE and LIST only add 4 buttons on top of STD, 2 of them are mostly rarely used combos
02:47:10Unhelpfulwhich target are you looking at? :)
02:48:31Unhelpfulthe CONTEX_CUSTOM thing is still going to be a problem, also.
02:49:11Unhelpfulif you chain to CONTEXT_TREE on gigabeat, core will chain to CONTEXT_CUSTOM|CONTEXT_TREE, which will get passed back to PF, even though it doesn't know what to do with that.
02:50:09Unhelpfuli suggested this back then, and ended up fixing it by using CONTEXT_STD as a base, but a "proper" soluting would be to replace CONTEXT_CUSTOM for "use the custom action function" with CONTEXT_PLUGIN
02:50:19Unhelpfulsince core also uses CONTEXT_CUSTOM
02:50:33 Quit KBH (Read error: 104 (Connection reset by peer))
02:50:45 Nick KBH- is now known as KBH (
02:56:38 Nick efyx_ is now known as efyx_zZz (
02:57:27kugelUnhelpful: Why does the beast that?
02:59:31 Join Thundercloud [0] (
02:59:54Unhelpfulcore selects from two tree contexts based on a setting, when passed CONTEXT_TREE. they're both based on CONTEXT_TREE|CONTEXT_CUSTOM.
03:00:43Unhelpfulipod has CONTEXT_TREE|CONTEXT_CUSTOM as well.
03:00:46kugelok, with CUSTOM_CONTEXT2, the thing works better, but it doesn't really chain
03:03:07kugelin the album view, it should go "album map (pf_contexts[0]) > normal map [pf_contexts[1]) > context_std", right?
03:04:02UnhelpfulCUSTOM_CONTEXT2 won't cause get_action_worker to call the context hook again. if you want to be able to chain to *any* core context, you need to create a context for that purpose that is not used for anything by core - like i said, CONTEXT_PLUGIN, and replace the CONTEXT_CUSTOM in PF and PLA with that as well.
03:04:34Unhelpfuland yes, that's how the PF contexts chain currently.
03:05:34Unhelpfuli still think adding a button manually might be easier. you'll need to think about how it would work on each target *anyway* :)
03:08:41kugelI see
03:11:21 Join shadoxx [0] (
03:12:43kugelUnhelpful: I guess we want pf to be able to go to the wps if it gets database integration
03:25:02 Join KBH- [0] (
03:26:37 Join toffe82 [0] (
03:27:35 Quit efyx_zZz (Remote closed the connection)
03:34:10Unhelpfulkugel: so that selecting a track plays the set and takes you to WPS, as it would from the DB?
03:35:05kugelUnhelpful: that could be done, but I wouldn't want that in the pf case
03:35:39kugelsince pf is purely eyecandy, one might want to stay longer there after selecting a file
03:36:00kugel(also since you can't get easily back to pf as you can from the database)
03:36:36kugelbut it shouldn't be more than 1 or 2 button presses away imo
03:37:55Unhelpfulwhat about a "quit to WPS" menu item? the menu is one press everywhere :)
03:38:24***Saving seen data "./dancer.seen"
03:45:17kugelUnhelpful: see the patch I just upload (before you mentioned it :) )
03:46:28Unhelpfulah. nice. :)
03:46:49Unhelpfuli think a menu item is fine, at least until somebody fixes CONTEXT_CUSTOM or adds keys for it.
03:47:41 Quit KBH (Read error: 110 (Connection timed out))
03:47:41 Nick KBH- is now known as KBH (
03:49:16kugelUnhelpful: would be nice if you look at the 3rd patch, I plan to commit it soon (without the tagtree and go to wps bits)
03:49:50kugelalso, we have to figure out how we get the core context menu attached
03:50:26Unhelpfuldid you change any of those CONFIG_ defines, or is that part just tab-nanny stuff?
03:51:07kugelthe third patch is the one above though :)
03:51:29Unhelpfulmapped to ACTION_NONE in PF doesn't mean "not used" - it usually means "masking a troublesome item from CONTEXT_STD"
03:51:45Unhelpfulsorry, i looked at the one that said "v3" ;P
03:52:17 Quit Thundercloud (Remote closed the connection)
03:52:44kugelah yea, the 2nd one wasn't really v2 worth :)
03:54:23kugelI hope the \n placeholder isn't too much of a hack :/
03:54:32Unhelpfulfn_offset can be calculated, by using strlen. track_names + tracks[track_index].name_idx + rb->strlen(&(track_names + tracks[track_index].name_idx))
03:54:57kugelfn_offset is calculated
03:54:57Unhelpfulor you could store the actual index for it and not have to sum with name_idx
03:55:02kugelit's the return val of snprintf
03:55:48Unhelpfulwhat's the purpose of the \n?
03:55:56kugelto make strcat work
03:56:03MikachuUnhelpful: are you sure about that &?
03:56:38UnhelpfulMikachu: you're right, the & doesn't belong... and that lets you drop a set of parens as well.
03:57:04Mikachuarguably &track_names[tracks[track_index].name_idx] is better
03:58:30kugelUnhelpful: strcat replaces the \0 in dest, hence a byte is lost
03:58:48kugela needed one
03:59:04Unhelpfulkugel: why use strcat at all? why not fetch the filename directly into the buffer?
03:59:29kugelI tried that, but that didn't really work
03:59:49Unhelpfulit's working fine for my local tree with album names.
04:01:55*kugel tries harder
04:02:52Unhelpfulafter you've got len, do the tag fetch with &names[string_index] as the target. then do strlen(&names[string_index]) to find out how long it was.
04:03:38Unhelpfulbtw, you should pre-check that MAX_PATH space is available, because tagcache_retrieve doesn't fail on EOM.
04:05:49Unhelpfulperhaps the "free buffer until we have X bytes" loop needs to be split out.
04:06:17kugelUnhelpful: that strlen after might get a bit trick then though
04:07:01kugelhm, I just need to substract actually
04:07:27kugelhm no
04:08:48kugelok, seems to work
04:09:18kugelUnhelpful: I thought passing avail to tagcache_retrieve was ok
04:10:51Unhelpfulkugel: it is, but if you look at tagcache_retrieve it uses the passed size as a length limit, but doesn't *fail* on hitting it - it can return true when you don't have the whole string.
04:11:22kugelnot nice
04:11:28kugelthat should be fixed
04:11:49Unhelpfulit returns false on failure to retrieve.
04:12:20kugelmaybe it thinks not having the full string isn't significant enough to fail at all
04:12:51kugelwe could diffentiate the return values with ints
04:13:52Unhelpfuleven so, the "free until we have X bytes" loop ought to be split out into a free_buffer() function so that we can use it to avoid failure due to lack of space on the filename fetch. you wouldn't want to free more buffer to load the title, but just fail if you can't fit the filename.
04:13:53kugelhow to handle avail to small anyway?
04:14:15Unhelpfulthat's what that loop is for :)
04:15:09Unhelpfulfree_slide_prio will free the lowest-priority slide. then buffer_out gets you the space that was freed, as a chunk cut from the front of buflib's buffer.
04:15:52kugelI didn't really look at that part of the function :)
04:16:32kugelUnhelpful: ok, the hunk looks like that now:
04:17:39kugelbtw, the colons should be dots, as per tagree imo
04:18:59 Join lude187 [0] (
04:19:05Unhelpfulum, the len>avail check is gone?
04:19:26Unhelpfulnevermind, pushed down :)
04:20:20 Quit lude187 (Client Quit)
04:20:23Unhelpfuli think that looks good. maybe retrieve should return length, -1 on fail?
04:20:25kugelfn_off should maybe substracted from avail :S
04:20:47kugelthat seems like a great idea
04:21:25Unhelpfulkugel: i don't think you need to subtract it from avail - you've added it to len.
04:21:35kugelhah, right
04:21:45kugelway too late here :(
04:22:04Unhelpfulmight be time to call it a night. it certainly is for me.
04:22:24kugelbut wait, it should, since it's passed to tagcache_retrieve
04:22:30Unhelpfulactually retrieve needs two error codes, -1 for fail, -2 for EOM.
04:23:07Unhelpfuljust use avail-fn_off, since you'll be subtracting len from avail later. or don't include fn_off in len.
04:24:20 Join __lifeless [0] (n=lifeless@
04:26:36kugelUnhelpful: will you change the error code soon (as in before I commit this or after)?
04:29:36Unhelpfulhrm, actually, couldn't retrieve just return needed len or -1 on error? caller can decide if len>avail is important.
04:30:45Unhelpfulthat would, aside from the -1, work "just like snprintf" :)
04:31:47Unhelpfuli can't bang on the error code issue right now. if you want to make it safe for now, do the free+retry on avail < fn_off +MAX_PATH
04:32:15kugel-1 includes the EOM case?
04:32:37kugelI'd be fine with -2 on EOM.
04:33:09Unhelpfulkugel: i'd prefer length-needed-on-EOM because then the caller *also* knows how much more space it needs. :)
04:33:10kugelone shouldn't ask for -1 at snprintf anyway, but < 0
04:34:55kugelhow would needed len look like?
04:37:01 Quit __lifeless (Read error: 54 (Connection reset by peer))
04:37:11 Join __lifeless [0] (n=lifeless@
04:38:50kugelUnhelpful: having it work like snprintf would work well. that makes EOM detectable
04:39:13 Quit _lifeless (Read error: 113 (No route to host))
04:39:42kugelalso, does tagcache_retrieve write the \0 byte?
04:43:17 Quit kugel ("exit(0);")
04:45:29 Join Andrew_ [0] (
04:45:55 Nick Andrew_ is now known as Guest57333 (
04:46:52advcomp2019Guest57333, hello.. you have question?
04:47:04Guest57333whats up dude
04:47:10Guest57333its AndrewJ from ABI
04:47:34 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
04:48:09 Nick Guest57333 is now known as AndrewJ (
04:48:33advcomp2019AndrewJ, look at the topic
04:49:23AndrewJadvcomp2019: sorry
04:52:52 Quit chandoo ("Leaving")
04:56:56 Join guest [0] (n=62d4a6c0@gateway/web/cgi-irc/
04:57:24 Nick guest is now known as Guest24253 (n=62d4a6c0@gateway/web/cgi-irc/
04:58:12Guest24253Is there anyone here that could point me to the proper link to the latest version of the source for Rockbox on the fuze?
04:58:36Guest24253I understand that it's not a proper port yet and unsupported, but I can't find any link to the current testing build.
04:58:41Guest24253Could anyone please help? Thanks.
05:01:27Guest24253Scorche, is this in response to my q? If so, could you please let me know where the file would be located. Sorry that I'm a little unsteady on my feet, but I'm a n00b at rockbox specifically.
05:02:28 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
05:03:15 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
05:03:50scorche|shGuest24253: asked where the source is, so i pointed you at the source...there is no "file" yet as it is not completed yet, as you mentioned
05:04:17scorche|shuntil it reaches that point, it is not really suitable for "n00bs" as you put it
05:06:26 Quit fyrestorm (Read error: 104 (Connection reset by peer))
05:11:24AndrewJDoes anyone here know how to get the Funny Languages ( to work? I have not the slightest idea how to
05:15:17Guest24253scorche, thanks for the input
05:15:32Guest24253however, all I meant was that I was unfamiliar with the rockbox website
05:15:53Guest24253I know my way around UNIX and Linux quite well, and I'm well-versed in several programming languages.
05:16:14Guest24253And I'm just a little put-off by your rudeness.
05:16:20 Join fyrestorm [0] (
05:23:40Guest24253Sorry to sound rude myself, but I came on here with a question and you immediately shut me down.
05:23:57Guest24253And I know people who have compiled source from the Fuze distro.
05:24:04Guest24253So I was just looking for a little help...
05:24:24 Quit shadoxx (Read error: 113 (No route to host))
05:25:11 Quit Guest24253 ("CGI:IRC (EOF)")
05:30:00 Quit RandomDestructn (Remote closed the connection)
05:34:09 Join shadoxx [0] (
05:36:11 Quit mirak ("Ex-Chat")
05:38:26***Saving seen data "./dancer.seen"
05:46:35 Join BabaChoowa [0] (
05:55:57 Join rethgir [0] (
06:10:54 Quit shadoxx (Read error: 60 (Operation timed out))
06:11:27rethgirI've recently registered to edit the rockbox wiki and would like to request edit privileges if possible. Thanks in advance!
06:12:10 Quit itcheg (" ajax IRC Client")
06:35:09 Join AndyIL [0] (i=AndyI@
06:36:57 Join AndrewRB [0] (
06:38:29AndrewRBcould anybody help me get my desired functionality with the WPS Keymap for the D2?
06:38:40AndrewRBhave a look at these two lines:
06:39:32AndrewRBwhat I want is to be able to press the holdmenu button once to "play/pause", or hold it briefly to skip to the next track
06:39:48kkurbjunrethgir: you should have privileges now
06:40:13AndrewRBwith those two lines as they are, it does that, except when I hold the button it skips the track *and* pauses.
06:40:21 Join timc [0] (n=aoeu@
06:40:35AndrewRBclearly, holding it is activating the first line and the second.
06:41:06AndrewRBplease could somebody explain how I could do that? (unless it isn't possible)
06:42:35rethgirkkurbjun: Thank you very much! :)
06:45:54AndrewRBhave i asked at a bad time, or did i say too much too fast? (i don't mean to sound demanding ><)
06:47:28kkurbjunAndrewRB, there are not too many people active at this time of the day, I'm not sure on the keymaps offhand, there is a way to make short presses and long presses do different actions, but again, I'm not sure on the method offhand
06:48:22BabaChoowacan someone here give me a hint how to start software developing for the iriver e100?
06:48:33kkurbjunI would suspect that it has something to do with the BUTTON_REL that you have since that should activate the action when you release
06:48:35 Quit AndyI (Read error: 110 (Connection timed out))
06:48:58kkurbjunI think you would want something that includes BUTTON_REPEAT immediately
06:49:14AndrewRBkkurbjun: ok, thanks for responding =) Well, I already have it responding to a "long press" - it skips to the next track when you *release* the button after *repeating* the button (holding it down)
06:49:43AndrewRBthe problem is that when it is pressed initially, it pauses it (then goes on the "repeat", then when released it "skips")
06:51:41AndrewRBhmmm. maybe if i changed the first line's button code to "BUTTON_HOLDMENU|BUTTON_REL"
06:51:43kkurbjunyeah, I think it's still with that REL line in there, that second line should always active on a release event, I think
06:52:56 Quit froggyman ("CGI:IRC (Ping timeout)")
06:53:14AndrewRBwell, the first line as it is is activating whenever the button is pressed, and by "pressed" i assume that is on press, not on release.
06:53:36kkurbjunI think what you might want is for the second line to have something like BUTTON_HOLDMENU|BUTTON_REPEAT, BUTTON_NONE },
06:53:39AndrewRBthe second line is activating whenever the button is released, *after being repeated*
06:53:49kkurbjunthat should activate when the button is pressed and held
06:53:57kkurbjunand there is no pre-req
06:54:20kkurbjunthe d2 might not be handling repeats properly though too
06:54:31AndrewRBI tried that initially - problem with that is it repeats that action while the button is held - so in reality it skips the track several times.
06:54:31kkurbjunit's touchscreeen right?
06:54:42kkurbjunoh, gotcha
06:54:44AndrewRByeah, but I'm not using the touch screen in this case
06:55:57kkurbjunhmm, maybe try oring in the REL with the repeat too, but I don't know if that is correct.. I'm sure there are some other targets that show how it should be done
06:56:04AndrewRBif the prereq is BUTTON_NONE, does that mean the pre-button doesnt matter, or it must be nothing?
06:56:12kkurbjunthe gigabeat should have an example on the menu button for going to the quickscreen
06:56:36kkurbjunI think it means that it doesn't matter
06:57:10 Quit intrados (Read error: 104 (Connection reset by peer))
06:57:14kkurbjunyeah, the gigabeat has an example in the WPS mappings for the menu
06:57:25AndrewRBright. because all i really want is for the first line to *only* activate when the button is pressed once, not repeated - so if the prereq was exclusively "nothing" then that would solve the problem
06:57:32AndrewRBok lemme have a look, thanks
06:58:10AndrewRBwhat does quickscreen do?
06:58:36kkurbjunit brings up a menu that only has 3 options that are customizable
06:59:32AndrewRBso in the gigabeat file, "BUTTON_MENU|BUTTON_REPEAT, BUTTON_MENU" is a "long press" and "BUTTON_MENU|BUTTON_REL, BUTTON_MENU" is a "short press", right?
06:59:38 Nick JdGordon|afk is now known as JdGordon (n=jonno@rockbox/developer/JdGordon)
06:59:57kkurbjunshould be
07:00:31AndrewRBthanks, let me compile and test
07:02:20AndrewRBperfect =)
07:02:22AndrewRBthankyou very much
07:02:33kkurbjun:), glad it worked
07:02:53AndrewRBi had actually been looking at other ports, but not the gigabeat one. besides, with no real knowledge of what things were meant to do, i was just guessing.
07:03:10AndrewRBnow i can fiddle with other things again, haha.
07:05:43 Join intrados [0] (
07:15:07BabaChoowaHi there! Did someone give a try to port rockbox to the iriver e100?
07:17:42JdGordonMarcGuay: pong?
07:20:16 Quit AndrewJ ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
07:26:54AndrewRBin short, yes they thought about it. no it probably isn't goign to happen
07:27:37BabaChoowaoh... too bad.
07:27:44BabaChoowathank you for the link!
07:27:51BabaChoowai'll check it.
07:28:00AndrewRBforums -> search -> type "e100"
07:32:01BabaChoowai see. unfortunately there is not that much. hmmm... Andrew, do you have an idea how "someone" could start porting rockbox to the e100? Where should that guy start?
07:32:06 Quit advcomp2019 ("Going!")
07:32:36JdGordonwe need someone to figure out the firmware encryption...
07:32:46JdGordonor find a way to run our own code on it....
07:33:12*AndrewRB was about to point at JdGordon =)
07:34:35AndrewRBreally, cowon makes it easy for us - the firmware update files can be used as a way to run code on the player
07:34:41BabaChoowathank you JdGordon. I see... encryption makes reverse engineering quite hard.
07:35:49BabaChoowaJust for my understanding, cowon is a company?
07:36:11AndrewRBsure, who make the only player i own (the Cowon D2)
07:36:41 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
07:37:11BabaChoowa*g* ... i see
07:38:29***Saving seen data "./dancer.seen"
07:40:56 Part toffe82
07:44:27AndrewRBgoodnight guys. thanks again for the help kkurbjun. (i say "night", but it is 6.44 AM... oops)
07:44:54 Quit AndrewRB ("ZZZZZZZzzzzzzzzzzzzzzzzzzzz..............")
07:48:12CIA-38New commit by 03jdgordon (r21140): fix FS #10261 - the files context menu (as an example) would get skipped and the first items context menu get shown instead
07:52:00 Quit jmillikin (Read error: 110 (Connection timed out))
07:58:21 Join Horschti [0] (n=Horscht@xbmc/user/horscht)
08:07:40 Join evilnick_home1 [0] (
08:15:45 Quit Horscht (Read error: 110 (Connection timed out))
08:19:24 Quit evilnick_home (Read error: 110 (Connection timed out))
08:24:31amiconnUnhelpful: Depends on the target. Coldfire supports reading & writing unaligned (although it's slower, hence not recommended for performance reasons). ARM and SH1 don't - you'll get an exception (Data Abort on ARM, CPUAdrErr on SH1). I don't know about MIPS
08:28:51amiconnIf you have a struct that is decalred with __attribute__((packed)), gcc will take care of unaligned accesses to multi-byte variables within the struct
08:29:34*amiconn used this gcc feature in ata_mmc.c - see send_cmd() (line 330)
08:29:55 Join Rob2222 [0] (
08:30:17 Join n1s [0] (n=n1s@rockbox/developer/n1s)
08:33:21 Quit kkurbjun (Read error: 110 (Connection timed out))
08:37:28 Quit BlakeJohnson86 (Remote closed the connection)
08:46:51 Quit dmb (Nick collision from services.)
08:47:52 Quit Rob2223 (Read error: 110 (Connection timed out))
08:49:49 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
08:50:15 Join BlakeJohnson86 [0] (
09:27:45CIA-38New commit by 03amiconn (r21141): Use bit-doubled mask everywhere in mono bitmap drawing. ~2% speedup, and smaller.
09:38:31***Saving seen data "./dancer.seen"
09:44:40JdGordonis the sorting order in the third parties forum screwed?
09:45:13JdGordonlooks like its going by first post date instead of last post date
10:11:42 Join bertrik [0] (
10:13:03 Join ender` [0] (
10:18:54 Join stoffel [0] (
10:22:11 Join flydutch [0] (
10:22:43 Quit kperri (" ajax IRC Client")
10:23:32 Part BabaChoowa
10:24:31 Quit BHSPitLappy (Remote closed the connection)
10:44:04 Join PaulJam [0] (
10:53:14 Join tvelocity [0] (n=tony@
11:02:12 Quit intrados (Read error: 104 (Connection reset by peer))
11:04:55 Quit fyrestorm (Read error: 104 (Connection reset by peer))
11:07:39 Join merbanan [0] (
11:10:18 Join robin0800 [0] (
11:13:42 Quit robin0800 (Client Quit)
11:14:02 Join robin0800 [0] (
11:15:17 Join fyrestorm [0] (
11:18:20 Quit planetbeing ("leaving")
11:28:47 Join Makuseru [0] (
11:29:19 Quit flydutch (Read error: 104 (Connection reset by peer))
11:30:05 Nick J-23 is now known as wioska_troglodyt (
11:30:16 Nick wioska_troglodyt is now known as J-23 (
11:38:35***Saving seen data "./dancer.seen"
12:00:00 Quit merbanan (Read error: 60 (Operation timed out))
12:09:01 Quit bmbl ("Woah!")
12:14:51 Quit __lifeless (Remote closed the connection)
12:15:07 Join __lifeless [0] (n=lifeless@
12:16:55 Quit timc (Connection timed out)
12:19:36 Join timc [0] (n=aoeu@
12:22:15Unhelpfulamiconn: it's not a struct. i'm looking at a hash for album titles, and it has a word-at-a-time optimization. unless a packed struct along the lines of struct unalignedwordarry { uint32_t words[0] } would work.
12:25:19 Join _lifeless [0] (n=lifeless@
12:35:46 Join PaulJam_ [0] (
12:39:15 Quit __lifeless (Read error: 110 (Connection timed out))
12:46:12 Quit PaulJam (Read error: 113 (No route to host))
12:47:18 Join robin0800_ [0] (
12:54:46 Join freeworldstat [0] (
12:55:59 Part freeworldstat
13:00:21gevaertsMaybe logf-over-HID would be useful
13:02:18Mikachucan you enable the hid without going into disk mode?
13:03:40gevaertsHID is always enabled during connection on players that support it. Currently you only get button handling in a full connection that includes storage, but that's not a fundamental issue
13:05:39martian67i imagine it will be fixed soon
13:06:19 Quit ender` (Read error: 104 (Connection reset by peer))
13:08:09gevaertsit's not a bug, so it can't be fixed
13:08:33gevaertsat some point there will certainly be more flexibility though
13:09:01 Join Lear [0] (i=chatzill@rockbox/developer/lear)
13:09:11 Quit robin0800 (Read error: 110 (Connection timed out))
13:09:25Mikachui couldn't work out how the debug hid entries were supposed to work if it was only enabled in disk mode :)
13:11:28 Quit rethgir (Read error: 110 (Connection timed out))
13:14:19 Quit Makuseru (Read error: 104 (Connection reset by peer))
13:17:11 Quit robin0800_ (Remote closed the connection)
13:17:36 Join robin0800_ [0] (
13:21:15 Quit _lifeless (Remote closed the connection)
13:21:31 Join _lifeless [0] (n=lifeless@
13:26:37 Join Thundercloud [0] (
13:28:40 Quit fyrestorm (Read error: 104 (Connection reset by peer))
13:29:11 Quit stoffel (Read error: 113 (No route to host))
13:30:42 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
13:34:29 Join ender` [0] (
13:35:25 Quit gevaerts (Nick collision from services.)
13:35:34 Quit Lear ("ChatZilla 0.9.84 [Firefox 3.5pre/20090529044053]")
13:35:34 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
13:38:38***Saving seen data "./dancer.seen"
13:39:32Unhelpfulhm... i'm storing the strings myself, i could just force the alignment.
13:42:40 Quit _lifeless (Read error: 110 (Connection timed out))
13:44:16 Quit perrikwp (Read error: 104 (Connection reset by peer))
13:44:29 Join stoffel [0] (
13:44:34 Quit saratoga (" ajax IRC Client")
13:47:58 Join PaulJam [0] (
13:53:01 Quit PaulJam_ (Read error: 104 (Connection reset by peer))
14:01:45Mikachuis there anyway to set a different color for the statusbar in the wps screen than the menu?
14:04:41 Join _lifeless [0] (n=lifeless@
14:05:08PaulJamMikachu: you could try to "rebuild" the statusbar in the WPS with your desired colour.
14:05:17Mikachuthat is what i wanted to avoid, yes
14:17:02Unhelpfulmight be worthwhile to have an example on the wiki for the statusbar as WPS ;)
14:23:43MarcGuayJdGordon: Figured it out myself - thanks.
14:32:38*Mikachu cheated;;a=commitdiff;h=f5218d5b6a
14:36:57 Join fyrestorm [0] (
14:56:41 Join LambdaCalculus37 [0] (n=rmenes@rockbox/staff/LambdaCalculus37)
14:59:25 Join Xerion_ [0] (
15:01:20 Join _Xerion [0] (
15:01:52LambdaCalculus37Getting this error when I'm trying to build for the GoGear HDD6330 as a separate target:
15:03:14 Join _Xerion_ [0] (
15:03:14***Alert Mode level 1
15:03:14DBUGEnqueued KICK Xerion
15:03:14DBUGEnqueued KICK Xerion_
15:03:14***Alert Mode level 2
15:03:14DBUGEnqueued KICK _Xerion
15:03:14DBUGEnqueued KICK _Xerion_
15:03:14***Alert Mode level 3
15:05:46 Quit Xerion_ (Read error: 60 (Operation timed out))
15:05:53linuxstbLambdaCalculus37: How have you defined CONFIG_KEYPAD ?
15:08:41LambdaCalculus37linuxstb: Let me check; where would I define it?
15:09:16linuxstbIn your config-targetname.h file
15:10:16 Join efyx_ [0] (
15:13:15***Alert Mode OFF
15:14:52 Quit Xerion (Connection timed out)
15:14:52 Nick _Xerion_ is now known as Xerion (
15:14:52DBUGEnqueued KICK Xerion
15:17:17LambdaCalculus37linuxstb: I do have it defined, as CONFIG_HDD6330_KEYPAD.
15:18:01linuxstbSo it's a different keypad to the HDD1630?
15:20:21 Quit _Xerion (Read error: 110 (Connection timed out))
15:21:05LambdaCalculus37linuxstb: It is a different keypad, but currently there doesn't seem to be a driver for it. What's there is just a re-use of the HDD1630 keypad.
15:21:48linuxstbThen you need to take care of button mappings - debug_menu.c is the first (search for CONFIG_KEYPAD)
15:23:05LambdaCalculus37linuxstb: Ahh, found it. Thanks!
15:23:20CtcpIgnored 5 channel CTCP requests in 11 minutes and 38 seconds at the last flood
15:23:20*LambdaCalculus37 adds PHILIPS_HDD6330_PAD
15:23:41 Join ved_ [0] (
15:23:41 Quit vedlith (Read error: 54 (Connection reset by peer))
15:24:43LambdaCalculus37linuxstb: Great success. :)
15:25:56LambdaCalculus37Snag #2:
15:27:04linuxstbLooks like the same thing...
15:27:35 Join cool_walking_ [0] (
15:27:42 Quit Xerion (Read error: 60 (Operation timed out))
15:28:52LambdaCalculus37linuxstb: I'm on it.
15:33:46 Join mishu [0] (n=5679d00f@gateway/web/cgi-irc/
15:34:03mishuhi to all
15:34:23mishui have a problem with my sansa
15:34:34mishusome help
15:34:48EtuMaybe, what's your problem?
15:35:23mishuon my screen appear
15:35:26mishuLoad main image failed Switch to Recovery mode
15:35:45mishuwhat shoud i do?
15:35:57EtuWhat did you do?
15:36:17mishumy brother has performed a format
15:36:21Etu(I think it's bricked or something, but Im not sure)
15:36:23mishufrom xp pro
15:37:11EtuI bricked my sansa some days ago and successfully unbricked it
15:37:12mishui searched all over the net and forums
15:37:16EtuIt was cool :)
15:37:28mishuno success
15:38:03mishuit is a sansa e280 8g
15:38:05LambdaCalculus37linuxstb: I disabled the USB stack for the time being. It builds now.
15:38:26LambdaCalculus37Bah, spoke too soon. :/
15:38:39***Saving seen data "./dancer.seen"
15:39:20Etumishu: I used this guide on my Sansa e250 some days ago:
15:39:55mishui tried some softwares but .. no luck
15:40:00mishulet's see
15:40:34linuxstbLambdaCalculus37: You shouldn't have needed to do that - just add definitions for the USB buttons in firmware/export/usb.h
15:40:37Etue200tool is alphasoftware, but it worked for me
15:40:45mishufrom windows dosen't work?
15:41:28EtuI don't know, I don't run windows
15:41:35linuxstbEtu: It's not really "alpha", just something designed for low-level hacking - not general users.
15:41:39LambdaCalculus37linuxstb: Okay.
15:41:46mishuplease escuse my english
15:41:58Etulinuxstb: The text told me so.
15:42:20linuxstbmishu: With a lot of effort I think it can be made to work in Windows, but it's far easier in Linux or OS X.
15:43:34Etumishu: The first part in the guide I sent you tells you how to make a liveCD to run it from
15:43:48mishui will try it
15:44:00EtuGood Luck :)
15:44:12mishui make this thing in windows
15:44:22mishuno luck
15:45:39mishuit work's from centos live cd?
15:45:47mishucentos 5.2
15:47:23EtuI think so
15:47:36Etuwhat linux system isn't so important
15:47:42EtuI did it from gentoo
15:50:53mishuEtu: what is e200tool and sansa.fmt?
15:51:23mishuthe rest of files i have them
15:52:42 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
15:59:34linuxstbmishu: "e200tool" is the e200tool program itself. "sansa.fmt" can be any file - the content isn't important, it just needs to have that name.
16:01:38 Join Xerion [0] (
16:05:13 Quit robin0800_ (Remote closed the connection)
16:05:37 Join robin0800_ [0] (
16:07:59 Join Rawkins [0] (n=52e74d63@gateway/web/cgi-irc/
16:08:03 Join flydutch [0] (
16:08:31 Join kugel [0] (n=kugel@rockbox/developer/kugel)
16:09:07 Join Moi123 [0] (
16:09:17Moi123Hi everybody
16:09:29mishuEtu: i see it, sorry
16:09:42*kugel is going to commit the pictureflow-start-playback patch
16:09:49Moi123I wanted to know, if it's possible to install Rockbox on a 4th generation iPod nano (chromatic) ?
16:10:34Moi123Oh :(
16:10:36Moi123Ok !
16:10:55Moi123And something is under developpement to support this iPod or it's just impossible ?
16:11:06Moi123(sorry for my English, I'm french)
16:11:54linuxstbIt's been impossible so far...
16:11:56 Quit LambdaCalculus37 ("Fwump")
16:12:23 Quit perrikwp (" ajax IRC Client")
16:13:02 Quit Rawkins ("CGI:IRC (Ping timeout)")
16:13:19Moi123thanks all
16:13:33Moi123It's the same for iPodLinux I presume ?
16:15:06krazykitdon't you think if they had it working, rockbox would too?
16:15:50 Join petur [50] (n=petur@rockbox/developer/petur)
16:16:13 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
16:17:34 Join gregzx [0] (
16:27:04 Quit robin0800_ (Remote closed the connection)
16:27:45 Quit mishu ("CGI:IRC")
16:27:57 Join mishu [0] (n=5679d00f@gateway/web/cgi-irc/
16:28:33linuxstbkugel: What exactly are you planning on committing? Wasn't there objections to committing some parts of that patch?
16:28:57kugelonly the playback starting part
16:29:14kugelthere was objections to the database integration part
16:33:37CIA-38New commit by 03kugel (r21142): playlist start can only have 1 return value (0), so make it return nothing.
16:33:40 Quit antitrons (Read error: 104 (Connection reset by peer))
16:33:54 Join antil33t [0] (n=Mudkips@
16:38:27 Quit Thundercloud (Remote closed the connection)
16:40:28CIA-38New commit by 03kugel (r21143): Commit the first part of FS #10263: Starting playback from within pictureflow, by creating a playlist from the tracklist and playing it. The database ...
16:42:32 Join lyngaas [0] (
16:43:06***Alert Mode level 1
16:43:06CIA-38New commit by 03kugel (r21144): Bump plugin api version.
16:53:07***Alert Mode OFF
17:03:02bertrikI was also under the impression that there was no agreement yet on using pictureflow to start playback
17:04:12gevaertsI don't see why it shouldn't. I agree about not integrating it too much before it's more or less complete, but what's wrong with starting playback?
17:06:55bertrikI didn't see any discussion about it
17:07:59linuxstbbertrik: You disagree that pictureflow should start playback?
17:08:09CIA-38New commit by 03kugel (r21145): Attach the playback control menu to pictureflow's main menu.
17:09:15linuxstbbertrik: Then what do you think needs discussing? I'm not sure if anyone has disagreed with the idea of that feature.
17:11:12bertrikok, I'm not against the feature, but I just remember stuff like this being discussed before being committed, now it seems it's the other way around
17:12:29 Join kkurbjun [0] (n=kkurbjun@rockbox/developer/kkurbjun)
17:13:11kugelwe discussed yesterday a bit, where no explicit objections against the playback feature were made (Llorean even explicitely said he isn't against it)
17:13:12linuxstbI think there's been some discussion about how to integrate it in the core, but I haven't really been paying attention. There's also a tendancy for -community to host discussions nowadays... :(
17:13:34kugelI don't see why a much-asked-for addition to a plugin needs much discussion
17:13:57kugelthere was no discussion in -community about it (not when I was there at least)
17:14:30 Join robin0800 [0] (
17:15:05kugelevil %BR% in MajorChanges :(
17:15:08n1si think that it's the general direction of pf and the database UI that should be discussed so we can agree how pf should be integrated in the database browser. (since it seems everybody want's to be)
17:16:08kugelintegration of plugins into the core is always controversial and needs discussion
17:16:20bertrikkugel, I don't disagree with the new feature, but I can't follow your reasoning that a much-asked-for-addition to a plugin should not need discussion
17:17:32kugelwhy do you think it needs discussion?
17:18:57Lloreanbertrik: The objection wasn't to having it start playback (from me at least) but rather from having it presented as part of the database while still missing a lot of functionality
17:19:43bertrikok, I see
17:20:06Lloreanlinuxstb: The current status of the plugin is that it can start playback, but it can't do just about everything else. I'm all for including it, but I think it should be a demo, and not part of the database list, until it can at least do the Playlist-related context menu stuff, and let you resume/stop/manipulate playback via the normal buttons that work in list contexts.
17:20:22kugelUnhelpful: What's the fastest speed pf could autoscroll without showing empty slides?
17:21:20Unhelpfulkugel: that obviously depends on how the scheduler handles the cache thread, and on how responsive storage is.
17:21:35kugelLlorean: it can also control playback via the plugin's playback control menu. And going directly to the wps is just a keymap issue (I've implemented directly from plugin to the wps)
17:21:42 Join preglow [0] (i=thomj@rockbox/developer/preglow)
17:22:31Lloreankugel: That still leaves playlisting, and if you can do it via the plugin's playback menu shouldn't using the normal keys for it also just be a keymap issue?
17:23:09kugelyou can't control playback in the list either (except for stopping)
17:23:12 Join shadoxx [0] (
17:23:24Lloreankugel: On the Gigabeat S you can skip forward and back
17:23:31LloreanAnd on the Gigabeat F and S you can adjust volume
17:23:49kugelwell, I think it should be possible to do that in pf
17:24:35kugelit's "just" a list too, so all core list keymaps should be possible to implement in pf
17:24:45LloreanThat's more or less what I expected.
17:25:08LloreanThat's why the main issue I've talked about is the playlisting one.
17:26:33kugelalso, if we want pf as a valuable alternative, we should reconsider the backlight settings and perma-boosts it uses
17:27:09LloreanI think perma-boost is okay, but it should follow the core backlight settings.
17:27:20Unhelpfulkugel: why don't i try a patch for CONTEXT_CUSTOM, and see if rebasing on CONTEXT_TREE works...
17:28:29Unhelpfulkugel: to get those other key mappings. even if it means some reworking of the existing PF contexts.
17:29:09kugelI didn't understand your sentence
17:29:11 Quit mishu ("CGI:IRC (Ping timeout)")
17:30:02Unhelpfulkugel: i'm going to try the changes i mentioned last night to make it possible to chain to CONTEXT_TREE from get_custom_action, and see just how badly doing so breaks things :)
17:30:20 Join cmwslw [0] (
17:30:21 Join mishu [0] (n=5679d00f@gateway/web/cgi-irc/
17:30:27kugelthat would be nice
17:31:17 Quit cmwslw (Client Quit)
17:31:25 Join cmwslw [0] (
17:31:25 Quit mishu (Client Quit)
17:36:28 Join Lear [0] (i=chatzill@rockbox/developer/lear)
17:37:24bertrikkugel: one thing I wonder is what the plans are for the pictureflow plugin. E.g. is it planned for it to become a full replacement for the current database?
17:38:40***Saving seen data "./dancer.seen"
17:39:08Lloreanbertrik: The idea is for it to have limited uses within the current database, I believe
17:39:47LloreanInitially, just an option to view all albums via cover, rather than simply as text, but (hopefully) later to be extended so it can be used in place of the "Album" filter anywhere albums are listed, just as an alternative within tagnavi.
17:40:47kugelbertrik: surely not a replacement, but hopefully a usable alternative
17:41:59Mikachupf is pretty useless if you don't have albumart in the first place :)
17:43:19*Unhelpful thinks an "alternative" is a much better idea, though including it in the standard tagnavi as "covers" without removing "albums" seems reasonable.
17:43:26*Llorean would very much like the pictureflow work end result be several interesting database visualization and custom filter plugins.
17:43:45Unhelpfulkugel: try this:
17:43:48LloreanBut that still requires the not-insignificant step of actually being able to use plugins as filters.
17:44:49bertrikok cool, like allowing capable targets to start pictureflow on power-on and have basic playback capabilities from it
17:45:29Unhelpfuli'm not really sure what to do on M3, since CONTEXT_TREE falls through to CONTEXT_STD (without any remote buttons :/)
17:45:46LearIn thread.c, there's a "#elif CPU_MIPS == 32" line. Shouldn't that be "defined(CPU_MIPS)" like everywhere else CPU_MIPS is used?
17:45:54Unhelpfulbertrik: "capable targets" are all bitmap ones :)
17:47:38UnhelpfulM3 might just need the ok/cancel/context/scroll remote maps pulled into pictureflow's buttons context
17:49:16 Join kugel_ [0] (
17:49:30 Join mcuelenaere [0] (
17:49:32 Quit kugel (Nick collision from services.)
17:49:40 Nick kugel_ is now known as kugel (
17:49:46mcuelenaereLear: I think I did that for being future-compatible with 64-bit MIPS processors
17:51:08Learmcuelenaere: I see. I get warnings from GCC 4.4 about it though. :)
17:51:23mcuelenaerehmm what warnings?
17:52:16LearAbout comparing 32 against something that isn't defined.
17:52:40Unhelpfuladd a defined(CPU_MIPS) && first :)
17:53:14Lloreann1s: Will that work (from your email)? I'm not too familiar with shell type things, but shouldn't it be something like &> because they're stderr?
17:53:48LearHm, there seem to be a flac problem. test_codec fails after the flac tracks...
17:54:49n1sLlorean: ah, you're right, wasn't thinking clearly
17:55:29CIA-38New commit by 03learman (r21146): Fix typo. Not sure how it will affect targets with that CPU though...
17:55:51kugelLear: probably related to the issue I was talking about yesterday
17:56:26Learkugel: Yep.
17:57:16n1sLear: testing gcc4.4?
17:58:06 Join bluebrother [0] (
17:58:07Learn1s: Yes, thought I should give it a go. Builds without much trouble and generates smaller code. That's all I know for now.
17:58:41LloreanLear: How's code speed?
17:58:49kugelUnhelpful: seems to work
17:59:04n1sLear: which targets are you testing?
17:59:23LearLlorean: Haven't gotten that far yet. n1s: ARM only (for now at least).
17:59:24kugelLear: I think we already tested gcc4.4
18:00:06n1ssome guy in the ml tested a gcc 4.4 rc for ar i think, he got slightly worse speeds overall iirc
18:00:08kugelone guy on the -dev ml tested 4.4-RCsomething, and I myself tested the last 4.3.x release
18:00:43LearThis is the release version though.
18:01:07Unhelpfulkugel: if existing controls work on supported target sims, then it probably doesn't break anything.
18:01:09kugeln1s: I think the glibc might actually be the problem? We know dreppers view on ARM
18:01:17n1syes, it may indeed behave differently
18:01:52mcuelenaerekugel: wasn't eglibc created for that?
18:02:03kugelI think so, yes.
18:02:19*n1s is confused
18:02:28kugelI didn't compile gcc though with eglibc
18:03:22kugelUnhelpful: my e200 seems fine
18:03:29n1sbut we don't use libc
18:03:46kugeln1s: we do when compiling gcc
18:04:01kugelI think
18:04:13LearVorbis didn't work at least...
18:04:46Unhelpfulkugel: then i guess you can try hooking up the extra button maps you wanted to something :)
18:04:48n1skugel: yes (basically *everything* on a unix system uses it) but why should it affect the performance of the output of gcc?
18:05:21kugelif the cross-compiler is broken, the generated code might be broken too
18:05:48kugelI accept that a newer compiler might be *a bit* slower, but not that e.g. several codecs don't work at all
18:06:08LloreanWe've seen cross-compilers known to work elsewhere build completely nonworking (or very strangely working) versions of Rockbox before.
18:06:42n1sI dont' understand what you mean or why some obscure glibc bug should cause this?
18:07:03*Llorean was chiming in without context, feel free to ignore me if that info makes no sense here.
18:07:05Unhelpfulkugel: i used distribution-built cross-compilers for a bit... found out when i tried to build for archos that they couldn't build a text_editor that would fit in the plugin buffer.
18:08:07kugeln1s: we build the cross-compiler using the stock glibc. It's known that Drepper gives a shit on e.g. ARM, so it's not unlikely that the ARM-cross-compiler is bugged
18:08:30*n1s had to patch quite a few places to get a working rockbox out of gcc 4.3 but i dont' really blame the compiler for that
18:09:03*Mikachu still uses his super old gcc 4.0.3 arm gcc
18:09:06 Join toffe82 [0] (
18:09:10kkurbjunto my knowledge glibc should not be effecting the cross compiled code at all, if there was a bug in glibc you would see some more serious problem on your system
18:09:13n1skugel: but your cross compiler uses your native glibc
18:09:21bertrikn1s, were those bugs in rockbox?
18:09:30n1srockbox doesn't use it at all on target
18:09:39n1sbertrik: well, sort of
18:10:16kugelkkurbjun: If that's true then I'm sorry
18:10:44 Join saratoga [0] (n=41becb3b@gateway/web/cgi-irc/
18:10:44 Quit saratoga (Client Quit)
18:10:47kkurbjunthe problem with using newer compilers to my knowledge with code breaking is how it approaches optimizations and how much you used compiler specific flags to get things working (code size can also be a problem as unhelpful mentioned)
18:10:52 Join saratogahome [0] (n=41becb3b@gateway/web/cgi-irc/
18:11:05n1sbertrik: mainly that newer gcc can decide to delete asm statements that it deems to have no side effect if they are not volatile and a subtle change in linker script syntax
18:11:37saratogahomelack of volatile keywords in rockbox sounds like a potiential bug on our part
18:11:40n1scode size was a problem too and a real bug in the m68k compiler that's fixed in 4.4
18:12:58n1sand a few of the issues we ran into when switching to Os on coldfire with gcc 3.4 (deletion of delay loops in c basically)
18:13:13kugelUnhelpful: are you going to try a few sims?
18:13:20bluebrotherhmm, dirskip crashes the h100 reproducably :(
18:13:24n1ssaratogahome: yes, i meant to commit that patch but never got around to it
18:13:44n1smaybe i should do that now :)
18:14:11Unhelpfulsaratogahome: depends, non-volatile asm that gcc is able to understand can give it a chance to optimize.
18:15:23saratogahomethats true, but i think the parts getting optimized away are probably delay loops and such, since the list of output registers for real functions should prevent this problem anyway i think
18:15:26kugelI wouldn't be surprised if gcc recognizes dozens of consecutive nosp and optimises them away
18:15:44bertrikn1s, ok, sounds like a good thing to fix (even if it doesn't occur with the current gcc)
18:15:47saratogahomethough i admit i have no idea how gcc works
18:16:09kugelUnhelpful: ACTION_TREE_STOP works too
18:16:25Unhelpfulkugel: OK?
18:16:46kugelwas that a question?
18:17:16Unhelpfuli tried on my beast, and it works ok... beast uses the windows button for the PF menu, though, and that's the return to WPS button :/
18:17:58Unhelpfulkugel: i wasn't sure what you wanted ACTION_TREE_STOP for... as a stop playback hotkey?
18:18:23bertrikmeh, I can't really get a grip on reliable i2c communication on as3525 when the CPU clock is much higher than the i2c clock. I wish I could find more detailed info about it than what is available in the as3525 datasheet (maybe have a look in the linux kernel)
18:18:52Unhelpfuli think i stole that button on more than a few targets for the "quit PF" mapping.
18:19:10Unhelpfulbut using cancel or the menu to exit would be acceptable, i suppose
18:19:26kugelif we're going to chain lists, then the ACTION_STD_CANCEL button shold handle pf quit IMO
18:19:35kugelwhich is e.g. |<< on e200
18:20:10 Quit killan (Read error: 54 (Connection reset by peer))
18:20:16 Join killan [0] (
18:20:49Unhelpfulkugel: ACTION_STD_CANCEL already *does* quit PF if in the album list. but the dedicated quit mapping quits from tracklist as well.
18:21:02 Part cmwslw ("Ex-Chat")
18:21:06kugelyep, I noticed that
18:23:58Unhelpfulseveral devs still thought a dedicated quit wasn't entirely redundant... but i'd say it's fair to replace dedicated quit with a playback control. you get to edit the manual, too, though ;)
18:25:14 Quit petur ("Zzzzzz")
18:26:01kugelUnhelpful: no, AlexP does :P
18:27:19kugelI don't think it's entirely reduntant either, just depends on whether the target has enough buttons
18:27:21 Quit n1s (Read error: 104 (Connection reset by peer))
18:27:29 Join n1s [0] (n=n1s@rockbox/developer/n1s)
18:28:36kugelUnhelpful: that giant CONFIG_KEYPAD is just for excluding scrollwheel targets, isn't it?
18:30:08Unhelpfulkugel: no, that's a large set of targets with the same button mappings.
18:30:54 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
18:31:11Unhelpfulbasically everything other than ondio and m3 - m3 because it should be using the remote buttons, ondio because it needs to explicitly mask some CONTEXT_STD mappings.
18:32:06kugelI'm thinking the ondio,m3,mrobe500 cases should be moved up, and the giant construct be replaced with #elif !defined(HAVE_SCROLLWHEEL)
18:32:38Unhelpfulthat should work fine.
18:33:25CIA-38New commit by 03nls (r21147): Old patch from FS #7832: Sprinkle 'volatile' in the various inline assembler statements, needed for the driver to work when compiled with newer gcc, ...
18:38:22kugelUnhelpful: does that look good?
18:38:52kugelI also put the ondio within the the !HAVE_SCROLLWHEEL, keeping the extra definitions
18:40:39Unhelpfulkugel: perhaps the album list context itself should be in !HAVE_SCROLLWHEEL?
18:41:17kugelthat should work
18:41:30kugelbut that needs another #ifdef at get_action, then
18:42:51Unhelpfuli still think it's a good idea, seeing as it reduces static data and code on scrollwheel targets.
18:44:30 Quit shadoxx (Read error: 113 (No route to host))
18:44:57 Quit cool_walking_ (Remote closed the connection)
18:45:03 Join webguest00 [0] (n=c3893ab3@gateway/web/cgi-irc/
18:45:45 Join jmillikin [0] (
18:46:07Unhelpfulif we're killing the dedicated quit button, you know, that actually means that we don't need to use custom actions at all on scrollwheel targets, and that there's only one custom action context on other targets.
18:47:28kugelright, I actually deleted the PF_QUIT for e200
18:47:39 Quit webguest00 (Client Quit)
18:48:24Unhelpfulthe touchscreen stuff can probably go?
18:50:41kugelthough, scrollwheel targets might still want a dedicated quit button
18:50:57kugelor cobo
18:50:58 Join archivator [0] (n=archivat@
18:51:02Unhelpfulwhat's wrong with cancel and the menu quit option?
18:51:39archivatorI'm getting Connection refused on, is the address changed or did you drop git support entirely?
18:52:29Unhelpfulgit's been down a few days now, i'm not sure what's happening with it. git-svn still works fine.
18:53:05 Join PaulJam_ [0] (
18:53:16archivatorHmm, I never set up git-svn, guess now is as good a time as any :)
18:54:45kugelUnhelpful: c200 still works, I'm going to commit the cleanup
18:55:06 Join PaulJam__ [0] (
18:55:11Unhelpfulkugel: i'm going to go over the supported-target keymaps and check for conflicts, and then go ahead and commit the rebase of the PF contexts from CONTEXT_STD to CONTEXT_TREE... it looks like it should be safe, though.
18:55:19 Quit mcuelenaere ()
18:55:29 Join VytenisS [0] (
18:55:38 Quit Chex (Read error: 110 (Connection timed out))
18:55:49 Quit MarcGuay (Read error: 110 (Connection timed out))
18:56:57CIA-38New commit by 03kugel (r21148): Restructure/cleanup the album list buttom mapping, removing it entirely for scrollwheel. No functional change (scrollwheel didn't use it before, ...
18:58:15 Quit PaulJam (Read error: 60 (Operation timed out))
18:59:54Nico_Pkugel: hi, nice work on pictureflow :)
19:02:35kugelNico_P: Thanks :)
19:03:03 Quit PaulJam_ (Read error: 113 (No route to host))
19:05:34 Quit toffe82 (Read error: 110 (Connection timed out))
19:09:24JdGordonkugel: 8799....
19:13:11bertrikyay, I think I have a workaround for the i2c problem on ams sansas with high cpu clock
19:14:54 Join PaulJam [0] (
19:14:59 Join MarcGuay [0] (
19:17:09kugelhm, there's a problem with the playlist created by pf
19:17:42 Quit robin0800 (Remote closed the connection)
19:18:18Unhelpfulwhat's the problem?
19:19:04 Quit Moi123 (Read error: 113 (No route to host))
19:19:18 Join fdinel [0] (
19:19:21 Join FlynDice [0] (n=FlynDice@
19:20:06 Quit PaulJam__ (Read error: 60 (Operation timed out))
19:21:31 Join toffe82 [0] (
19:24:43FlynDicebertrik: I've been playing with i2c delays a lot lately with varying results, I see you think you found a workaround?
19:26:03 Join PaulJam_ [0] (
19:26:20bertrikFlynDice, yes, I'm testing it more thoroughly now. The essence is that instead of waiting for the busy bit, we wait for the DACNT register to become 0. The DACNT contains the number of pending i2c bytes to be read/written.
19:27:28kugelUnhelpful: you can't resume it after reboot
19:27:50Unhelpfulkugel: i wonder why?
19:28:09kugelso I do
19:29:11 Join SirFunk__ [0] (
19:29:33 Quit gregzx ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
19:29:34 Join dmb [0] (n=dmb@unaffiliated/dmb)
19:30:14JdGordonLlorean: usb charge patch? in for 3.3 or not? and the thread sorting in the 3rd party forum is on first post not last post...
19:30:41 Join __lifeless [0] (n=lifeless@
19:31:16FlynDicebertrik: Great!! Any chance this would benefit the radio function on the e200's that fails when the mmu is enabled or is that going to be separate?
19:31:25LloreanJdGordon: We either need the charge patch, or USB should still be disabled for the iPods (but can be disabled for the Sansas I believe). I think the charge patch should at least be committed one way or another as long as someone's willing to sign off on "it shouldn't damage any hardware, we think"
19:31:30 Join SeaWeed [0] (
19:31:38LloreanAFAIK it's the same basic thing the H300 does, but I'm hardly in a position to be too knowledgeable about it.
19:32:08Unhelpfulkugel: a lot of targets seem to do funky redefines of left/right with various release or repeat flags. i think rebasing on CONTEXT_TREE will have to mean moving the masking items for those to all targets.
19:32:11JdGordonget artound to testig it?
19:32:27LloreanJdGordon: Oh, thought you meant a different patch.
19:32:39LloreanIf you mean *your* patch, I didn't notice you'd updated a new one since I said it didn't compile?
19:32:45 Quit PaulJam (Read error: 113 (No route to host))
19:32:51JdGordonoh, i didnt remember that
19:32:54JdGordonill fix it up now
19:32:56LloreanThe LANG issue?
19:33:02kugelUnhelpful: why? you're overriding their left/right definitions?
19:33:10LloreanAnyway, as long as your patch does what it says it does (and it seems like it ought to), I'm for it being included.
19:33:13gevaertsI think we should enable USB on the sansas, leave it disabled on ipods, and think about the others
19:33:18LloreanJdGordon: Which third party forum?
19:33:29Lloreangevaerts: Any opinions on the charging patch?
19:33:35JdGordon3rd party tools or whatevr its called
19:33:45LloreanI'd *really* like it if we could enable it on the iPods as well, but charging is necessary.
19:34:00LloreanJdGordon: Other utilities, I see
19:34:28Unhelpfulkugel: for example, several map BUTTON_LEFT|BUTTON_REL, BUTTON_LEFT to cancel, with the result that the button down might start PF scrolling, but the button up would quit PF.
19:35:06kugelUnhelpful: pf should map this to scroll anyway, imo
19:35:18LloreanJdGordon: Forum should be fixed, assuming I'm looking at the right one. Check for me?
19:35:40JdGordonyep, cheers
19:35:54 Quit archivator ()
19:36:14 Join froggyman [0] (n=47ba0b80@gateway/web/cgi-irc/
19:36:15Unhelpfulkugel: the problem is that these defines are not universally using the same preconditions, and that they can change based on a setting. safest bet is for the PF scroll button to be down without precondition, and repeat without precondition to repeat, and to mask any with-precondition mappings.
19:36:30bertrikFlynDice, you mean the radio issue on e200v2s? that is completely separate from ascodec i2c
19:36:41 Join gregzx [0] (
19:37:21FlynDicebertrik: yes I see that now :(
19:38:01gevaertsLlorean: I'll try to find some time to integrate that properly with the usb code. I have no opinion on whether it's correct though
19:38:03bertrikthat can probably be fixed by using a larger/smarter delay loop in fm_delay in fmradio-i2c-as3525
19:38:40Lloreangevaerts: Well, as long as it's what the H300 does (and we offer the same option name for USB charging, whatever it's called) I think it should be okay to include it.
19:38:43***Saving seen data "./dancer.seen"
19:38:52Lloreansoap offered some concerns as to whether it could cause damage to accessories.
19:39:40*soap offered his interpretation of what dreamlayers appeared to be warning against. ;)
19:40:13 Quit SirFunk_ (Read error: 110 (Connection timed out))
19:40:21gevaertsLlorean: H300-style is the minimum, which should be easy. Integrating it into proper USB handling should be reasonably easy as well. Evaluating whether things will blow up is outside my area of expertise
19:40:49soapI use the patch everyday. What sort of potential problems should I be looking for?
19:40:52Lloreangevaerts: Well, the OF for the H300 didn't warn things might blow
19:41:06soap(outside hubs catching on fire, dogs and cats living together...)
19:41:14LloreanOne would expect iRiver to at least provide warnings to help prevent lawsuits for blown hubs
19:41:16 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
19:41:39Lloreansoap: From what I was told (I think by LinusN) worst case should just be if the port can't provide enough power, USB won't connect period.
19:42:53soapLet me reread the comments (as I don't grok the code), but I thought the issue was with the patch that 500 was being taken, without request, so the connection might very well happen "under false pretenses".
19:43:11JdGordonLlorean: updated
19:44:45 Quit Lear ("ChatZilla 0.9.84 [Firefox 3.5pre/20090529044053]")
19:46:25 Quit _lifeless (Read error: 113 (No route to host))
19:46:54 Quit stoffel (Read error: 113 (No route to host))
19:47:59CIA-38New commit by 03unhelpful (r21149): Replace use of CONTEXT_CUSTOM by get_custom_action with new CONTEXT_PLUGIN, to prevent conflicts with core contexts using CONTEXT_CUSTOM, and use ...
19:48:21CIA-38New commit by 03bertrik (r21150): Use I2C2_DACNT register (number of pending i2c bytes to read/write) to determine if an ascodec i2c transfer is done. This should fix i2c problems with ...
19:50:36 Quit antil33t (Read error: 104 (Connection reset by peer))
19:50:37Unhelpfulkugel: now you can go mapping tree-context playback controls and such :)
19:50:50kugelah wonderful :)
19:50:51 Join antil33t [0] (n=Mudkips@
19:52:29Unhelpfulin the long run, i'd say remove pf_context_buttons entirely, except on M3, where it can be a copy of the core remote standard context so that we get the right buttons for core cancel etc on remote
19:59:05kugelUnhelpful: got the playlist thing
20:02:20VytenisSHello, can anyone have a look at and commit if everything is ok ?
20:03:24saratogahome bertrik: is that commit mostly for the radio or do other things use i2c?
20:04:03bertriksaratogahome, this commit was only for the i2c towards the "ascodec" part in the as3525
20:04:24bertrikthe radio in the ams targets uses i2c-over-GPIO
20:05:26kugelbertrik: any chance it fixes existing prolems?
20:05:39bertrikthe as3525 does have another i2c bus, but so far it appears it's not used for anything
20:06:33bertrikkugel, maybe, at least it should take away another unsure variable from getting the MMU to work
20:08:44FlynDicemaking mmu build with it right now...
20:09:37bertrikthe latest patch from funman on FS #10048 gives me a black screen
20:14:54FlynDiceWell, similar problems with playback still but it does solve the problem we were using the i2c_busy delays for.
20:16:21 Quit SirFunk__ (Connection timed out)
20:17:15bertrikmy gut feeling is there's still something wrong with pcm on ams sansa, maybe a problem when both the sd and pcm want to DMA for example
20:17:32 Join SirFunk [0] (
20:18:11FlynDice2 different channels though, should that be a problem?
20:18:47*JdGordon requests testers for FS #10198 so it can hopefully go in before the freeze...
20:25:15LloreanJdGordon: It doesn't seem that major, why not just commit it?
20:25:34JdGordonbecause i might have stuffed it...
20:27:37saratogahomewe'll find out if you commit it!
20:28:08LloreanHave a little confidence in yourself
20:28:32LloreanIt's a relatively simple patch, and it works for you, right?
20:28:55LloreanYou can get a little widespread testing in the few days before freeze, that's much better! :-P
20:29:38JdGordonyou're doing reverse pycology on me arnt you!!...
20:30:40LloreanSeriously though, I read it, it's not terribly complex. If you stuffed it up it's most likely just gonna be a typo you can fix in a moment or two anyway, right?
20:32:19bertrikFlynDice, you are testing on a sansa fuze right, I can't get my clip's screen to show with funman's latest patch
20:33:42FlynDicebertrik: e280V2 I'm runing part funman part my code right now, I'll try and post a patch later tonight.
20:34:59 Quit SeaWeed (Remote closed the connection)
20:44:20gevaertsLlorean: what I think we should do long-term (for 3.3 if possible, but there's not that much time...) is to (a) properly negotiate with the host, and (b) have a "charge over USB" setting which specifies whether we should try to get 100mA or 500mA in that negotiation.
20:45:02gevaertsThat would mean that if you have charge over usb enabled, and the host says no, the player doesn't connect, but it can't blow up stuff either
20:45:25gevaertsAt the same time, it provides a workaround if you have one of those 100mA only ports
20:45:37Lloreangevaerts: I *think* that's more or less the H300's behaviour
20:45:52LloreanIt has a charge setting, and I think the hardware itself handles the negotiation because it's hardware USB. But I'm not sure.
20:46:13gevaertssounds plausible anyway
20:46:45LloreanEither way, I'd like to see USB and I think it should be enabled for everything that won't have the battery die while it's connected
20:47:43gevaertsI'll do my best to have something during the next week. We can then decide later on if we find the speed acceptable
20:48:11LloreanI don't think speed should be a major concern.
20:48:14 Quit Nico_P (Remote closed the connection)
20:49:11gevaertsI don't think we're slower than disk mode, so that shouldn't be a problem
20:50:00*Llorean nods
20:50:45gevaertswe're still be a bit slower on e200, but that's compensated by this database update thing
20:51:03gevaerts4.2MB/s vs 4.8MB/s in my measurements
20:51:55LloreanIt's fine by me as long as dual boot remains an option, and data won't be lost/corrupted from battery failures. :-P
20:52:22 Join webguest36 [0] (n=47f0bd95@gateway/web/cgi-irc/
20:53:02gevaertsdo we know if charging over USB works properly on other PP502x devices?
20:53:08 Quit webguest36 (Client Quit)
20:53:32gevaertsi.e. H10 and mr100
20:54:10 Join chandoo [0] (
20:54:49LloreanI haven't heard reports either way on those.
20:55:21LloreanI imagine it probably doesn't. It had to be setup on the Sansa too, that just happens to have been done before we had USB IIRC.
20:57:00 Join matsl_ [0] (
20:57:20 Quit kugel (Read error: 110 (Connection timed out))
20:57:54gevaertsso what do we do with those?
20:58:17gevaertsH10 users aren't very happy about their disk mode right now
20:58:41LloreanWell, in my opinion slow/fast charging doesn't matter as much as "will the battery die in long transfers"
20:59:06LloreanSo I guess we need a test. A long transfer and a check to see if the battery is lower or higher afterward.
20:59:36LloreanAs long as the battery isn't going to die mid-transfer, I think it's okay to enable USB disk mode with a warning that if they want to charge they should follow some alternative procedure.
20:59:50 Join kugel [0] (n=kugel@rockbox/developer/kugel)
21:00:26gevaertsthat also depends a lot on battery state
21:00:58LloreanState how? Shouldn't it simply be "if the current used to run the device is less than the current received over USB, the battery won't drain"?
21:02:00gevaertswell, does the charger get its allowed 100mA when not configured, and is 100mA enough to keep the battery topped up?
21:02:21gevaertsA lot depends on exact charger design and setup I think, and I know nearly nothing about that
21:02:22 Quit froggyman ("CGI:IRC (EOF)")
21:02:47*gevaerts thinks he should build a measurement USB cable
21:02:50LloreanWell, either way they need investigated so we at least know the current behaviour.
21:02:53 Join froggyman [0] (n=47ba0b80@gateway/web/cgi-irc/
21:03:42 Join PaulJam [0] (
21:03:42gevaertsI can test H10/5 as soon as I get this cable cut correctly so I can measure current, which should be a lot more reliable than seeing if it goes up a bit
21:04:33LloreanDon't H10s have user-replaceable batteries?
21:05:00gevaertsI may be able to test mr100 as well, but it needs an OF restore first as it currently doesn't boot at all. It also *may* have a bad disk...
21:05:13gevaertsthe h10 battery is replaceable, yes
21:06:53LloreanWould it be possible to see if USB power is enough to run it without the battery present, or is it unable to run without the battery period?
21:07:15gevaertsI should test that
21:09:48gevaertsI can't boot rockbox yet (empty battery), but the ROM usb mode aparently keeps working if I remove the battery
21:10:22 Quit PaulJam_ (Read error: 60 (Operation timed out))
21:11:24 Quit saratogahome ("CGI:IRC (EOF)")
21:17:09 Join _lifeless [0] (n=lifeless@
21:17:18 Quit bmbl ("Woah!")
21:17:20 Join merbanan [0] (
21:17:37kugelheh, the bot is nice
21:17:48kugelno message means my commit probably didn't come through
21:19:37CIA-38New commit by 03kugel (r21151): A bit more work on playback controlling pictureflow: ...
21:21:45Lloreangevaerts: I think it's safe to have any button work, since otherwise what button is harmless changes from screen to screen
21:22:31bertrikkugel, are you still seeing some kind of positive effect on your player from the DBOP FIFO patch?
21:22:38LloreanFor example, holding "left" is extremely harmless in the root menu, while holding "menu" or "select" (traditional buttons for USB avoidance on many of our players) leaves you in other menus, one of which (the quickscreen) often leaves people with changed settings the first time they try to leave it and press the wrong thing
21:22:44gevaertsLlorean: usually the chosen button is menu or equivalent. That should be harmless on any screen
21:22:49kugelbertrik: not anymore, after I did svn up it went away
21:22:59gevaertsah, yes. The quickscreen one...
21:23:03 Quit __lifeless (Read error: 60 (Operation timed out))
21:23:09kugeland I don't have a record of the local changes anymore :/
21:23:15Lloreangevaerts: I suspect every time someone changes file view from "supported" to "playlists" or "folders" it's because they tried to leave the quickscreen by pressing menu again.
21:23:18gevaertsWell, menu is always harmless on its own :)
21:23:39gevaertsbut yes, you have a point there
21:24:23LloreanPerhaps the manual entry should include the suggestion "In most screens, the safest button to hold while inserting USB will be select. This will open the context menu, but perform no further actions while you hold it." or something similar
21:24:46gevaertsyes. That should avoid most problems
21:24:48bertrikkugel, ok, too bad :P I find it a bit surprising that it doesn't seem to help at all after all
21:25:02kugelme too
21:25:10kugelsvn is slower than what I had with your path
21:25:36bertrikI think I'll look into it further once we have stable high speed with MMU on ams sansas
21:26:13bertrikI was assuming that the FIFO is always enabled, but maybe it needs to be explicitly enabled somehow
21:33:22 Join PaulJam_ [0] (
21:38:46***Saving seen data "./dancer.seen"
21:39:51bertrikI see we don't use the timer in the as3525 yet, right?
21:40:29 Join pp0000 [0] (n=621cc6c1@gateway/web/cgi-irc/
21:40:40kugelbertrik: we do
21:41:00kugelat least 1, I'm not entirely sure if the 2nd one is setup
21:41:00*gevaerts doesn't entirely understand the beast USB ccharge logic
21:41:56 Quit pp0000 (Client Quit)
21:41:59 Join pp0000 [0] (n=621cc6c1@gateway/web/cgi-irc/
21:42:04gevaertsI thought it handled non-PC (i.e. pure charger) connections, but I can't see that in the code
21:43:09bertrikkugel, ah ok I see, it's handled in kernel-as3525.c
21:43:46gevaertsand that's kind of essential on sansa and ipod, where there are no other chargers usually
21:44:49tmztthat's just vbus detect though?
21:45:24Lloreangevaerts: Pure chargers used to be handled on iPods at least.
21:45:39gevaertstmzt: not really. To do it properly (which you really can't, but never mind) it's vbus and no data
21:46:31tmztbut how do you know how much current you can draw with a pure charger?
21:46:40tmztis the only safe value 100mA?
21:46:40gevaertsLlorean: they are handled in the sense that it doesn't go into any sort of USB mode, but the 100mA vs 500mA bit doesn't handle it
21:46:59 Quit pp0000 ("CGI:IRC (Ping timeout)")
21:47:00LloreanPure chargers shouldn't need to attempt to draw any specific amount, right?
21:47:19gevaertstmzt: the only safe value is 100mA, yes. Being strict about that isn't going to gain any friends though
21:47:25LloreanI know that iPod chargers actually provide 1000mA
21:47:47soap(an hour late) but even if we were 1/4th the USB speed on Sansa - the fact Rockbox can see a SDHC card (and the OF can't) should be compelling enough, no?
21:47:48bertrikkugel, I was thinking about using the timer as an microsecond or 0.1 microsecond free running counter to create small delays, so we rather arbitrary NOP loops
21:47:51 Quit PaulJam (Read error: 113 (No route to host))
21:48:21kugelbertrik: I think we can do that
21:48:22tmztgevaerts: I have a usb charger capability on my laptop, even in suspend or poweroff (depending on bios setting)
21:48:40kugelthat's done on several other targets too
21:48:41tmztgevaerts: I have no idea what the max current it can handle is though
21:48:56gevaertssoap: well, at 1/4th the speed I wouldn't be sure, but we're at 85% on write and 140% on read, so that's fine
21:50:40gevaertstmzt, Llorean: I suspect that it's safe to assume that if we detect vbus and we don't detect any bus activity for a few seconds, it's a charger and you can grab whatever you need
21:51:05gevaertsSo the logic we need is this "wait a few seconds and see if anything happens"
21:51:19kugelbertrik: IIRC, TIMER2 is used for the tick tasks, and TIMER1 somewhere in timer.c, but I don't know whether it's actually used there
21:51:44soapWhat /would/ the consequences be grabbing too much? My brain isn't set up for thinking about the 400ma difference as dangerous.
21:51:45gevaertsAnyway, if we can control precisely, I wouldn't go over 500mA in any case. I'm not sure if we can do that however
21:53:11gevaertssoap: what if you have 4 ipods on a bus powered hub? That's suddenly 1600mA too much, and that is the case the USB limits are guarding against
21:55:31 Join saratoga [0] (i=9803c6dd@rockbox/developer/saratoga)
21:55:46gevaertsThe problem we have left (after implementing everything as well as we can, which I'll try next week) is that you either don't support "USB" chargers, or you assume that anything that doesn't talk on the bus is a charger. That assumption will be false every now and then, as in tmzt's BIOS setting case, but I'd expect those cases to be able to handle 500mA anyway
21:56:36saratogaits surprising theres no way to detect a device thats failing to deliever enough current
21:56:46saratogayou would think that would be a neccesary safety precaution
21:57:00n1sgevaerts: does the spec say what to do if a device tries to draw too much?
21:57:31gevaertsn1s: yes. You're allowed to cut it off. Unfortunately the electronics needed are not mandatory
21:58:04n1sah, that's unfortunate
21:58:48 Quit kugel ("exit(0);")
22:00:23 Join {phoenix} [0] (
22:01:41 Join PaulJam [0] (
22:02:57 Quit jhulst ("Read error: EOF")
22:04:37FlynDicekugel: Do you know where I can check if the backlight is on for the AMSSansas? I tried to use "bool lcd_active(void)" but that doesn't seem to work.
22:05:28bertrikkugel, indeed TIMER1 is used in timer.c as a general timer and TIMER2 in kernel-as3525 to do call_tick_tasks ...
22:06:33 Quit PaulJam_ (Read error: 60 (Operation timed out))
22:08:39bertrikhm, this timer stuff is confusing
22:12:21 Join Mac [0] (n=c3cc377a@gateway/web/cgi-irc/
22:14:09 Quit Mac (Client Quit)
22:15:31 Quit martian67 (Remote closed the connection)
22:16:02 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
22:18:42 Quit FlynDice (Remote closed the connection)
22:21:13 Join PaulJam_ [0] (
22:21:42 Join BdN3504 [0] (n=55b22781@gateway/web/cgi-irc/
22:23:06 Quit Bagder (Read error: 110 (Connection timed out))
22:23:36BdN3504who comitted the new pictureflow playback capability?
22:24:49BdN3504Whoever it was: I want to thank you! This is one of the sexiest features of Rockbox!
22:27:14BdN3504Vielen Dank Thomas du bist der Beste!
22:28:09 Quit PaulJam (Read error: 60 (Operation timed out))
22:31:12 Quit BdN3504 ("CGI:IRC (EOF)")
22:35:49 Quit thegeek (Read error: 104 (Connection reset by peer))
22:36:00 Join thegeek [0] (
22:36:04*bluebrother doesn't understand why a feature can be sexy at all
22:46:31 Join beta_ [0] (
23:05:43 Join Thundercloud [0] (
23:11:38 Quit moos ("Rockbox rules the DAP world")
23:17:20 Join lee321987 [0] (
23:17:36 Quit lee321987 (Client Quit)
23:25:54 Nick ved_ is now known as vedlith (
23:26:01 Join safetydan [0] (n=deverton@rockbox/developer/safetydan)
23:28:48 Join wincent [0] (
23:33:45 Quit bluebrother (Read error: 113 (No route to host))
23:38:48***Saving seen data "./dancer.seen"
23:45:17 Join Res1 [0] (
23:45:34 Quit n1s ("Lämnar")
23:46:08Res1hello; does anyone know if there are windows drivers for rockbox available? When I plug in the USB cable to my ipod and hold the menu button, windows IDs my ipod as a rockbox device; is there anything special about this?
23:47:09Mikachuwhat version of rockbox do you have?
23:48:35Res1the latest stable version (3.2-090323)
23:49:37Mikachui'm not sure, but i don't think that version supports disk mode from rockbox
23:54:52UnhelpfulMikachu: i'm pretty sure it doesn't, the portalplayer USB driver still had some issues that were not worked out yet, so it was disabled for the release.
23:55:42Mikachuit seems to work okay for me with the dailies, but maybe i'm slowly corrupting my disk
23:56:13MikachuRes1: for now just use the disk mode or apple firmware i guess
23:56:17UnhelpfulMikachu: i believe that issue has since been resolved.
23:56:28Res1Well it is good to know this is in development, but I've never needed to use it I suppose
23:56:32Mikachuyeah, from what i understand the only issue is charging
23:56:51MikachuRes1: if you install a daily, you'll be able to control your pc audio player with your ipod ;)
23:56:57Res1BTW, I just tested the recording feature, way cool :)
23:57:13*Mikachu doesn't have a dock or line-in cable
23:57:37Res1I am using a 4g U2 ipod, I can use my headphones as a mic.
23:58:51Mikachunanos can't

Previous day | Next day