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

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

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

#rockbox log for 2009-07-05

00:00:21Zagor$revision-id.zip is pretty fixed isn't it?
00:00:42Bagderwell, for that revision yes
00:00:48Zagorah you mean the last?
00:00:52Bagdertoday the urls are fixed for _all_ revisions
00:01:11Bagderyes, so that we can point to a url for the latest build
00:01:17Bagderno matter which revision it happens to be
00:01:18Zagorah, we don't host more than the latest build
00:01:24pixelmabluebrother: we also wondered if it would be possible to have checks like in the code for LCD_WIDTH > 128 etc. in the manual to use for \opt or \nopt
00:01:26Bagderexactly
00:01:27Zagorwhy would we... reverting :)
00:01:53 Quit mt (Remote closed the connection)
00:02:08Zagorrasher: that sounds reasonable
00:02:13pixelmabluebrother: and how to get the LCD_WIDTH from the c file then
00:03:30Bagderso is there then a grace period after each commit?
00:03:37Bagderthat could be... annoying
00:04:16ZagorBagder: do you rather think every commit should be queued and built?
00:04:17rasherBagder: That's how it is now, isn't it?
00:04:36Bagderrasher: no, now there's just a time period between the checks
00:05:00rasherBagder: That almost amounts to the same thing.
00:05:05rasherWith some fluctuation
00:05:05Mikachuand if there's a lot of commits during a build, those all get grouped together (i assume)
00:05:06BagderZagor: no, that's not very good either
00:05:11Bagderrasher: I disagree
00:05:12bluebrotherwell, I bet it's possible though I don't think we have someone around who could do that. Had the idea of using the preprocessor for the manual as we need it for features.txt anyway, but haven't thought about details yet.
00:05:28rasherBagder: How long is there between checks now?
00:05:49Bagderthere's a sleep 60 between each check, and after a build round
00:05:54Mikachuwouldn't it be pretty simple to trigger it with an svn hook?
00:06:02Bagderyes it will
00:06:08Mikachusorry, wrong tense :)
00:06:14rasherBagder: So after a commit, it'll be 0-60 seconds before the build starts, in which time more commits may happen
00:06:23Bagderyes, but max 60 seconds no more
00:06:30Bagderwhatever happens
00:06:40Bagderif it started in idle
00:06:42rasherI was under the impression that it waited after a commit
00:06:47rasherTo see if more were coming
00:06:49Bagderit doesn't
00:06:55Mikachumaybe you could check if the commit is by the same person, then it's likely a hotfix commit?
00:07:39rasherMaybe it could just start instantly, and then check after 60 seconds if there had been more commits, and restart?
00:07:42Mikachudo the clients run svn up with a specific revision btw? otherwise there could be weirdness
00:07:51Bagderyes they do
00:07:58 Join homielowe [0] (n=homielow@unaffiliated/homielowe)
00:08:01rasherThat would have the same effect as the current system, but shave 0-60 seconds off the time from commit to the builds are ready
00:08:03Bagderthey get told exactly what rev to build
00:08:27Bagderrasher: except the time when a few people commit with 40 seconds interval
00:08:36 Join AndyI [0] (i=AndyI@212.14.205.32)
00:08:37rasherBagder: No
00:08:48Bagderyou said it would restart then
00:08:50rasherBagder: It should only re-check 60 seconds after the build that set the ball rolling
00:08:56 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
00:09:00Bagderafter the first commit then
00:09:02rasherYes
00:09:12Bagderright, that's what I asked first ;-)
00:09:23rasherI changed my mind halfway through
00:09:44Bagderthe only little nit against this concept is that its less simple
00:10:12rasherSo what I'm saying now: Start instantly on commit. 60 seconds after *that* commit, see if there are more commits, and restart with that rev.
00:10:19rasherThis is true.
00:10:59kugelwhy not just start on any commit (with the post-commit hook) (restart of a build round is going)?
00:11:26Bagderso if we're in a commit craze it'll just not complete a round?
00:11:29rasherkugel: because that could end up postponing the result if people keep committing every 59 seconds
00:11:35gevaertsWhy not just try to get as many clients as possible so the round is done in 60 seconds? ;)
00:11:45*AlexP preps his atom
00:12:05Bagderbecause we can still only hand out full builds
00:12:11Mikachudoes arm-gcc run on rockbox yet? everyone could hook up their daps to the build system!
00:12:21 Join Hillshum [0] (n=chatzill@unaffiliated/hillshum)
00:12:31gevaertsMikachu: we want to make the builds *faster*!
00:12:34Zagorrasher: is that really a big problem? we rarely get long bursts of every-minute commits
00:12:36Mikachu;)
00:13:00kugelI actually can't remember such a rage, except funman's recently
00:13:05rasherZagor: It might not be worth it if it's much more complicated
00:13:09kugelbut that was 5s between each commit
00:13:20BagderZagor: but honestly we also rarely get "follow-up" commits just afterwards that are strictly related
00:13:32ZagorI think "restart if new commit within 60 seconds of previous" is the simplest way
00:13:41kugelWe tell people to watch the build table after each commit, so it will calm down automagically :P
00:13:43BagderI think not
00:13:48*gevaerts would just let the build complete
00:13:51Bagderme too
00:13:55ZagorBagder: silly typos hopefully come in within 60 seconds
00:14:03BagderI still don't like that
00:14:04Zagorgevaerts: so what about the new commit?
00:14:26gevaertsZagor: it gets scheduled after the current build is done
00:14:33Bagderafter the build round, if there are queud builds, build the newest one
00:14:39Zagorso we build each and every commit then
00:14:45BagderI think not
00:14:47ZagorBagder: ah ok
00:14:56kugelgevaerts: so that we have to wait 30min after a commit rage?
00:14:57gevaertsFor two-commit bursts, yes. For longer bursts, no
00:15:18Zagorwe only queue one build, so a new commit during a build replaces the already queued one. sounds fair to me.
00:15:18gevaertskugel: no. I'd just do the new current revision
00:15:19rasherkugel: hopefully the build system should be done faster than 15 minutes
00:15:31*gevaerts wasn't clear
00:15:32BagderZagor: a right, we can just make the queue command work that way
00:15:54Zagorqueue command? isn't that the BUILD command?
00:16:02Bagderthere's a queue one too
00:16:03Zagor(from commander to server)
00:16:13Bagderalthough they could probably be one
00:16:24ZagorI think so, yes
00:17:42gevaertsWhat I'd do is something like let the post-commit hook set a flag and if no builds are running, start a build. At the end of the build, see if any new flags have been set, and if so, start another build with the new latest revision
00:18:49Zagorgevaerts: well the hook sends the revision number. so instead of a flag we simply keep/queue the latest revision.
00:19:37Bagderyeah, so if no buildround is active, it'll build immediately, otherwise keep the rev number for after this round
00:19:43gevaertsZagor: if it looks the same from the outside, it's the same to me :)
00:19:52Zagor:)
00:20:01ZagorBagder: exactly
00:20:35*gevaerts would probably implement something horrible with a directory that has files in it named after revisions...
00:20:39rasherI just thought it was a bit wasteful to potentially wait 60 seconds before building
00:20:55kugelSlasheri: ping
00:21:03 Quit petur ("Zzzzzz")
00:21:10gevaertsme too. I think we should optimise for the single correct commit case. That's most of them
00:21:48Slasherikugel: o/
00:22:21kugelSlasheri: can dircache smash the database?
00:22:21Bagderrasher: this system won't wait a single second
00:22:41Slasherikugel: hmm, what do you mean with that?
00:22:51kugelI have a very annoying problem that I don't have on my fuze (which doesn't have dircache). If I boot without µSD, the database is almost empty
00:23:31kugelmost of my songs are on it, but I regularly exchange it between my fuze and e200. The Fuze's db stays intact without µSD, the e200's not
00:23:38rasherBagder: Great - so my suggestion was just to keep the same feature of possibly grouping commits in the first build round. But if that's not a requirement, no need to complicate things
00:24:14Slasherikugel: hmm, interesting. running the "update now" manually doesn't fix it?
00:24:34pixelmakugel: is auto-update enabled too?
00:24:39kugelit does, but it's annoying
00:24:46kugelpixelma: no
00:25:14kugelI expect the database to not change at all without auto-update, hence I don't want to enable it
00:25:38Slasherikugel: hmm, it shouldn't change
00:26:40kugelit does, and I suspected dircache. I haven't tried disabling it yet, though
00:27:13Slasheriah, yes. Do you load db to ram?
00:27:19kugelyes
00:27:22Slasheriin that case it does change indeed
00:27:41CIA-71New commit by zagor (r21639): Clear upload dir after build round ends.
00:27:54Slasherithe ramloader checks with dircache (if that's available) whether those entries still exists
00:28:05Slasheriand if not, those are removed from the db
00:28:46 Quit AndyIL (Read error: 113 (No route to host))
00:28:49Slasherihmm.. maybe the loader should be fixed by a check if auto-update is enabled
00:28:56kugelI consider this as a bug, actually
00:28:57 Join tvelocity [0] (n=tony@adsl3-207.her.forthnet.gr)
00:29:01Slasheriif not, in that case it would not remove those entries
00:29:08kugelthat sounds good
00:29:21Slasherithat should be easy to do
00:29:38kugelI'll gladly test it
00:29:54Slasherii will give it a try - soon :)
00:30:24saratogaSlasheri: about my previous question, where should I look in the source if I want to hack up how microsd cards are mounted in the file browser?
00:30:36kugeleh, I know what soon means in your "timezone" :P Can you point me to the code?
00:31:17Slasherisaratoga: hmm, in that case maybe the file browser itself? the tree/filetree.c
00:33:02Slasherikugel: hmm.. weird, in fact there is a check for the autoupdate in the code
00:33:14Slasheritagcache.c:3986
00:33:53Slasheriah, no, not for dircache
00:34:09Slasherijust adding the check for the upper block too should be enough :)
00:34:41kugelit's definitely off here
00:35:10saratogaSlasheri: would it be better to do it in the file browser or at a lower level by faking one giant disk that spans two volumes?
00:35:22kugelthe whole if..else.. should be within that, correct?
00:35:24saratogai don't understand how rockbox works yet
00:35:55Slasherikugel: in fact, no
00:36:29kugelbecause of the 2 extra lines in the dircache block?
00:36:31Slasherikugel: normally you want to load the dircache pointer to the tag entry. But if the file doesn't exist, just skip the loading and delete
00:36:41Slasherikugel: so you have to refactor the ifs a bit
00:38:39Slasherikugel: well, just skip the delete_entry() unless auto-update is enabled
00:38:42Slasherithat should be enough
00:38:47pixelmabluebrother: how do I get the file info into the first line of a new tex file? "% $Id : Revision Date Author $ (if I "translate" the filled out ones back)?
00:38:53kugelalright
00:40:08pixelmaor would
00:40:14rasherpixelma: Just adding % $Id$ and do svn propset svn:keywords Id filename.tex
00:40:22kugelI think $Id:$ is enough
00:40:32pixelmaok
00:41:19ZagorBagder: there seems to be something not initialized right in startround(). the second BUILD round doesn't get distributed as the first.
00:42:17Bagdersounds like a bug then
00:42:42Zagoryeah. "5 builds not complete, 15 clients. 2 builds in progress"
00:43:17Bagder!
00:44:35CIA-71New commit by zagor (r21640): Queue last build request recieved during a build run.
00:45:20bluebrotherpixelma: you need the svn property to at least include the value "Id". Usually the values "Author Date Id Revision" are set. And you can leave in other stuff (like old value when copying the line from another file), everthing between the $'s gets replaced
00:48:20ZagorBagder: shouldn't $client{$_}{'building'} be decreased in kill_child()?
00:49:03Zagoror in _CANCEL perhaps
00:49:37Bagderoh indeed
00:49:51Zagoradding it to _CANCEL
00:50:32rasherIs http://rockbox.pastebin.com/m304d3a76 the idea with that?
00:51:20 Join bluebrother^ [0] (n=Dom@g224237222.adsl.alicedsl.de)
00:51:55 Quit Hillshum (Read error: 60 (Operation timed out))
00:52:24***Saving seen data "./dancer.seen"
00:52:43Unhelpfulkugel: re exit(1)... i think there ought to be an "exit" that converts all non-zero into PLUGIN_ERROR and zero into PLUGIN_OK, and a rb_exit or similar that passes the value exactly as-is. the latter is still of value to avoid passing return values up through levels, but is better for things that are not pure ports.
00:53:28rasherZagor, Bagder: that was directed at you two, sorry (http://rockbox.pastebin.com/m304d3a76)
00:53:34kugelUnhelpful: I thought so too
00:53:55Bagderrasher: yes, that looks like what we've intended
00:54:01Zagorrasher: yes. feel free to commit.
00:54:05 Quit ender` (" Marriage is not a word, it's a sentence. A life sentence...")
00:54:31kugelWhy not just print $cpu?
00:54:31rasherI don't know what values are possible for that, but hopefully it should be easy to fix
00:55:12*rasher does not question Bagder and zagor
00:55:16Bagderkugel: that wouldn't really help anything as then the logic would instead be put in the server
00:55:19kugeland why is >64 and <32 not supported? I planned to attach a emulated 8086...
00:55:32rasherI'd personally have put the logic in the server
00:55:37rasherBut meh
00:55:47rasherIt's already there, so
00:55:54 Quit bluebrother (Nick collision from services.)
00:55:58CIA-71New commit by zagor (r21641): Decrese {building} for cancelled builds. ...
00:55:59 Nick bluebrother^ is now known as bluebrother (n=Dom@rockbox/developer/bluebrother)
00:56:13kugelBagder: is the HELLO message parsed by the server or is that only for printing?
00:56:23CIA-71New commit by rasher (r21642): Detect 32/64-bit targets and report (and exit with an error if the CPU is unknown).
00:56:31Bagderfully parsed and dealt with
00:56:32Zagorkugel: very much used by the server
00:56:52bluebrotherkugel: your 8 bit build server would be too slow so there is no point in that. Get a real CPU!
00:56:57 Nick linuxstb is now known as _____stb (n=linuxstb@rockbox/developer/linuxstb)
00:57:11kugelahm, I see the $sock now
00:57:16Unhelpfulwhat about a 65816? :P
00:57:33kugelbluebrother: 8086 is 16bit
00:57:37bluebrother6502 of course!
00:58:16bluebrotherkugel: even worse! 16 bit is *total* crap
00:58:17Unhelpfulbluebrother: that's only 8-bit :)
00:58:31bluebrotherwho wants to use 16 bit at all?
00:59:08*kugel hopes bluebrother noticed that kugel was kidding
00:59:41Unhelpfulbluebrother: you can do quite a lot only being able to count to 65535. :)
01:00
01:00:02bluebrotherUnhelpful: as in getting headaches because you don't have enough RAM :(
01:00:25kugelSlasheri: that works, I'm going to commit it
01:00:31Unhelpfulbank-switching, boyeeeeee :)
01:00:48kugelor do some overlay
01:01:03 Quit DarkDefender (Remote closed the connection)
01:01:07kugelthis is getting OT
01:01:12Slasherikugel: great :)
01:01:36kugelalso, that's going to be my first *intentional* git commit :)
01:01:36 Join robin0800 [0] (n=robin080@cpc3-brig8-0-0-cust436.brig.cable.ntl.com)
01:03:04saratogado we still have .map files available for svn builds?
01:03:51*gevaerts still wants the .elf files as well...
01:05:13saratogais there a script that figures out where an address is in rockbox?
01:05:57CIA-71New commit by Ubuntuxer (r21643): FS #10291: use lib highscore.h and add a new highscore table
01:06:26 Nick _____stb is now known as linuxstb (n=linuxstb@rockbox/developer/linuxstb)
01:07:37Zagorsaratoga: grep? ;)
01:08:08saratogaZagor: i thought we had a perl script for it somewhere in svn
01:09:22Zagoroh. If we do, I have missed it.
01:09:33saratogahmm no it only finds asm at an address
01:14:55saratogawe only call cpuflush on codec load if theres more then 1 CPU
01:15:13saratogais that correct for a CPU like ARM with seperate instruction and data cache?
01:15:49saratogai would expect that the instruction cache needs to be flushed if we modify code
01:17:07 Quit Ubuntuxer ("Leaving.")
01:18:34saratogaand google confirms that it does need to be flushed
01:21:17CIA-71New commit by kugel (r21644): Do not delete tagcache entries on bootup with dircache enabled but auto-update disabled. ...
01:22:38 Quit barrywardell ()
01:22:44 Join perrikwp [0] (n=Keith@74.167.148.160)
01:23:28kugelsaratoga: there are maps for daily builds
01:23:43kugelsorry, "archived" builds
01:23:43saratogakugel: is there anyway to flush cache for ams in svn right now?
01:23:50 Nick fxb is now known as fxb__ (n=felixbru@h1252615.stratoserver.net)
01:24:28kugelsaratoga: yes, the same way as for other targets using caches
01:24:43saratogado you know the function?
01:25:27kugelcpucache_flush
01:25:57saratogathats only defined in PP50*.c files as far as I can tell
01:26:19saratogaguess i could just do a mcr 15, 0, r0, c7, c5, 0 in an ASM block
01:26:30kugelsaratoga: that's defined in mmu-arm.S
01:26:48kugelit's an alias for clean_dcache()
01:27:45kugelalthough there's only the prototype for clean_dcache() in mmu-arm.h, so you better try that
01:28:11saratogawell lets see how well i remember my arm asm
01:28:21 Quit robin0800 ("Leaving")
01:28:39 Join robin0800 [0] (n=robin080@cpc3-brig8-0-0-cust436.brig.cable.ntl.com)
01:28:40 Join barrywardell [0] (n=barrywar@86-43-167-232-dynamic.b-ras2.prp.dublin.eircom.net)
01:30:16kugelsaratoga: just use clean_dcache(), no need for asm
01:30:57saratogakugel: ASM fixed it anyway
01:31:04kugelfixed what?
01:31:07 Quit robin0800 (Client Quit)
01:31:10saratogathe crash on codec load
01:31:22kugelwhere did you insert it?
01:31:30saratogaits funny i was reading an arm9 textbook yesterday and they mentioned self modifiying code and I thought "huh I'm pretty sure we don't deal with that properly when loading codecs I wonder why it doesn't matter for rockbox"
01:31:50saratogabefore the call to codec_main in codec_crt0
01:32:07kugelhaha, that's nice
01:32:14Zagorgood catch!
01:32:27kugelsaratoga: seems sensible, we're doing that in plugin_load too
01:32:57 Quit bluebrother ("Leaving")
01:34:20saratogakugel: there is no clean_icache() function that I can see
01:34:29 Join donutman25 [0] (n=Dagni@65.75.86.64)
01:34:31saratogais it called something else
01:34:34kugelno, but clean_idcache
01:34:40kugelthe very last function
01:35:08kugeli.e. flush both
01:35:30kugelI'm not sure what the difference between "invalidate" and "clean" is, both are doing the writeback
01:35:44saratogashould be the same
01:35:50saratogawhere is this defined? grep is finding nothing
01:36:08kugelinvalidate_idcache, sorry
01:36:15kugelthe last function in mmu-arm.S
01:37:20kugelsaratoga: btw, is this really a data abort, or rather undefined instruction?
01:38:32saratogakugel: try playing an mp3 followed by another format and see for yourself :)
01:38:37saratogabut probably undefined instruction
01:39:28kugelhm
01:39:40kugelI actually got a prefetch abort at skipping back from the ogg
01:40:04saratogado we really want to flush the dcache? it should actually be coherent at this case
01:40:07saratogaat this point
01:40:34saratogaand why does this not cause problems on the gigabeat F
01:40:39saratogaits ARM9 too . . .
01:41:13kugelsaratoga: I'm not sure. we're switching codecs
01:41:35kugelw/o flushing, the new might use the cached data from the old, right?
01:41:41saratogathe switching is handled by the CPU, so any data it caches is still correct
01:42:09saratogaits only a problem for the icache because write backs from the dcache don't update teh icache
01:42:21saratogaso if data is cached in both the icache will be wrong following a write
01:42:41saratoga[buying this arm arch book was a great idea!]
01:43:45kugelI don't think flushing the dcache also is much of an issue
01:44:37kugelbut if you're sure you can try. you should create a function in mmu-arm.S though
01:45:26saratogaok
01:46:27kugelsaratoga: the file is splitted in routines for arm11 and lower though. take care of that
01:47:48saratogakugel: it looks like the beast takes care of its own mmu stuff
01:47:56saratogabut i don't understand why this wasn't a problem on the F
01:48:14kugelthey have the same function names, but different code
01:48:21saratogaalso is there some reason not to make this a #define ?
01:49:34kugelI don't think so. It seems it's the same for imx anyway
01:50:40saratogai don't understand the asm here
01:50:47saratogawhat does mov r0, #0 do at teh start of the function?
01:50:58 Part donutman25
01:51:02kugelyou won't be able to add it to the plugin/codec api then, though
01:51:12CIA-71New commit by pixelma (r21645): Start of an Iaudio M3 manual: add the necessary platform files (name the remote keymap file 'iaudio' because the keymap is shared with M5 and X5 and ...
01:51:20 Join notlistening [0] (n=tom@94-195-105-95.zone9.bethere.co.uk)
01:51:28kugelsaratoga: which function?
01:51:29saratogaoh they're just preloading a constant before the coprocessor commands
01:52:41notlisteningcan a e200v2 boot rockbox from the SD card?
01:53:13kugelwith some source code changes, yes
01:54:10 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
01:54:16kugeli.e. change BOOTDIR in config-e200v2.h, and pass −−rbdir=/<micrsd1>/.rockbox to configure
01:54:40notlisteninghumm i will try just to see what can it hurt ;)
01:54:47saratogakugel: this look right to you? my asm sucks http://pastebin.com/m1f8cc43d
01:55:16saratogait should work on arm11 as well so i'm going to put it outside the ifdef
01:55:32kugelI you should make sure r0 is 0 imo
01:56:10saratogaoh good catch
01:56:31saratogawhich header does this get added to?
01:57:31pixelmaugh, I think I put the labels the r
01:57:34pixelmawr
01:57:49pixelmawrong way... sorry
01:58:08kugelI'm not sure why invalidate_dcache does the "mov r1, lr" ... "mov pc, r1" stuff though
01:58:52kugelbl is saving the return address on the stack, IIUC
02:00
02:00:34saratogai guess they don't want to touch the stack during the flush?
02:01:27kugelmaybe
02:01:34Unhelpfulkugel: i thought that bl saved the return to lr?
02:03:32kugelyes, but I meant the previous value of lr. But I'm not too sure
02:03:54saratogashould i add this function to the codec api or is there some other way to get to it?
02:04:46kugelmake it a macro, and let cpp insert it directly, or compile mmu-arm.S in
02:05:10kugelbut adding it to the codec api is probably the cleanest
02:05:50saratogayou think i should add this for the gigabeat F as well?
02:06:21 Quit Thundercloud (Remote closed the connection)
02:06:38kugelit's added automatically
02:07:00kugelthe F uses the part you're editting. The beast uses the upper part of the file
02:07:55saratogakugel: i mean should I add the call to flush on codec load
02:08:16kugelyes
02:08:34kugelah you were about to #if CONFIG_CPU == AS3525?
02:09:03saratogahaha found my F, still running r185xx
02:09:08kugeli'd just put #ifdef HAVE_CPUCACHE_INVALIDATE around, that's defined in mmu-arm.h
02:09:33kugelsaratoga: you could try to reproduce it on it
02:09:41saratogathats what i'm doing
02:09:52kugelI'm curious too why it never hit F/X
02:10:42 Quit mcuelenaere ("Gnight")
02:11:05saratogakugel: the F does not have this problem
02:11:45kugelI can't really reproduce it on my fuze
02:12:09kugelI'm playing a mp3, ffwd to ~10% before the end, then press next
02:12:25kugelmy playlist is 1mp3, then 12 or so oggs
02:12:43saratogai did it a few times on my e200v2
02:13:01saratogai'll try the fuze
02:13:21kugelit should be added for all targets with icache nevertheless, I think
02:15:27saratogakugel: it happens on my fuze
02:17:50CIA-71New commit by zagor (r21646): Rename builds logs $rev-$id.log for keeping.
02:18:54 Quit HellDragon (Read error: 54 (Connection reset by peer))
02:20:02 Join HellDragon [0] (i=jd@modemcable178.248-201-24.mc.videotron.ca)
02:22:10kugelsaratoga: I fail at reproducing
02:25:39saratogakugel: i used the current svn build switching from vorbis to mp3 and got it
02:25:52saratogafrom the website
02:26:45kugelI get a prefetch abort sometims
02:26:48kugelsometimes*
02:26:49 Quit Zagor ("Clint excited")
02:27:03kugelat a very weird address (0x849.....)
02:27:05saratogathats what i see
02:27:58kugelok
02:34:33Unhelpfulkugel: ok, why is RFAC a test case for exit-to-wps?
02:34:59kugelnot a test case
02:35:15Unhelpfuli suppose i mean, why would i want to go straight to WPS from it?
02:35:26kugelthe recently added "Play shuffled" puts all folders into signle playlist
02:35:40kugeland starts playing it
02:36:04kugelbut it apparently needs to exit for doing this. Exiting to the wps makes more sense than exiting to the main menu
02:36:31Unhelpfulalso, this patch seems to contain playback controls for PF as well?
02:36:37gevaertsit doesn't really *have* to exit, but if it doesn't, it has to reset a lot of internal state
02:37:11kugelUnhelpful: it just "imports" the std wps button. if it's free, it's used. If it's not free, then not
02:37:12Unhelpfulor... just a stop button?
02:37:23gevaertsRFAC grabs the audio buffer, so in order to keep running (and doing something useful) it would have to stop playback again
02:37:32kugelstop? I have you the wrong diff then
02:37:44Unhelpfulhttp://pastie.org/531932 <- this one
02:40:05Unhelpfulit has RFAC changes for exit-to-wps, properties, pictureflow... i must've misread the PF_STOP stuff.
02:40:06kugelUnhelpful: thats the latest one anyway: http://pastie.org/534481
02:40:28kugelyea, that was an old one
02:40:36Unhelpfulwhy would you want to exit to WPS from properties?
02:40:50rasherUnhelpful: because you can press the "go to wps" button
02:41:07rasherWhich is expected to work, since the properties plugin blends into the menu structure
02:41:14kugelin the core
02:41:19*rasher asked these same questions a few hours ago
02:41:25saratogaso arm-mmu isn't even available in apps right now
02:41:25notlisteningjust updated my svn and i am getting error on make full\ip : can't open database.ignore at /home/tom/rockbox/tools/buildzip.pl line 215.
02:41:33saratogais it ok to add it to system.h?
02:41:34Unhelpful...ah! in other words, it uses the new plugin feature to enable a button that usually fails (strangely) in the properties screen, which pretends not to be a plugin.
02:41:52kugelsaratoga: it should. plugin_load calls cpucache functions
02:42:00rashernotlistening: did you run make first?
02:42:16kugelUnhelpful: exactly
02:42:23notlisteningyeah make -j ;P
02:42:59rashernotlistening: Erm, I'd just try blasting the build dir (rm -rf *) and doing make && make fullzip
02:43:04Unhelpfulkugel: the only thing i'd say ought to change is the bit we talked about before, about exit() vs an rb_exit()
02:43:05rasherSounds like something went wrong somewhere
02:43:21kugelUnhelpful: it doesn't include any exit() business anymore
02:43:27saratogakugel: it calls cpucache_invalidate which is actually defined in one of the PP system files
02:43:42notlisteningCC apps/plugins/mpegplayer/header.c has some a whole load warnings ;)
02:43:51Unhelpfulkugel: good deal then. also, consider putting your work up on repo.or.cz or such. ;)
02:44:02kugelsaratoga: weird. it's the same function, though
02:44:12saratogabut i guess aliased onto invalidate_idcache too
02:44:25rashernotlistening: Is this a clean build? And are you using "our" compilers?
02:44:28saratogamaybe I should just use that function and accept that we'll waste a flush
02:44:30 Quit linuxstb (Remote closed the connection)
02:44:31rashernotlistening: A clean sourcetree, I mean
02:45:05notlisteninghumm not so clean but almost sorry my mistake is it the boom pligin causing the problems
02:45:32rasherNot quite sure what you're saying
02:45:43Unhelpfuli don't think there's a "boom" plugin?
02:45:50kugelsaratoga: PP has it's own set of functions, it doesn't compile mmu-arm.S
02:45:51notlistening/home/tom/rockbox/apps/plugins/doom/rockdoom.c:216: warning: (near initialization for ‘wads_builtin[6]’)
02:46:03kugelsaratoga: yes it is aliased
02:46:11notlisteninglol wrong key;)
02:46:30saratogaugh and HAVE_CPUCACHE_FLUSH is actually defined in somewhere else in some PP code
02:46:39saratogawhat a mess
02:47:28kugelthe integration of PP into the target tree is indeed a mess
02:47:35saratogawho should I talk to about the target tree defines?
02:47:41saratogai'm clueless as to how this should be setup
02:47:58notlisteningthere a whole load. I have been trying to set the microsd from the e200-v2 to be the boot device
02:48:04 Join Far^Side [0] (n=farside@88.84.187.18)
02:48:11Unhelpfultry a make veryclean first, perhaps?
02:48:24 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
02:48:29saratogaif you're going to mess with configure you probably want to do a fresh folder
02:48:49kugelsaratoga: just add a stub for PP, and use it with #ifdef HAVE_CPUCACHE_INVALIDATE (or flush=
02:48:52kugel)
02:49:35notlisteningGood idea will co again
02:50:00Unhelpfulno no not co again
02:50:06Unhelpfula fresh build folder.
02:50:20saratogakugel: thats not defined for anything but PP in /apps
02:50:39kugelwhat? It's defined in mmu-arm.h
02:50:46saratogawhich isn't included in apps
02:51:01Far^Sideis my iPod 5.5G dead? My iPod suddenly started showing the sad face icon after being left in the sun for a short time. I have an 8 GiB CF-card in it, an that card is working when removed from the iPod. Tried starting it in recovery mode, but it won't show up in iTunes and it restarts shortly afterwards
02:51:14kugelhm. then include maybe in cpu.h
02:51:34kugelor just directly, which is cleaner if you're going to add this functions directly to the api
02:51:46saratogaall this to include one line of asm :)
02:51:57kugelI don't see why you shouldn't include mmu-arm.h in apps
02:52:02kugelheh
02:52:26***Saving seen data "./dancer.seen"
02:53:21notlisteningright will try that first Unhelpful
02:53:24saratogakugel: maybe as3525.h should include arm-mmu.h?
02:53:33saratogasince it uses the arm mmu
02:54:06saratogathat way we don't need stubs for arm7 targets with no arm mmu
02:55:09kugelprobably
02:59:25notlisteningUnhelpful, same warnings & error message on make fullzip
03:00
03:01:27Unhelpful...but not on just "make"?
03:01:57kugelsaratoga: I think I see what the problem is
03:01:59rashernotlistening: You need to run "make" first, then "make fullzip"
03:02:52kugelsaratoga: F/X include mmu-arm.h in system-target.h, which is in system.h
03:02:52 Quit Lss (Read error: 104 (Connection reset by peer))
03:02:57kugelwe don't do that
03:03:25saratogakugel: yeah but i dont' see where they flush the cache?
03:03:31kugelthat way we have a stub for cpucache_invalidate() in system.h instead of the actual function
03:03:33notlisteningi am running make -j first then make fullzip
03:03:45kugelapps/ never calls cpucache_invalidate for AMS
03:03:55rashernotlistening: And does make -j complete succesfully?
03:04:02saratogabut it does for F/X?
03:04:03notlisteningi have been making rockbox before
03:04:10kugelcpucache_invalidate is a codec api member
03:04:11Unhelpfulnotlistening: so, "make" is completely successful, but "make fullzip" produces errors?
03:04:19notlisteningyeah it does just a few warnings as i said
03:04:27notlisteningthat right
03:04:29rashernotlistening: that's not completely succesful.
03:04:33notlisteningsprry if i am being unclear
03:04:36saratogakugel: yeah but only for PP
03:04:38kugelsaratoga: the stub in system.h is within #ifdef HAVE_CPUCACHE_FLUSH
03:04:46rashernotlistening: Which compiler are you using?
03:04:58kugelsystem.h doesn't see that AMS has this, because we never include mmu-arm.h
03:05:05saratogayes I know that
03:05:22notlisteningthe one you guys setup for me from the script
03:05:38rashernotlistening: which target?
03:05:51saratogaoh crap does plugin.c handle loading codecs as well?
03:05:53kugelsaratoga: playback calls it, before calling some callback
03:06:20notlisteningarm sorry
03:06:28notlisteningUsing arm-elf-gcc 4.0.3 (400)
03:06:28notlisteningUsing arm-elf-ld 2.16.1
03:07:09rasherThat seems correct.
03:07:23saratogakugel: arrrgghhhh
03:07:24rasherMight -j produce errors that don't appear on -j X?
03:07:31rashernotlistening: try make veryclean and make -j 2
03:07:40saratogaso all thats really needed is to include that one damn header
03:07:49notlisteningok ;)
03:07:52kugelsaratoga: can you look if #include "mmu-target.h" in our system-target fixes it?
03:08:08kugelthat's badly documented, really
03:08:21saratogayes
03:12:03notlisteningsame error :P
03:12:26 Part n00b81 ("Leaving")
03:12:29notlisteningshould i undo my changes?
03:12:29kugelnotlistening: you're probably doing it wrong
03:12:33kugelwhat exactly did you change?
03:12:52saratogakugel: yes it fixes
03:12:53notlisteningthe BOOTDIR bit
03:12:56saratogai'm going to commit it now
03:13:19kugeladd a comment to mmu-arm.h also please
03:13:35kugelsuch as "INCLUDE THIS INTO SYSTEM-TARGET.H OR YOU WILL DIE"
03:13:43saratogaok
03:13:45rashernotlistening: try svn diff in the root and post the results on a pastebin
03:14:00notlisteningok ;)
03:16:39notlisteninghttp://pastebin.com/d3979c86b
03:17:30CIA-71New commit by saratoga (r21647): ARM922T's icache isn't coherent with its dcache, so we need to ensure that its flushed before changing codecs. Playback takes care of this for us, ...
03:18:28kugeldamn, we wasted too much time on that one
03:19:17rashernotlistening: I'm fairly sure the microsd thing is going to work anyway..
03:20:20drakonikooh
03:20:24drakonikI think I found something interesting
03:20:45kugelnotlistening: you have a typo there
03:20:59rashernotlistening: er, is not going to work
03:21:07kugel"/<micrsd1>/.rockbox" should be "/<microsd1>/.rockbox", but I don't think it's causing the compiliation failure
03:21:12kugelrasher: it works, I tried it
03:21:29rasherkugel: Shows how much I know.
03:21:38drakonikSo I took the rockbox sim directory and put it onto my ipod in disk mode. I moved everything on the ipod into the simdisk folder. I ran the db scan, and eventually, the db scanned froze up. I could still back out of the menu, but the prompt window that opens along with the simulator gave an error message "Too many dirs open"
03:22:01kugelsaratoga: I'm wondering if that automagically also fixes the clip playback issues
03:22:30notlisteningI'll try changing it
03:23:04kugelnotlistening: you have quite some other changes which are more likely to break it, actually
03:24:31notlisteningyeah i know he radio bit is not important but the rest of the work is with rbutil
03:25:08notlisteningand possibly the configure script ;)
03:25:40notlisteningback to the drawing board I think
03:25:42saratogakugel: yeah i was just thinking that, but greping around I don't see anything directly related to buffering using it
03:26:11rashernotlistening: I think we're down to "Rockbox compiles fine. Your changes broke it"
03:26:21kugelit apparently happens on rebuffering. there's quite a few callbacks involved I think
03:26:24rashernotlistening: unless you can show us that a clean tree still breaks
03:27:21notlisteningok doubt that sorry for causing trouble and thanks for being patient
03:29:08kugelsaratoga: so, we basically didn't get fs corruption reports so far, did we?
03:29:57kugelI think we should figure out what to do with the voltage stuff (either revert for now, or pump the volts when accessing the microsd), and then put under supported asap
03:30:29saratogakugel: none that i know of
03:30:48saratogaand I think we should revert the voltage changes until we've gotten people who reported issues to completely test any fix
03:31:12saratogaor leave them in but disable voltage switching for now since it clearly does not work correctly
03:31:22saratogai've also noticed it now on my fuze, but it doesn't happen everytime
03:31:22kugelI wish a dev would have a problematic clip / microsd
03:32:07saratogaat least since that patch went in i sometimes get a white screen on boot with my fuze
03:32:10saratogai presume they're related
03:32:43saratogabut yeah, aside from that what bugs are left on the fuze?
03:33:19kugelI'm not aware of an open bug
03:33:46kugelI just checked my fs again. fsck still only reports problems for the files created by the OF
03:34:07 Quit BlakeJohnson86 (Read error: 104 (Connection reset by peer))
03:34:15saratogado we need to do anything special to create the canidate bootloaders then?
03:34:47kugeljust tag them I think
03:34:54 Join BlakeJohnson86 [0] (n=bjohnson@c-24-118-162-123.hsd1.mn.comcast.net)
03:35:32 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
03:35:33saratogabut the binaries don't need a version number like PP ones?
03:36:28saratogawe should probably remove the SVN revision and replace it was a version 1.0 line
03:36:29 Join Blue_Dude [0] (n=chatzill@uslec-66-255-178-43.cust.uslec.net)
03:36:48rasherSame needs to be done for the pp bootloaders.. I didn't do it for the ones in FS
03:37:18kugelsaratoga: you mean make should create a bootloader-fuze-rXXXX.sansa?
03:37:46kugelor put the version into the header?
03:37:59rasherkugel: I think he means the version string the bootloader displays
03:38:09kugelah I see
03:38:31notlisteningrasher, my bad coding there damn i will have to hunts that down
03:40:16kugelsaratoga: we just need to define APPVERSION accordingly
03:42:36 Quit Minthe ("Leaving...")
03:43:51kugelI don't quite see where that comes from currently though
03:44:28saratogayeah i'm lost too
03:44:49 Quit Horscht ("Verlassend")
03:45:43saratogait looks like its just the revision and date right now
03:46:06saratogarasher: do we just redfine it for this one build ?
03:47:18rashersaratoga: I'm not sure
03:47:42kugelI think APPVERSION comes from a script during compilation
03:47:52kugeland yes, I think we could redefine it for a release
03:47:53rasherMaybe a branch for bootloader release should be made
03:50:08 Quit notlistening ("Leaving")
03:53:07 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
03:55:09saratogarasher: that sounds good
03:55:15 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
03:55:15 Quit pixelma (Nick collision from services.)
03:55:18 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
03:55:18 Quit amiconn (Nick collision from services.)
03:55:27rasherWe can do AMS and PP at the same time, to simplify things
03:55:30 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
03:55:33rasher(in the same branch)
03:55:36 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
03:55:53kugelrasher: good idea
03:57:45kugeland mkamsboot also in the same branch
03:58:55 Quit killan ("( www.nnscript.com :: NoNameScript 4.22 :: www.esnation.com )")
03:59:26saratogasomeone needs to make mkamsboot binaries for windows
03:59:29kugelthen we can have mkamsboot also take the svn revision
03:59:47kugelsaratoga: did someone actually try that already?
03:59:54saratogayeah funman i think
04:00
04:00:04saratogai cannot get this bootloader to display what i want
04:00:57kugel#ifdef APPVERSION .. #undef APPVERSION .. #endif ... #define APPVERSION "Version 1.0" ?
04:01:27saratogai tried doing it with printfs like the PP bootlaoders
04:01:29saratogai will try that
04:01:58kugelsaratoga: version stuff is handled in show_logo.c
04:02:13saratogaPP has printfs too
04:02:32kugelyes, but the version display is in show_logo.c for all players I think
04:02:35rasherkugel: also sansapatcher, ipodpatcher
04:03:29kugelwe call that branch sansa_inferno then :)
04:03:37saratogai'll make the bootloaders then post then to FS
04:03:56rashersaratoga: I don't think there's any sense in doing that until everything's fixed in the branch
04:04:03kugelor let me quickly commit a mkamsboot change, and branch now?
04:04:15saratogacommit what change?
04:04:17rasherSo we can test the bootloaders that are *actually* going to be released. As in, the same exact files
04:04:42kugelthen have some sort of RC phase to test the output on any targets we plan to support (or update the bootloader for) so that we can tag
04:05:06kugelsaratoga: http://www.rockbox.org/tracker/task/10397#comment31453
04:05:44kugelfunman said we should do 1.1 after committing this anyway
04:06:11kugelI don't really see a reason to wait with branching. We aren't expecting changes to the bootloader, are we?
04:06:32saratogaok i'll just post a patch a then with the text changes for the branch
04:11:23saratogahow wide is a char in the default font?
04:11:25saratoga8 pixels?
04:11:58kugel8x8, yes
04:14:17rasherI'm quite sure it's not 8x8
04:15:24saratogaSYSFONT_WIDTH
04:15:53saratogaok i'm too tired to do this tonight
04:19:26saratogais rbutil ready to go once theres bootloaders?
04:21:35 Quit saratoga ("CGI:IRC (EOF)")
04:22:12kugelit should
04:25:49kugelstupid clip
04:25:54kugelit's so small I always lose it
04:28:06 Nick fxb__ is now known as fxb (n=felixbru@h1252615.stratoserver.net)
04:28:29 Quit fxb (SendQ exceeded)
04:28:42 Join fxb [0] (n=felixbru@h1252615.stratoserver.net)
04:30:35CIA-71New commit by kugel (r21648): Slightly rewrite some parts of mkamsboot for better output and some sanity in the early array declarations. No functional change.
04:39:11kugelrasher: any idea for the branch name?
04:40:54kugelcan a branch be renamed? :)
04:49:05CIA-71New commit by kugel (r21649): Create a branch for updated bootloaders for USB-enabled PortalPlayer devices and initial Sansa AMS bootloaders, including their installers
04:52:28***Saving seen data "./dancer.seen"
04:56:25 Quit kugel (Remote closed the connection)
05:00
05:09:44 Nick fxb is now known as fxb__ (n=felixbru@h1252615.stratoserver.net)
05:27:37 Join Jax184 [0] (i=Jax184@d75-153-89-18.bchsia.telus.net)
05:27:56Jax184Will rockbox ever be coming to the v2 e200 DAPs or any newer sandisk players?
05:28:34 Join saratoga [0] (n=9803c6dd@rockbox/developer/saratoga)
05:28:46scorche|shit is quite likely that it would be declared supported in time
05:29:16saratogaJax184: kugel just branched the e200v2 installer and bootloaders an hour ago
05:29:29Jax184Really?
05:30:41saratogayes theres an email about it on the front page of this website
05:37:10Jax184I'm having some trouble finding this, care to link me?
05:37:54 Quit froggyman (" HydraIRC -> http://www.hydrairc.com <- IRC with a difference")
05:38:22Jax184Or is that http://www.rockbox.org/mail/archive/rockbox-dev-archive-2009-07/0017.shtml ?
05:39:30 Quit fdinel ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org")
05:55:23Jax184So on the current build page, did somebody go and pixel push every little picture of the mp3 players there?
05:59:40 Quit CaptainKwel (Remote closed the connection)
06:00
06:18:59 Join _lifeless [0] (n=lifeless@188.16.68.82)
06:24:36 Quit linuxguy3 (Read error: 110 (Connection timed out))
06:34:24 Part Jax184
06:35:27 Quit __lifeless (Read error: 113 (No route to host))
06:52:30***Saving seen data "./dancer.seen"
06:57:41 Quit barrywardell ()
07:00
07:23:06 Quit goffa__ (Read error: 104 (Connection reset by peer))
07:27:58 Join Horschti [0] (n=Horscht2@xbmc/user/horscht)
07:29:31 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
07:37:27 Join hd [0] (i=jd@modemcable178.248-201-24.mc.videotron.ca)
07:37:47 Quit HellDragon (Read error: 104 (Connection reset by peer))
07:45:59 Quit Horscht (Read error: 110 (Connection timed out))
07:54:07 Join Grahack [0] (n=chri@ACaen-156-1-75-178.w90-51.abo.wanadoo.fr)
08:00
08:12:26 Quit Llorean (Read error: 104 (Connection reset by peer))
08:13:00 Join Llorean [0] (n=DarkkOne@adsl-99-182-52-92.dsl.hstntx.sbcglobal.net)
08:15:20 Part toffe82
08:23:55 Join robin0800 [0] (n=robin080@general-ld-216.t-mobile.co.uk)
08:41:02 Join stoffel [0] (n=quassel@p57B4ED67.dip.t-dialin.net)
08:41:38 Quit stoffel (Remote closed the connection)
08:42:16 Join stoffel [0] (n=quassel@p57B4ED67.dip.t-dialin.net)
08:52:06 Quit robin0800 ("Leaving")
08:52:33***Saving seen data "./dancer.seen"
08:58:52 Join Rob2223 [0] (n=Miranda@p4FDCE92E.dip.t-dialin.net)
08:59:00 Quit beta_ (Read error: 60 (Operation timed out))
09:00
09:00:34 Join beta_ [0] (n=beta@d24-36-78-223.home1.cgocable.net)
09:02:38 Join Dhraakellian [0] (n=ntryon@cpe-72-226-197-191.rochester.res.rr.com)
09:17:35 Quit Rob2222 (Read error: 110 (Connection timed out))
09:26:25 Join flydutch [0] (n=flydutch@host87-202-dynamic.15-87-r.retail.telecomitalia.it)
09:40:44 Quit advcomp2019 (Read error: 104 (Connection reset by peer))
09:41:17 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
09:47:56 Quit r0b- (Read error: 110 (Connection timed out))
09:48:13 Join r0b- [0] (n=rob@adsl-76-235-168-174.dsl.klmzmi.sbcglobal.net)
09:49:48 Join JdGordonn [0] (n=irchon@32.156.190.68)
09:50:08 Quit JdGordonn (Remote closed the connection)
09:51:29 Quit jordan` (Read error: 110 (Connection timed out))
10:00
10:01:51pixelmaAlexP: morning... I'm currently go through your changes in the manual related to the H100 remote control columns and have some questions
10:09:16pixelmaor basically just one - why not use \opt{HAVEREMOTEKEYMAP} for all the \ActionRC stuff. This way you don't need the \opt this pad and \opt that pad in the button tables which was the point of putting them into the keymap platform files (as I see it)
10:11:18pixelmayou also had that in the first commit in some parts (and with underscores though)
10:25:48CIA-71New commit by funman (r21650): mkamsboot: change version string to 1.1, move devices list in the header
10:31:58 Join robin0800 [0] (n=robin080@general-ld-216.t-mobile.co.uk)
10:43:11 Join wincent [0] (n=wincent@host-091-097-067-213.ewe-ip-backbone.de)
10:48:05 Join {phoenix} [0] (n=dirk@p54B466BD.dip.t-dialin.net)
10:48:15 Quit desowin (Remote closed the connection)
10:49:58 Join desowin [0] (n=desowin@atheme/member/desowin)
10:52:36***Saving seen data "./dancer.seen"
10:53:42 Quit homielowe (Read error: 110 (Connection timed out))
10:54:27 Join funman [0] (n=fun@2001:0:53aa:64c:cfc:53b2:a9b7:588b)
10:54:47 Quit flydutch ("/* empty */")
11:00
11:00:16 Join _Auron_ [0] (n=DarkAuro@adsl-76-203-192-240.dsl.rcsntx.sbcglobal.net)
11:00:53 Join homielowe [0] (n=homielow@unaffiliated/homielowe)
11:00:55 Join ender` [0] (i=krneki@foo.eternallybored.org)
11:07:04 Join alexteq [0] (n=5e461110@gateway/web/cgi-irc/labb.contactor.se/x-064345255963b9d3)
11:07:46 Quit robin0800 ("Leaving")
11:07:49alexteqhee
11:08:55alexteqhello is this the right place to ask if rockbox will be supported for sandisk sansa view?
11:10:32JdGordonthis is as good as any.... but its techinally not the right place
11:10:45JdGordoneither way, the answer will always be "maybe"
11:10:57pixelmathere's a port started but no-one can tell if and when it will be finished. Best place to read up on it is probably the corresponding "New Port" thread in the Rockbox forums
11:11:50pixelmaAlexP: I'd also like to revert the changes to nim.tex (Player only plugin) and split_editor.tex (bitmap Archos only)... objections?
11:17:50 Join TechOutsider [0] (n=ca7968e7@gateway/web/cgi-irc/labb.contactor.se/x-604a17ecc0ed1933)
11:19:01alexteqi see! if i wanted to get involved in the development of rockbox what technical or programming skills would it require ?
11:20:48TechOutsidercan i dualboot original ipod firmware w/ original ipod firmware?
11:23:56pixelmacan you rephrase, the question doesn't make sense to me, do you mean dualboot the original firmware with Rockbox? Also, please don't use shorthands like /w as per the guidelines (this is to help screenreaders and automated translation)
11:24:58CIA-71New commit by funman (r21651): Sansa AMS : don't reinvent adc_read(), patch by FlynDice
11:26:12Grahackalexteq: I'm quite new here but I know the areas where we can contribute are vast: translations, RB docs, C code, electronics, DAP docs and charts...
11:27:40 Join matsl [0] (n=matsl@host-90-233-180-166.mobileonline.telia.com)
11:28:17TechOutsideri mean dual boot original firmware with the original firmware
11:28:24 Join homielowe_ [0] (n=homielow@S01060011954e0432.no.shawcable.net)
11:28:26TechOutsiderso it shows up as two ipods in iTunes
11:29:16funmanno, it doesn't make sense. and also the original firmware is not on topic on #rockbox
11:29:38pixelmaI don't know why you would want to do that and I also doubt you'll find the answer here (what funman said)
11:29:54TechOutsiderwell maybe i could use the rockbox bootloader
11:30:03 Join lilltiger [0] (n=lilltige@82.145.152.217)
11:30:30alexteqGrahack: thank you
11:31:25bertrikdoes anyone have a problem with using a typedef in target-specific driver code?
11:32:39funmani think it depends how spread the structure usage is, and if it's small an additional 'struct' isn't very annoying
11:32:40 Quit TechOutsider ("CGI:IRC (EOF)")
11:35:08 Join homielowe__ [0] (n=homielow@66.183.72.25)
11:35:16 Quit homielowe (Read error: 110 (Connection timed out))
11:37:12 Quit alexteq ("CGI:IRC")
11:37:56Grahackalexteq: wandering around your question, I just read that you can contribute setting up a build server too http://www.rockbox.org/twiki/bin/view/Main/BuildServer
11:39:01 Quit funman ("free(random());")
11:45:03 Quit homielowe_ (Success)
11:53:02*pixelma wonders why splitedit.c has buttons for the H100s defined when it is excluded for swcodec targets in apps/plugins/SOURCES
11:54:01pixelmamaybe that's something from before this control file was invented (or someone wanted to port it at one point)
11:57:35 Join n00b81 [0] (n=taylor@unaffiliated/n00b81)
11:58:09gibbon_is there away to find out, if my new-style buildhost has produced anything meaningful yet?
12:00
12:00:14GodEatergibbon_: The new build client results aren't being distributed yet as far as I know
12:01:37gibbon_but thes are sent back to the server?
12:02:00GodEateryes, but I'm not sure they're being kept at the moment
12:02:54pixelmaI guess one of the Swedes could tell you
12:07:49gibbon_well ok... its nothing urgent... i just wanted to make sure, that i did everything right
12:08:07gibbon_the debugging output currently is not that clear about if anything went wrong
12:08:25 Join kugel [0] (n=kugel@rockbox/developer/kugel)
12:08:38gibbon_but thank you
12:08:45kugelhm, I had other plans for mkamsboot v1.1
12:09:03gibbon_i mainly wanted to make sure that the given performance is enough
12:10:41gibbon_s/enough/sufficient/
12:13:44 Quit lilltiger (Read error: 110 (Connection timed out))
12:17:35 Join KBH [0] (n=hbk@pool-71-96-74-73.dfw.dsl-w.verizon.net)
12:17:58 Part homielowe__
12:19:24AlexPpixelma: For the second part, no problem
12:19:53AlexPpixelma: For the first part, I did it specifically on the H100 as some remotes are so short of buttons that they won't have all the actions
12:20:09 Join barrywardell [0] (n=barrywar@86-43-162-146-dynamic.b-ras2.prp.dublin.eircom.net)
12:20:13AlexPThat's why I was asking about the archos remote and how many buttons it has
12:20:25AlexPThe gigabeat remote is also slightly short
12:20:45pixelmaI also wonder if we need to include the complete licenses text in the manual - that's 15 pages in our current layout that won't be read much I guess
12:20:54AlexPBut if there are actions that are covered by all the remotes, then yes it is better to opt on have_remote
12:21:22 Join Thundercloud [0] (i=thunderc@persistence.flat.devzero.co.uk)
12:21:36kugelUnhelpful: ping
12:21:56AlexPpixelma: In fact, thinking about it, that was silly :)
12:22:11pixelmaAlexP: the first case could be also handled iwith \Actions in the platform files - just "declare" them as empty
12:22:16AlexPpixelma: Yes, they should all be opted on haveremote
12:22:28pixelmanot sure which is the cleaner solution
12:22:33AlexPI didn't think about that - silly old me :)
12:22:44pixelmaalready did ;)
12:23:22AlexPNo, I agree with you - it is better to do it on haveremote
12:23:35AlexPand then just have it blank in the platform file
12:24:02 Quit HBK (Read error: 60 (Operation timed out))
12:24:27pixelmaok, commit probably coming up after lunch
12:24:35AlexPOK, sounds good :)
12:32:59 Join jordan` [0] (i=gromit@78.235.252.137)
12:35:29 Join elliott [0] (n=a6526b8a@gateway/web/cgi-irc/labb.contactor.se/x-16ce312143828e56)
12:35:50CIA-71New commit by kugel (r21652): Change to versioning for mkamsboot to <rXXXXX>-<DATE> for svn builds. A fixed number like 1.1 can (and should) be used for releases.
12:37:19 Join elliott1 [0] (n=elliott@h138.107.82.166.ip.windstream.net)
12:38:47 Part elliott1 ("WeeChat 0.3.0-dev")
12:40:28 Quit elliott ("CGI:IRC (Ping timeout)")
12:46:19CIA-71New commit by kugel (r21653): Backport r21652 and give mkamsboot "1.1RC".
12:52:38***Saving seen data "./dancer.seen"
12:53:01 Quit stoffel (Remote closed the connection)
12:53:30 Join ender [0] (i=krneki@foo.eternallybored.org)
12:54:30kugelWhat should release bootloader actually say?
12:54:41kugelBoot Ver. 1.0?
12:55:17 Quit {phoenix} (Read error: 110 (Connection timed out))
12:56:34pixelmawhen I first saw the "Boot Ver." on my c200, I thought that it doesn't sound good (language wise). Maybe a native speaker can comment
12:57:29kugelpixelma: what did it say after the ver.?
12:57:40 Quit daurn ("Cyas")
12:58:26pixelmacurrently 5.0 for me here
12:59:26kugelso just "Boot Ver. 5.0" ? I don't have seen a released bootloader for a long time :/
13:00
13:01:18 Quit Grahack ("Leaving.")
13:02:43pixelmayes - at a similar looking screen as the real boot one. On the M5 on the other hand it is "Rockbox boot loader" and on the next line "Version 4" - everything on the top of the screen (and I think deliberately without logo which was only in use on the Sansa, I think not everyone agreed though)
13:03:03 Join daurnimator [0] (n=daurnima@unaffiliated/daurnimator)
13:04:18kugelhm
13:05:43kugelI think we'll be going with "Boot Ver. 1.0" then for now
13:09:37 Quit daurnimator ("Cyas")
13:11:29 Quit ender` (Read error: 110 (Connection timed out))
13:12:21CIA-71New commit by kugel (r21654): Version 1.0RC for the Sansa AMS bootloader.
13:12:55 Join mcuelenaere [0] (n=mcuelena@78-21-191-122.access.telenet.be)
13:13:04kugelso, they just need a quick verification now (the fuze bootloader works)
13:24:12bertrikkugel, you were referring to r21647?
13:24:37bertrikdoes that patch fix the playback stops on the clip?
13:25:53 Join elliott1 [0] (n=elliott@h138.107.82.166.ip.windstream.net)
13:26:13kugelbertrik: yes, I am. I don't whether it fixes it. I can't find my clip
13:26:36kugelthe whole apps/ code didn't see any cpu cache related functions before
13:26:40elliott1little bastards are too easy to lose...i just found mine after 2 months...
13:26:53kugelindeed
13:28:01kugelelliott1: this is the exact revision. I guess if you still have it then it didn't fix it
13:28:30 Join daurnimator [0] (n=daurnima@unaffiliated/daurnimator)
13:29:29kugelbertrik: it would be nice if we can test the branched bootloader/mkamsboot quickly, so that the branch is free for the PPs
13:31:15 Quit gevaerts (Nick collision from services.)
13:31:27 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
13:31:48kugeland we should poke FlynDice about the voltage thing
13:32:19elliott1i can't seem to reproduce the problem, it seems to play certain songs just fine once then the next time it will freeze half way through
13:32:35elliott1i haven't found a song that it fails every time on
13:37:50elliott1of course, now that i said that, it has stopped twice in a row on this one
13:40:31elliott13 times and this time it is a hard lock...it is 225KBit, 44100Hz
13:40:43 Quit barrywardell ()
13:41:30*bertrik is playing a 2 second music clip in the rockbox bootloader on the meizu m3
13:42:11mcuelenaerebertrik: \o/
13:42:13kugelwow, where's the ladies and gentlemen main?
13:42:18kugelmail*
13:44:21bertrikit's only 2 seconds raw wave (for space reasons) and requires various hacks, but it works :P
13:44:36n00b81Nice job, bertrik :)
13:50:24 Join flydutch [0] (n=flydutch@host87-202-dynamic.15-87-r.retail.telecomitalia.it)
13:50:32 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
13:51:06 Join barrywardell [0] (n=barrywar@86.43.162.146)
14:00
14:00:47 Quit daurnimator (Read error: 110 (Connection timed out))
14:02:47 Quit barrywardell ()
14:04:09 Join stripwax [0] (n=Miranda@87-194-34-169.bethere.co.uk)
14:04:54 Quit elliott1 ("WeeChat 0.3.0-dev")
14:07:31 Join donutman25 [0] (n=Dagni@65.75.86.64)
14:10:17 Join daurnimator [0] (n=daurnima@unaffiliated/daurnimator)
14:14:42pixelmaAlexP: you also have the h100 remote, right? Could you tell me (or make a sketch) how the buttons are called so I could label them?
14:15:22AlexPpixelma: Sure
14:15:30AlexPDo you have your picture?
14:15:48pixelmalet's see
14:17:14AlexPbertrik: Awesome :)
14:19:53bertrikAlexP, thanks. The biggest stumbling block after this is reading (and interpreting) the flash for the file system I think
14:20:26CIA-71New commit by alex (r21655): Swap order of H100 remote keys in a table.
14:20:42AlexPTo be clear - one set of buttons on one table :)
14:21:06AlexPbertrik: Does it share anything with the other direct flash access targets?
14:21:09Torneis the website down?
14:21:25AlexPTorne: Works here
14:21:47Tornei get a timeout :(
14:22:02bertrikAlexP, not sure, but I think so. It uses the 'whimory' flash stack if I understand correctly (also used in the iPhone).
14:22:09Tornei think i worked out what gpio3 bits 10/11 on the beast mean, though :)
14:22:29AlexPTorne: do_stupid_format ?
14:22:34Torneunfortunately no
14:22:37Tornenothing *relevant*
14:22:43Torneif they're zero it sets up flash and ram differently in a way that seems to probably match the devboard
14:22:50Torne(more banks of ram etc)
14:22:52AlexPbertrik: So hopefully that'll help a little
14:22:59AlexPTorne: woo!
14:23:11pixelmaAlexP: only the picture with all remotes currently online: http://home.infocity.de/m.arnold/temp/Remotes.svg
14:23:35AlexPpixelma: OK, so looking at that:
14:23:49AlexPAlong the top are two rocker switches, that can also be pushed in
14:24:09bertrikAlexP, I hope it's not too complicated for reading at least
14:24:11pixelmayep, got that
14:24:25AlexPThe one on the left is rewind when rocked left, fast forward when rocked right, and pushed in it is labelled Navi/Menu
14:24:39AlexPI called those Rewind, Forward, and Navi
14:25:16AlexPThe top right rocker switch is labelled Source/Type when pushed left (I called it Source), Bitrate when pushed right, and Rec when pushed in
14:25:34AlexPAlong the bottom, on the left is the hold switch
14:26:18AlexPOn the bottom right is a third rocker switch - labelled Vol - left, Vol + right, and A-B/Mode when pushed in (I called this A/B)
14:26:40AlexPThen there is play/pause and stop on the face (as you can see)
14:27:14AlexPvoila!
14:27:37 Quit n00b81 ("Leaving")
14:28:02pixelmaalright, thanks. It's there in the logs when I find the time to label it :)
14:28:08AlexPyep :)
14:29:56 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
14:34:39pixelmaAlexP: sorry to bother you once more... can you take a look at the manual's main_menu/fmradio.tex - can you tell me why you called the action \ActionRCFMVolDown (as for the main unit - and also in the c code it's actually using the settings inc/dec)?
14:36:27pixelmaah hmm... the h100 remote has a second set for those in the radio screen
14:41:22*pixelma wonders how to deal best with that - opting in fmradio.tex or something in the keymap file :\\
14:41:33AlexPyeah, not sure
14:42:06AlexPSorry, got to pop off for a bit n ow
14:42:10AlexPWill be back later
14:42:35pixelmathe Iaudio remote uses the same as for settings inc/dec as everywhere else (which is volume up/down)
14:43:13 Part wincent ("Kopete 0.12.7 : http://kopete.kde.org")
14:51:52 Quit kugel (Read error: 110 (Connection timed out))
14:52:39***Saving seen data "./dancer.seen"
14:53:03 Join kugel [0] (n=kugel@rockbox/developer/kugel)
15:00
15:03:26 Nick fxb__ is now known as fxb (n=felixbru@h1252615.stratoserver.net)
15:05:10 Join n00b81 [0] (n=taylor@unaffiliated/n00b81)
15:09:34 Join mt [0] (n=MTee@41.206.134.155)
15:12:56 Quit kachna|lappy (Read error: 113 (No route to host))
15:17:32 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
15:20:42mtI've added another version of the latest patch in FS #10182
15:21:16mtsaratoga, linuxstb : ^^^
15:23:50 Quit daurnimator (Remote closed the connection)
15:24:09mtOne persistent issue is that, when compiling bitstream.c, gcc warns about signed and unsigned types (comparision/in conditional expression), does the build server have these warnings enabled ?
15:31:47 Quit Xerion (" ")
15:33:27 Join funman [0] (n=fun@rockbox/developer/funman)
15:34:43funmankugel: ping
15:36:59 Join Xerion [0] (i=xerion@82-170-197-160.ip.telfort.nl)
15:41:19CIA-71New commit by bertrik (r21656): S5L8700: initial framework for PCM (using DMA transfers)
15:55:32kugelfunman: ping
15:55:36kugelpong*
15:55:48funmannice to see the bootloaders stabilize ;)
15:56:04funmani had a question about r21652 but i answered it myself
15:56:53kugelfunman: we haven't had any fs corruption reports, did we?
15:56:53funmanAPPSVERSION is only used in bootloader/*.c , and the "Boot Ver. 1.1RC" string fits on Clip/m200v4 screens, unlike "Boot Ver. rxxxxx-yyyyyy"
15:56:59funmani don't think so
15:57:23kugelWe should probably revert (or #if 0) the voltage stuff for now, so that it doesn't block the supported status
15:57:52funmanis there any functional change with using adc_read() ?
15:58:09funmanwell yes, it reads the top bits of ADC at least
15:58:11kugelno idea. I don't experience any problems
15:58:49funmanwe aren't in a hurry so we can let it like it is, waiting to know if it's still buggy and what FlynDice thinks about it
15:59:01kugelmc2739 said adc_read doesn't fix his problems
15:59:39kugelyea, we definitely wait if FlynDice has ideas
16:00
16:00:17kugelbut I don't think lowering voltage is major enough in the long term
16:00:31kugel in the short* term
16:01:52funmankugel: can you build a 32bits linux binary of mkamsboot ?
16:01:57kugelyes
16:02:40funmani have win32 & x86_64 (can't access an intelmac atm). Where should we put these binaries, in a 1.1RC folder of the download server or a FS ticket?
16:03:15kugeldon't know. should we put them anywhere at all?
16:03:27mtAnyone with a big endian target willing to test RM patch ?
16:03:48kugelI planned that we quickly verify mkamsboot and the bootloaders (at least on the targets to be supported), and then release
16:04:07funmanmt: i had a look, i think only coldfire (irivers) are big endian.
16:04:18funmankugel: makes sense
16:04:20kugelBut I don't think we should release before really putting the main build up too
16:05:06funmanwhy is a svn branch needed for AMS ?
16:05:34kugelbecause we can hack in version strings better. Plus, the branch is also for PP bootloaders
16:05:46 Join linuxguy3 [0] (n=timj@68.253.209.227)
16:06:34kugelI rather branch, instead of hacking in the version string, then release, and then revert the whole thing agian
16:08:45kugelwe can tag off the branch then, and don't mess with it at all in the trunk
16:08:50mtfunman: If I compiled a simbuild for an iriver, do you think the results would be significant ?
16:09:11funmanmt: no, unless you compile it on a big endian system (like OSX PPC)
16:09:54mthmm.. I'll just wait then till an iriver owner could test the patch.
16:11:53 Join stripwax5443 [0] (n=Miranda@87.194.34.169)
16:12:12stripwax5443someone need help testing on iriver target?
16:12:53mtstripwax5443: Yes, FS #10182, latest patch. :)
16:13:04 Join jon-kha [0] (i=jon-kha@kahvi.eu.org)
16:13:08stripwax5443hm, rockbox.org down for me
16:13:53 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
16:14:17stripwax5443rockbox.org working for everyone else?
16:15:03n00b81works for me
16:16:24 Join stripwax_ [0] (n=Miranda@87-194-34-169.bethere.co.uk)
16:16:34pixelmaworks for me too... I see testing on a coldfire target is needed?
16:17:13pixelmabut I have no .rm files
16:17:22kugelfunman: I think we should slowly start to agree on a release date. We aren't in a hurry but it would be nice to get this done
16:17:57*stripwax_ goes for the reboot option ..
16:18:13 Quit stripwax_ (Read error: 104 (Connection reset by peer))
16:18:52funmankugel: release date is mid september ?
16:19:12funman3 months after 3.3
16:19:28mtpixelma: Here are some samples http://samples.mplayerhq.hu/real/AC-cook/
16:22:03kugelfunman: I don't mean that release. We don't need to wait for 3.4 to put the AMSes under supported
16:22:07 Quit stripwax5443 (Read error: 104 (Connection reset by peer))
16:22:08 Quit stripwax (Read error: 104 (Connection reset by peer))
16:23:02funmankugel: there is only 1 kind of release and it's rockbox releases.
16:23:14dzwhat C library should I be using for arm? (trying to do a toolchain using crossdev on gentoo, complains about missing keyword for newlib)
16:23:29funmanI'm not sure we need to define a date, rather we should mark the AMS builds as supported when there is no major bug in them (which we still aren't sure?)
16:23:35kugelfunman: yes, but we can always offer current builds + installation (etc) without a release
16:23:50funmandz: use the rockboxdev.sh script in tools/ (no libc is needed)
16:24:10funmankugel: of course, but i still don't think we need to give us a deadline
16:24:24kugelhence the "slowly"
16:24:43dzfunman: alright, I suppose someone should take the crossdev instructions off of the wiki page then?
16:24:45 Quit matsl (Read error: 110 (Connection timed out))
16:24:58funmandz: i think so, which page is it ?
16:25:21kugelWhat I basically mean was that we should watch for issues now, and if we don't find a major one in a timeframe of say 2 weeks, let's put the builds up
16:25:47dzfunman: http://www.rockbox.org/twiki/bin/view/Main/CrossCompiler
16:25:48 Join matsl [0] (n=matsl@host-90-233-230-237.mobileonline.telia.com)
16:25:57kugelthe fs corruption appears to be gone with your bank switch fix, the only issue I know is the voltage thing
16:27:20funmani think the instructions for manual build should be removed as well (not the instructions for GDB though)
16:27:27dzfunman: actually, looks like adding −−without-headers may fix it
16:28:01kugelfunman: why?
16:28:08funmankugel: some people mentioned corruptions and deadlock at least before r21550
16:28:31funmankugel: are these instructions supported ? (and maintained). This is not the case for the gentoo instructions at least
16:28:51crashdis there any way to remap the ipod hold function to be something else other than the hw hold switch?
16:28:53kugelit's a wiki, it has many things we don't maintain
16:28:56crashdor is that going to be a meaty code hack
16:29:36kugelfunman: hence I said it appears to be gone since that commt
16:29:39funmankugel: this page is linked to from the docs index so the instructions are official
16:29:55kugelI don't think they should be removed
16:30:14kugelwhy should they? Are they not working?
16:30:24funmankugel: just look what dz said
16:30:32kugelgentoo doesn't work, yes
16:30:57kugelbut they still are more helpful than no instructions at all, or am I wrong here?
16:31:04pixelmacrashd: there already is a software hold feature for targets that don't have a hold switch, maybe you can reuse that
16:31:12kugeland I'm fairly sure the manual way also still works
16:31:13dzgentoo seems to be working with the −−without-headers flag added
16:31:20crashdpixelma: interesting
16:31:31crashdmy hw hold switch is broke ysee so im looking into a workaround
16:31:39crashdhow does it work on software hold targets?
16:31:46pixelmacrashd: I believe there were already threads in the forums about it and there were some more answers
16:31:51funmanhttp://www.anythingbutipod.com/forum/showpost.php?p=370168&postcount=157 mentions deadlock after r21550
16:32:03kugelfunman: if that page is so official, then the instructions are probably also supported.
16:32:22pixelmacrashd: you push a certain key combo to enable softwaree hold (in the WPS) and press it again to disable it
16:32:27funmankugel: i just meant that only rockboxdev.sh is supported
16:32:40kugelI don't think that's true
16:33:03kugelit's a convinient script, but we're not forcing anyone on how to build the crosscompilers
16:33:08crashdseems pretty trivial to cchange the HAVE_SW_HOLD define
16:33:10crashdthanks pixelma :)
16:33:32funmankugel: not forcing, but we're not helping people who don't want to use rockboxdev.sh
16:33:49kugelfunman: how come you think he was using a build after r21550?
16:34:04pixelmacrashd: the presses aren't discarded completely (backlight lights up but you can't do accidentally changes... you probably need to edit your keymap file to assign a key combo
16:34:10funman"latest patch"
16:34:35funmanand he says incorrect display was fixed
16:34:36kugelthat doesn't mean anything
16:34:44crashdpixelma: that's great
16:34:45funmantrue
16:34:46crashdthanks for the help :)
16:35:15kugeland even if, do we know he reformatted before updating to "the latest patch"?
16:35:41funmanit doesn't matter
16:35:58kugelI don't really give a shit on reports of 1 individual who provided such loose informations
16:36:38kugelIt doesn't matter? You said yourself in the commit message to only report bugs on freshly formatted devices
16:36:47funmanwhat i want to say is that no bug report doesn't mean no bug
16:37:12funmanit only matters when seeing filesystem corruption (incorrect code loaded like codecs or plugins)
16:37:39funmannot for a lock (which comes from an unrecoverable error in an hardware SD operation)
16:38:02pixelmacrashd: you can look up ACTION_STD_KEYLOCK in one of the Archoses' keymap files
16:38:26 Quit Nico_P (Remote closed the connection)
16:38:37funmanif we make mkamsboot/bootloaders builds we could ask on the forum thread for testers (I think there is a lot of people waiting to test)
16:38:39kugelfunman: I can't follow you
16:38:52funmankugel: what's wrong?
16:39:05kugelI didn't understand what you were trying to tell me
16:39:25funmanin which sentence?
16:39:46kugelthe three after it doesn't matter
16:40:29funmanthe 1st sentence was separated from the 2 following
16:40:45kugelAlso, what's the point of making test builds of the bootloader and mkamsboot and asking for tests (i.e. support those)? A current build is a test build, so we can put all that stuff up, making it supported. It's the same afterall
16:41:24funmanwe need testing before we support it
16:41:38funmanand for testing we need bootloaders and mkamsboot binaries so people can use the current build
16:42:15kugelThat's no different from putting the binaries on the download server
16:42:36funmani'm not saying something different
16:44:01kugelI don't think we need to mess with putting something in the forum. If we don't know of a major bug, then we can do all the supported stuff, and receive testing that way (that's what the current build is for afterall)
16:44:12funmani know of a major bug
16:44:34funmanand i need lots of testing to be sure it was fixed by r21550
16:45:11kugelwhat major bug (aside from the voltage thing?)
16:45:16funmanif it wasn't, i have no idea of how much time is needed for fixing this
16:45:22funmanSD driver deadlocking
16:45:25kugels/?)/)?/
16:45:43kugelThat's caused by the voltage change
16:45:49funmanno
16:46:01kugelWhat sd driver deadlocking then?
16:46:37funmanthe sd controller reporting a DATA TIMEOUT indefinitely
16:46:58kugelthat's useless
16:47:30funmani don't think that's useless, if the hardware reports this error it really happened.
16:48:12kugela) I don't have that b) "the sd controller reporting a DATA TIMEOUT indefinitely" is information only you can provide so I can connect it to any other report in the forum for example
16:48:47funmanhttp://forums.rockbox.org/index.php?topic=14064.msg152510#msg152510
16:49:41kugelI can't find where that guy tells something about data timeout
16:49:54funmanI detailed the problem in my edit of http://forums.rockbox.org/index.php?topic=14064.msg151588#msg151588
16:50:21funmancurrently the code will just go into an infinite loop, I got this information by adding debug code
16:51:12funmani want to make the driver print a useful error message if this happen, but it's still on my TODO list
16:51:17kugelthat's long ago. I thought it's fixed with aligned buffers and the bank swithc fix
16:51:32kugelcan you still reproduce it with current svn?
16:51:43funmankugel: i don't know if it was fixed, we can only tell with wide testing
16:51:51funmanno since i have no method to reproduce it
16:52:17kugelSee. And we get the best testing by supporting it with all downloads and stuff
16:52:31kugelI don't see why we need to make it locally to the forums first
16:52:41***Saving seen data "./dancer.seen"
16:52:46funmankugel: but if it's still buggy, what's the point of supporting the buggy builds?
16:53:21kugelit's not a known bug as such, because there were many commits that are not unlikely to fix it and we don't get reliable bug reports for that issue
16:53:38kugeland if we have no known bug, there's no reason to hold anything back
16:53:46funmanlike i said, no bug report doesn't mean no bug
16:54:12funmanthat, because the bug appears randomly
16:54:34kugelso we never put it supported?
16:54:39kugelbecause it may be buggy?
16:54:49 Join mt_ [0] (n=MTee@41.206.137.59)
16:54:59funmani suggested that we ask people to test it before supporting the Sansa AMS
16:54:59kugelThat's the definition of the current build, and we have to get wide testing for potentially buggy stuff
16:55:14funmanwith enough testers we can make sure this bug was fixed
16:56:01 Quit mt_ (Client Quit)
16:56:15kugelI think you're putting a bit too much into the "being supported" thing
16:56:23 Join mt_ [0] (n=MTee@41.206.137.59)
16:56:31funmanwhat do you think is wrong with some testing before supporting the builds ?
16:56:38kugelit doesn't mean we're guaranteeing a flawless build
16:57:08pixelmawhat's the point in calling it supported too quickly? If there are potentially still enough users who can damage their filesystem (maybe lose important data) I think I'd give testing another round. And my impression is that there are enough testers around
16:57:19kugelnothing as such, except that it's effectivly the same as supporting, with the difference that we get even more testers with supporting
16:57:53kugelas you said yourself, we are not doing a release
16:57:57funmankugel: well if there is no difference for you, then I prefer calling for testers before calling Sansa AMS supported
16:58:42kugelthen let's do it that way
16:59:25funmanwhen do you want to make builds of mkamsboot and bootloaders ?
16:59:54kugelI could make it right now
17:00
17:00:28funmango for it, and mail them to Bagder so he can put them on the download server
17:00:54kugelI'll do them off the branch so that they're still RC
17:00:57 Quit mt (Read error: 104 (Connection reset by peer))
17:01:05 Nick mt_ is now known as mt (n=MTee@41.206.137.59)
17:01:19funmankugel: so you want to do a candidate release ?
17:01:20kugelwe should probably also announce it on abi
17:01:53kugelfunman: do you want to have them final before putting the players supported?
17:02:28 Quit mt ("ChatZilla 0.9.84 [Firefox 3.0.11/2009060215]")
17:02:30funmanwhy not? There is no known bugs in mkamsboot and the bootloaders; only in the "normal" rockbox.sansa build
17:02:52kugelit's just a matter of changing the version string
17:03:06funmanNote that the c200v2 bootloader is buggy (missing buttons and buggy LCD driver), but the others work fine
17:03:15kugelbut I don't see why we should already make them final before. I'd rather do that in parallel with the supported status
17:03:39funmankugel: the choice is up to you ;)
17:03:52kugelI also thought we start off with e200v2 and fuze only, since the others have indeed major bugs
17:04:52funmanbootloaders are unaffected, but there is no point of having bootloaders without a supported (or planned shortly) build
17:05:46kugelso what now?
17:06:09funmani don't know, what's your opinion?
17:06:17kugelwhat I said above
17:06:40funmanok with me
17:06:46kugelwe're calling for testing since we want to know about bugs we don't know aout yet
17:07:08kugelthere's no use if we get repeated reports of "missing buttons" or "stopping playback"
17:08:21 Join Lss [0] (n=Lss@cm33.zeta237.maxonline.com.sg)
17:13:08pixelmamt left?
17:13:39funman17:02 < mt> I'll be back in an hour or so.
17:16:10*dz hopes he is not causing too much distress to the svn server
17:16:35 Join toffe82 [0] (n=chatzill@ppp-71-138-19-215.dsl.frs2ca.pacbell.net)
17:17:00bertrikAFAIK, denes was the last guy to have a serious look at the Meizu flash, but it's been 4 months since he was last online :(
17:18:14pixelmafor the logs: the sample .rm (took the vetenskap ones) don't sound correctly on my M5 (there is noise but not a continuous thing). I also notice in my WPS that the reported playback times "jump around" like mad, even negative numbers).
17:18:20pixelma^mt
17:19:04kugelfunman: so I'll mail him the bootloaders and mkamsboot for linux x86, when will you build for win32 and linux x86_64?
17:20:53 Join mt [0] (n=MTee@41.233.147.176)
17:21:04funmannow, should I use the 1.1RC version in the branch?
17:21:23 Join AndyIL [0] (i=AndyI@212.14.205.32)
17:21:24kugelwould be nice
17:21:33funmani can send it to you so you mail the whole thing to him
17:22:09kugeltoo late :>
17:22:38kugelI wrote that he'll receive those from you
17:23:11kugelif you're interested, I put mkamsboot into "ams-testing/mkamsboot-v1.1RC/" in the zip file
17:26:17funmanwhat else is in the zip?
17:26:38kugelthe bootloaders in "ams-testing/bootloaders-v1.0RC", but you don't need to rebuild them
17:27:51kugelhow do you build for win32? Cygwin?
17:28:14funmancross compiler
17:28:17gevaertsfor sansapatcher, I've used mingw in the past
17:28:21funmanmingw-* packages in debian/ubuntu
17:28:24pixelmamt: put my results on my M5 in the logs for you... I also checked now that the files play correctly on PC (with VLC) and check the build log, there is a warning in get_metadata and about signedness in libcook/bitstream.c. To be sure I compiled after a make clean and rerun of configure
17:28:25kugelah I see
17:29:05kugelgevaerts: mkamsboot needs cross-compiler, though, that complicates things
17:29:24gevaertskugel: in what way?
17:29:43kugelyou need arm-elf-gcc
17:29:53kugeli.e. plain mingw won't work
17:30:12kugeland we still don't have the dualboot binaries in svn
17:30:14gevaertskugel: I don't run mingw on windows :)
17:30:38kugelwe should probably do that when we tag mkamsboot 1.1
17:30:42funmankugel: arm-elf-gcc is not needed anymore ?
17:30:50kugelno?
17:31:09funmandualboot.{c,h} were committed in r21118
17:31:28kugeloh, I forgot about that
17:31:48kugelplain machine code in C arrays, right
17:31:58funmani think gevaerts meant "mingw" as in "the mingw package provided by debian" ?
17:32:05gevaertsI did indeed
17:32:07pixelmamt: that's with the rm_playback.v1.1.patch one
17:32:21kugeldid anyone actually try if it runs on windows too?
17:32:30funmanthere is also a "mingw" which is similar to cygwin (runs on windows), I don't know if they are related
17:32:33funmankugel: it runs in wine ;)
17:32:53gevaertskugel: last time I did it, yes. Still sansapatcher though, not mkamsboot
17:33:11CIA-71New commit by mcuelenaere (r21657): Lua: add script which wraps not-yet ported C functions to Lua
17:33:12 Quit AndyI (Read error: 110 (Connection timed out))
17:33:17mtpixelma: I know about the signedness warning, I asked before if the build server would complain about them too.. Just wanted to check the parsing of the file is correct and that the codec plays. I'll check the codec again to see what could be causing the noise you mentioned. Thanks alot, much appreciated. :)
17:33:19kugelfunman: we'll see in the forum thread :)
17:34:46pixelmamt: well, the track isn't recognisable. On the upside, it doesn't crash my player :)
17:35:22mtpixelma: Was just writing a question about it being recognizable. :)
17:39:11 Join fdinel [0] (n=Miranda@modemcable204.232-203-24.mc.videotron.ca)
17:39:20CIA-71New commit by mcuelenaere (r21658): Add a Rockbox header..
17:41:14Unhelpfulkugel: what? :)
17:41:19 Part donutman25
17:42:20kugelUnhelpful: uhm, that was shortly after waking up, I totally forgot what I wanted
17:43:37 Quit alexbobp (Connection timed out)
17:43:53kugelUnhelpful: ah I think I remember: I get audio dropouts and the empty slide often when scrolling in PF. I never get either of those on my e200, a much slower dap
17:44:37kugelthere's probably something yielding not enough, any idea what that could be?
17:45:29CIA-71New commit by mcuelenaere (r21659): Fix actions.lua & buttons.lua make errors
17:45:44Unhelpfulthere's a yield per slide rendered... don't really think we should make that more often. i don't remember precisely how often the load thread yields, but it has rather high priority.
17:46:12kugelI'm wondering why this isn't a problem on the e200
17:46:33gevaertsThe current sansa bootloader is 5.0, right?
17:46:38kugelyes
17:47:14Unhelpfulvery hard to say. display runs ahead of load on beast also... perhaps that happens more easily on targets that can finish rendering a screen faster?
17:47:26kugelgevaerts: are you looking into updating the PP bootloaders and patchers?
17:47:32gevaertsyes
17:47:55kugelgood. the branch is though for those also
17:48:02gevaertsI'm mainly trying to work out how to set the bootloader version properly
17:48:14kugelI just hacked into show_logo.c
17:48:39kugelpassing VERSION=XX to make didn't seem to work
17:49:12gevaertsSo this would be 6.0, right?
17:49:18kugelyea
17:50:05Unhelpfulmaybe we ought to adjust the thread priorities dynamically - boost the load thread if we're running out of slides?
17:50:22kugeleven more audio dropouts?
17:50:24gevaertsThe problem I see is that it's probably not 6.0 on mr100 and h10
17:51:01rasherMaybe it should be
17:51:02kugelcan't you just do #ifdef IRIVER_H10 for example?
17:51:16kugelor what rasher said
17:52:10CIA-71New commit by mcuelenaere (r21660): Fix sim_* errors when compiling Lua
17:56:21bertrikI wonder, in the current meizu framework we reserve room for the FIQ and IRQ stacks, but this doesn't seem to be done for the sandisk targets
17:56:38bertrik(looking at boot.lds)
17:56:46Unhelpfulkugel: or maybe reduce the render thread priority instead? or not thread loads at all, and just attempt to load threads in the render thread?
17:56:51Unhelpfulload *slides*.
18:00
18:01:39 Join BryanJacobs [0] (n=bryanjac@cpe-74-67-191-154.rochester.res.rr.com)
18:04:24kugelUnhelpful: also, what happened to your scrolling work?
18:06:23BryanJacobsplugins take a long time to build, is there some easy way to turn them off with tools/configure?
18:06:27 Quit matsl (Read error: 60 (Operation timed out))
18:06:37kugelI'd like to make pf more "sane" maybe, i.e. not boosting all the time, not updating while it's rendering nothing, no backlight which is constantly on
18:06:39rasherBryanJacobs: plugins=no
18:06:46BryanJacobsrasher: thanks
18:06:48rasherBryanJacobs: In the target's section
18:07:13rasherBryanJacobs: Line 850 onwards
18:07:14Unhelpfulkugel: not updating while not scrolling was something i had working on color targets. :/
18:08:01BryanJacobsrasher: found it, I was hoping there'd be a parameter instead of editing the file <shrug>
18:08:20rasherBryanJacobs: I don't believe there is
18:08:29gevaertsI can compile proper v6.0 bootloaders by editing the makefile, i.e. I can't commit this. Is that good enough?
18:08:38Unhelpfulit was part of the core scrolling work... disappeared when my laptop got formatted. it should be much faster to write again. :)
18:08:41BryanJacobsah well, I'll just turn that off so I don't have to build Doom to test a codec
18:08:55kugelah I see
18:09:10mcuelenaereBryanJacobs: you could do make bin
18:09:28mcuelenaere(and make bin && make codecs if you want codecs too)
18:09:29kugelthat doesn't build codecs, which is probably not what he wants :)
18:10:03BryanJacobswhat exactly is "make bin", just the rockbox monolithic core?
18:10:15kugelyea
18:10:22mcuelenaeresee make help
18:10:23Unhelpfulprobably it should unboost when it enters the static scrolling mode...
18:10:28 Quit mt (Read error: 104 (Connection reset by peer))
18:10:30kugelstops after rockbox.<target>
18:11:25BryanJacobsarr, why is move_handle called on a pretty-much-empty buffer?!
18:12:14mcuelenaerehmm is it normal that one can't do rb->audio_play(foo); in sim builds?
18:12:43mcuelenaereresults in a error: ‘const struct plugin_api’ has no member named ‘sim_audio_play’
18:12:56 Join webguest24 [0] (n=d1c345e7@gateway/web/cgi-irc/labb.contactor.se/x-af4ef0e42cbd3078)
18:13:16BryanJacobsdoes anyone know why move_handle doesn't check whether the handle being moved crosses another one?
18:13:23CIA-71New commit by kugel (r21661): Sansa Fuze: Correct some solitaire controls.
18:13:39 Quit webguest24 (Client Quit)
18:13:41BryanJacobsI mean, there's no RINGBUF_ADD_CROSS(loc,moveamt,handlebeingmoved->next) in the source
18:14:55BryanJacobsalso, WHY are handles not located right before their data all the time?
18:15:11BryanJacobsI mean, is there a good reason not to have the struct memhandle border that handle's data?
18:15:44kugelUnhelpful: it seems the fuze's lcd is just too slow
18:15:57kugelI don't get drop outs with a faster driver
18:15:59rasherBryanJacobs: the people who know about the playback/buffering system are unfortunately few and far between
18:16:08BryanJacobshm.
18:16:37BryanJacobsI think I pretty thoroughly understand it now, but I see what I perceive to be poor design choices occasionally... The system looks to have kind of spitballed up
18:17:08Unhelpfulkugel: basically, i had it set up so that the scroll functions returned a pointer to the struct scrollinfo, and some extra style flags and a global pointer for custom scroll styles.
18:17:10BryanJacobslike, for example, shrink_handle moving a struct memhandle to be next to its data but other code assuming that already is so
18:17:40kugelno, no they're not totally gone
18:19:25kugelBryanJacobs: feel free to improve ;>
18:20:41BryanJacobskugel: I've got a system that allows multiple handles almost ready
18:20:53BryanJacobserr, multiple "current" handles
18:21:18BryanJacobswhich will in turn allow multiple files and more efficient utilization of buffer space
18:22:55kugelBryanJacobs: I wonder if that could be reused for crossfade. The current implementation is rather awful
18:23:53BryanJacobskugel: well, the old buffering code actually could have been
18:24:16BryanJacobspreviously when you called bufopen() a second time, the buffering system allocated space and requested that buffering the previous file be finished
18:24:27BryanJacobswhich means the old file would be memory-resident and thus available to play
18:24:49 Quit Sajber^ (Read error: 104 (Connection reset by peer))
18:24:57BryanJacobsthe major change I've made is that a newly allocated handle won't necessarily reside at the buffer write index position
18:36:57 Join aaron424 [0] (n=chatzill@adsl-065-013-002-216.sip.asm.bellsouth.net)
18:38:35*bertrik fails at setting up a simple interval timer on the meizu
18:41:19CIA-71New commit by mcuelenaere (r21662): Lua: implement gui_syncyesno_run
18:41:20funman"to use this software, you must press a specific button every 10ms"
18:42:05BryanJacobsfunman: lol
18:44:00 Join killan [0] (n=nnscript@c-0efa70d5.06-397-67626721.cust.bredbandsbolaget.se)
18:45:11 Quit funman ("free(random());")
18:46:24 Join Horscht86 [0] (n=Horscht2@p4FD4DAB9.dip.t-dialin.net)
18:51:29 Quit flydutch ("/* empty */")
18:52:43***Saving seen data "./dancer.seen"
18:53:44BryanJacobssee, this is the kind of code I'm talking about:
18:54:01 Join flydutch [0] (n=flydutch@host87-202-dynamic.15-87-r.retail.telecomitalia.it)
18:54:01BryanJacobscopy_mp3entry((struct mp3entry *)h->data, (const struct mp3entry *)src);
18:54:14BryanJacobsh->data is a size_t >_<
18:54:30BryanJacobsyou can't cast it to struct mp3entry* and have it work
18:54:40BryanJacobsthis code must not ever be exercised
18:54:53rasherOh dear
18:57:47 Quit bluebrother (Read error: 104 (Connection reset by peer))
18:58:10 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
18:58:26 Quit tvelocity (Remote closed the connection)
18:58:40 Join n1s [0] (n=n1s@rockbox/developer/n1s)
19:00
19:03:31 Join AndyI [0] (i=AndyI@212.14.205.32)
19:03:49 Quit Horschti (Read error: 110 (Connection timed out))
19:07:02 Quit flydutch ("/* empty */")
19:07:30 Join efyx_ [0] (n=efyx@lap34-1-82-224-140-171.fbx.proxad.net)
19:08:21 Quit AndyI (Client Quit)
19:09:30 Join flydutch [0] (n=flydutch@host87-202-dynamic.15-87-r.retail.telecomitalia.it)
19:10:35JdGordonkugel: pong
19:11:40kugelJdGordon: ping
19:11:55JdGordonyou pinged me yesterday?
19:12:21kugelyea
19:12:33JdGordonbertrik: congrats!
19:12:43kugelI can't recall why though
19:12:47bertrikJdGordon, thanks!
19:13:46gevaertskugel: maybe the exit to WPS thing?
19:14:11kugelright!
19:14:16JdGordonanyone want to test out 10406? I'm pretty sure its ready for commit, assumine ghtere is no objection
19:14:39 Quit AndyIL (Read error: 110 (Connection timed out))
19:14:56kugelI won't object, unless you still can't tell me more about your doubts regarding going to the wps from a plugin :)
19:14:59rasherJdGordon: why would you want the statusbar to be enabled on some screens (which?) but not others?
19:15:27rasheroh, display. Ignore me.
19:15:55 Join mt [0] (n=mt@41.233.147.176)
19:15:55JdGordonlovely terminolody we have isnt it? :p
19:16:12rasherIt makes sense, I was just not thinking
19:22:30 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
19:24:18pixelmabluebrother: is there a macro I could use to have e.g a file called iaudio-remote.svg included for all Iaudios? I know there is something similar for the screenshots but didn't know whether and how I could use it for the player's images (and didn't even see where either \specimg or \genericimg are defimed, one is in preamble the other a mystery to me)
19:24:24kugelOBJECT! You have trailing spaces in the recscreen hunk
19:25:28 Nick Horscht86 is now known as Horscht (n=Horscht2@p4FD4DAB9.dip.t-dialin.net)
19:25:51bertrikweird, timer A works, but timer B, C, D doesn't quite work
19:26:33 Join AndyI [0] (i=AndyI@212.14.205.32)
19:27:15kugelJdGordon: why aren't you using VP_SB_ONSCREEN (for statusbar_enabled in viewport.c, for example)?
19:27:30mtpixelma: ping
19:27:45pixelmapong
19:27:59mtAre you free for a bit more testing of rm ?
19:28:23JdGordonkugel: instead of?
19:28:39kugel|= 2 in that case
19:28:48pixelmamt: a bit (as I'm also working on the manual and compile with cygwin, so a bit slower ;) )
19:29:32kugelyou could also extent the macros to have another bit for top or bottom and save a whole lot of #ifdefs
19:30:56 Join Sajber^ [0] (n=Sajber@h-142-185.A213.priv.bahnhof.se)
19:31:06bluebrotherpixelma: \genericimg and \specimg are defined in the platform files
19:31:18mtpixelma: OK, I'll just upload the new patch so that you could test it whenever you have time.
19:31:52bluebrotheryou should be able to use \opt{iaudio}
19:34:49pixelmaI'd like to have it included automatically like the xxx-front images are, just that it also uses iaudio-remote instead of only m3-remote so I can use one image for all 3 Iaudios
19:35:14 Quit flux (Read error: 54 (Connection reset by peer))
19:35:16pixelmacan I assign \specimg twice?
19:35:36pixelmaguess not
19:35:59bluebrotheras in two values? No
19:36:02 Join saratoga_home [0] (n=41becb3b@gateway/web/cgi-irc/labb.contactor.se/x-4eeb644110ee8e6e)
19:36:37bluebrotheryou could use \playerman though that uses a capital as first letter.
19:36:42saratoga_homemt: wma needed a define to convert ROCKBOX_BIG_ENDIAN to WORDS_BIGENDIAN for the ffmpeg bistream parsing
19:38:21mtpixelma: When you're free to test, you could try the new patch on a clean checkout or just do this on the v1.1 patched one : http://pastebin.com/m4a6113c5
19:38:49mtsaratoga_home : Just changed WORDS_BIGENDIAN in bswap.h in the new patch.
19:39:05saratoga_homeah then i am too slow
19:39:09 Quit BryanJacobs ("Java user signed off")
19:40:11pixelmabluebrother: the thing is that it will be different for the Irivers but the Iaudios could use the same image. I remember that genericimg is a bit flexible there - e.g. use xxx-ondiofm if it exists otherwise xxx-ondio etc. I hoped there could be a similar thing there (the other one uses specimg but that's tied to screenshots, right?)
19:40:35mtsaratoga_home: I'd say it's me that's slow (very late reply) :)
19:40:58pixelmabluebrother: sorry, genericimg in the end too
19:41:15mtpixelma: Still didn't fix the signedness warnings though, so you'll still be getting them.
19:41:24bluebrotherwell, we could of course invent something similar here :)
19:41:50mtsaratoga_home: Do you know if the build server complains about comarisons between signed and usnsigned or not ?
19:41:52pixelmawould be nice :)
19:42:51kugelmt: why don't you just fix it instead?
19:43:02 Nick n00b81 is now known as n00b81|afk (n=taylor@unaffiliated/n00b81)
19:43:27 Join Blue_Dude [0] (n=chatzill@uslec-66-255-178-43.cust.uslec.net)
19:43:41pixelmamt: will do
19:44:18 Join flux [0] (i=flux@jolt.modeemi.cs.tut.fi)
19:44:22mtkugel: I'm working on it now, just focusing on fixing the BE issue first.
19:44:28mtpixelma: Thanks.
19:45:15saratoga_homemt: yes it will complain
19:45:21bluebrotherpixelma: I'm out the next hour or so. Will try to give it a look later if time permits.
19:45:43saratoga_homethe build servers are just various developer's PCs, so they should behave exactly as your own
19:46:08pixelmabluebrother: nice, take your time - have to do other things myself either way :)
19:46:47pixelmamt: a make clean to be safe first or is it not needed?
19:46:55 Join andrea [0] (i=4f345713@gateway/web/freenode/x-3ae63a4014ef72d7)
19:47:12mtpixelma: If you followed the pastebin it is not needed.
19:47:23 Nick andrea is now known as Guest33434 (i=4f345713@gateway/web/freenode/x-3ae63a4014ef72d7)
19:49:40 Quit Guest33434 (Client Quit)
19:55:31JdGordonkugel: see anything else in the diff that should be fixed?
19:57:21pixelmamt: I don't notice any difference, but I just notice that bswap.h wasn't listed as compiled again - shouldn't it?
19:57:23 Join sakuramboo [0] (n=sakuramb@ool-43504efe.dyn.optonline.net)
19:58:15kugelI'm not sure whether "remote_statusbar" is needed, I'd rather see the VP_SB_* macros to be used, but I guess that's not really doable with the current settings code
19:58:28saratoga_homepixelma: its a header, so only the files that include it are recompiled
19:58:44mtpixelma: it should recompile libcook/bitstream.c and libcook/cook.c
19:59:14pixelmayes, it did
19:59:40pixelmaalso codecs/cook
19:59:48CIA-71New commit by mcuelenaere (r21663): Also make rocklib_aux.c depend on $(LUA_OBJ)
19:59:56pixelma.c
20:00
20:01:05mtI'll recheck.
20:05:35sakurambooon the iPod versions, is there a function to stop the arm on the hard drive from moving when the "battery dead" symbol flashes on the screen?
20:06:10CIA-71New commit by mcuelenaere (r21664): Take 2 at 'Consolidate all fixed point math routines in one library' (FS #10400) by Jeffrey Goode
20:06:21 Join alexbobp [0] (n=alex@ppp-70-253-66-27.dsl.austtx.swbell.net)
20:07:57mtpixelma: could you pastebin the output of make ?
20:08:03CIA-71New commit by jdgordon (r21665): FS #10406 - split the statusbar setting into one for each display, and allow the bar to be at the top or bottom of the display
20:11:09pixelmamt: from the second run?
20:11:37Bagderhm bootloader/sandisk-sansa/[model] is the path to the sansa PP bootloaders
20:11:48Bagderanyone has a suggestion for the ams ones?
20:12:02kugelmcuelenaere: many files had svn:executable
20:12:04Bagderor should we have them in the same spot?
20:12:06mtpixelma: yes, after editing bswap.
20:12:10Bagderthe model names are different
20:12:14mcuelenaerekugel: bah, git..
20:12:40kugelmcuelenaere: I mean in the patch, I don't know what git does to those
20:13:05mcuelenaereah, I don't know how to set SVN properties in git
20:13:05saratoga_homemodel is e200v2 right?
20:13:13BagderI guess, yes
20:13:20kugelBagder: Same spot I'd say.
20:13:23mcuelenaerepatch should ignore SVN properties; but git has its own version (git apply)
20:13:59kugelbut IIRC git svn doesn't handle setting properties
20:15:51pixelmamt: http://rockbox.pastebin.ca/1484862 but I try again in case I made a mistake somewhere
20:16:19mcuelenaerekugel: then I suppose no SVN property will be set/changed :)
20:17:53mtpixelma: The audio is still unrecognizable, right ?
20:19:12kugelmcuelenaere: seems so :)
20:20:04pixelmaAlexP: hope that change merged correctly now...
20:20:05dzalright, only 1600 revisions to go until I stop abusing the svn server -_-
20:20:42kugelwhat are you doing?
20:21:03dzgit svn clone
20:21:19kugeldz: there's a git repo that already has this done
20:21:24dz:<
20:21:25CIA-71New commit by pixelma (r21666): Manual: clean up some things related to the inclusion of the remotes. For button tables that use the 'Action' defines use HAVEREMOTEKEYMAP instead of ...
20:21:43kugeldz: http://www.rockbox.org/twiki/bin/view/Main/GitVersionControl#The_public_Rockbox_git_repositor
20:22:43JdGordonsweet, I get to blame the delta from my patch on mcuelenaere :)
20:22:44kugelI cloned that and got the full svn history
20:23:09mcuelenaereJdGordon: heh, no you don't; my patch doesn't change binsize ;)
20:23:16mcuelenaerewell not really 'mine'
20:23:20Bagderok, the ams bootloaders and mkamsboot are now placed on the download server
20:23:23JdGordonnot at all? :(
20:23:33Bagderbut the version numbers are going to look even crazier now
20:23:35 Join Rondom [0] (n=Rondom@dslb-084-057-179-018.pools.arcor-ip.net)
20:23:36mcuelenaereit shouldn't, only decrease plugin binsize
20:23:51AlexPpixelma: They should probably be moved to be after all the main unit ones and just before the description
20:24:21kugelBagder: I don't see then, is this due to the mirror business?
20:24:26kugelthem*
20:24:27AlexPpixelma: Otherwise any targets with the main unit keys opted after the HAVEREMOTEKEYMAP will have the main and remote keys swapped in the tables
20:24:36Bagderyes, it'll take up to an hour for the mirrors to show them
20:25:03pixelmaAlexP: how so?
20:25:17 Quit SirFunk (Remote closed the connection)
20:25:51pixelmamt: yes, still unrecognisable
20:25:56CIA-71New commit by gevaerts (r21667): bump sansapatcher version
20:26:03AlexPpixelma: If there is an \opt{HAVEREMOTE}{Remote key} then \opt{Some other keymap}{Main Key} Description
20:26:11Unhelpfulmcuelenaere: you can't set svn props in git. ask somebody who doesn't hate perl to fix that. ;)
20:26:23AlexPpixelma: Then it'll expand to Remote Key Main Key Description, and not Main, Remote, Desc
20:26:30*JdGordon curses the player target
20:26:31mtpixelma: ok. Thanks.
20:26:50AlexPpixelma: And I put the h100_remote_keymap opts right after the H100 opt
20:27:17AlexPpixelma: So if there are targets with remotes that have the key for that table opted after the h100 was, they will be in the wrong order
20:27:19Unhelpfulwhen adding files, i generate a patch from git and apply it to an svn checkout so i can add properties.
20:27:42AlexPpixelma: Anyway, I'll go through and check in a bit
20:28:04pixelmaAlexP: ah, yes. For targets with remote
20:28:44 Quit _Auron_ ("Infinity repeatedly denies rumours of plotting with zero to bring down the Universe.")
20:29:19pixelmahrrmm... forgot to mention that with this there are already some tables filled out in the Iaudio M3 manual...
20:29:34AlexPI'll just move the remote ones to the end
20:29:38AlexPThat way it is sure
20:30:03CIA-71New commit by jdgordon (r21668): fix yellow
20:30:19 Join matsl [0] (n=matsl@host-90-233-230-237.mobileonline.telia.com)
20:30:39rasherJdGordon: that's a really poor commit message
20:30:48JdGordonwhich one?
20:30:53rasher"fix yellow"
20:31:06rasher(and yeah, I know you're not the only one)
20:31:16pixelmaAlexP: so which tables will you change now? I ask because the ones that don't use \ActionSomething still need changing for IAUDIO_RC_PAD
20:31:17JdGordonmcuelenaere: you sure your patch doesnt affect bin size? there is no way mine could have done that much of a hit
20:31:35mcuelenaereaccording to Blue_Dude it shouldn't..
20:31:36JdGordonrasher: welcome to rockbox... you know we've been doing that sort of message for *years* right?
20:31:39kugelJdGordon: haha, it was so totally clear that this happens :D
20:31:52JdGordon?
20:31:54rasherJdGordon: What did I just say? Doesn't make it any better.
20:32:11rasherI think it's a very poor "tradition"
20:32:18kugelJdGordon: that a commit of yours clashes with the player
20:32:23JdGordonah
20:32:25AlexPpixelma: Definately the action ones - for the others do you think it is best to have all the remote ones together at the end, or to keep main and remote together?
20:32:35Blue_DudeYeah, I just noticed that the bins did grow some.
20:32:43pixelmaJdGordon, mcuelenaere: looks like BlueDude's change from the last time - as the Archos binsize hit is a lot smaller
20:33:08Blue_DudeWhat I'm not sure if is *why* they should grow.
20:33:14JdGordonyeah, less though but its still there
20:33:21Blue_DudeI mean, it's mostly moving code around, not adding it.
20:33:33JdGordonI dont care about it, just making sure I'm not going to have to revert because of it :)
20:34:01kugelBlue_Dude: take a peek at utils/analysis/bloat-o-meter.py or objdiff.py
20:34:02Unhelpfulrasher: but it's so thoroughly entrenched that as a newcomer to the project i felt kind of silly giving verbose descriptions of build fixes
20:34:39rasherUnhelpful: Yeah. I don't like it one bit.
20:34:42Blue_Dudekugel: if I had a clue what you just said, I might. I don't know how to use those tools though.
20:35:34kugelBlue_Dude: go back 1 revision, build the main binary (rockbox.elf), then to HEAD again and build the main binary, and pass both elf files to one of those scripts
20:35:48pixelmaAlexP: I think at the end is best there too (so it's column after column and it also leaves the possibility to combine them if thy use the same RC button)
20:35:56kugelthen you'll see what exactly caused the binsize increase
20:36:03AlexPpixelma: Yes, I think so too
20:36:20AlexPpixelma: I'll do actions first, then come to plugins later :)
20:36:34Blue_Dudekugel: over my head. I'll take a look though.
20:36:40mcuelenaerekugel, Blue_Dude: on it
20:36:58Blue_DudeThanks.
20:37:03mtsaratoga_home: I'll try using wma's bswap instead.
20:37:08 Join lilltiger [0] (n=lilltige@82.145.152.217)
20:38:00pixelmaAlexP: and I thought I could let you add the IAUDIO_RC_PAD there too while you are at it ;)
20:38:08AlexPWhere?
20:38:12AlexPplugins?
20:38:20pixelmayes ;)
20:38:31mtpixelma: One last try - http://pastebin.com/m16935b6f :)
20:38:35mcuelenaereBlue_Dude: http://pastie.org/534925
20:38:46AlexPhehe, I have h100 plugins to do first :)
20:38:53mcuelenaerebut fp_factor & fp_sincos aren't build for main Rockbox, are they?
20:39:39JdGordonlooks like they are
20:40:09*mcuelenaere wonders about the __cmpdi2 routine, looks like a floating point one.. ?
20:41:35pixelmaAlexP: that we don't work on the same files, I'm a bit confused what you are doing currently
20:41:47pixelmaor should I just wait for a commit?
20:41:49saratoga_homewhy is replaygain.c doing fp mul and fp divs anyway?
20:42:54Blue_Dudereplaygain need to calculate a gain factor from the replaygain decibel tags.
20:43:10saratoga_homewhy can't that be done in fixed precision?
20:43:26Blue_Dudefp is fixed point here.
20:43:42kugelmcuelenaere, Blue_Dude: fp_sincos is within #if (!defined(PLUGIN) && !defined(CODEC)) actually, so for core only
20:43:48AlexPpixelma: Nothing in actual fact
20:43:58mcuelenaerekugel: yes I know, that's why I'm wondering what's wrong here
20:43:59kugelfp_factor I mean
20:44:00AlexPpixelma: I was going to move all the action related ones to the end
20:44:09AlexPpixelma: But they are already there
20:44:11Blue_Dudekugel: right. Only replaygain needs it for now.
20:44:21mcuelenaerekugel: perhaps I switched the arguments to bloat-o-meter.py
20:44:30CIA-71New commit by gevaerts (r21669): fix bin2c.c path
20:44:41AlexPpixelma: I'll do plugins another day, so don't worry for now
20:44:51Blue_DudeBut replaygain had something like it already. Why the delta?
20:44:53kugelmcuelenaere: no, it would show a overall decrease then
20:45:05 Join rob [0] (n=rob@adsl-76-235-168-174.dsl.klmzmi.sbcglobal.net)
20:45:12saratoga_homeBlue_Dude: are you sure? I can't even tell what fp_mul is doing
20:45:15CIA-71New commit by gevaerts (r21670): fix bin2c.c path
20:45:35mcuelenaereBlue_Dude: replaygain actually shows a decrease in size (convert_gain)
20:45:48pixelmaAlexP: well, if I add the Iaudio one anyways... although moving is a bit more work than search and replace...
20:45:57Blue_Dudefp_mul is a macro. As far as I know it's not even used by anything. fp_mul64 is an internal macro in fixedpoint.c that is used.
20:46:10kugelI'm also unsure why fp_factor is so big
20:46:17AlexPpixelma: yeah :/
20:46:26AlexPpixelma: But sure, if you are going to add them :)
20:46:27saratoga_homeon the functions are defined out
20:46:33saratoga_homein that case maybe we could just delete them?
20:46:43Blue_DudeI did rewrite fp_factor. It wasn't legacy code. But it's not that big either...
20:47:16Blue_DudeYou can comment out fp_mul. But fp_factor is being used. I commented out everything else that's not in use.
20:47:20kugelprobably because fp_exp10 is inlined by gcc, and it contains lot of 64bit data
20:47:24Blue_DudeOr conditionally compiled them.
20:47:52Blue_Dudefp_exp10 was in replaygain.c before though.
20:48:09Blue_DudeMy version is actually a bit leaner. One less function call.
20:48:32kugelBlue_Dude: the old one doesn't have 64 datatyes though, it seems
20:48:55Blue_DudeWell I did rewrite it to be capable of more than a single fixed point precision though.
20:49:14Blue_DudeUm. I think it did internally...
20:49:19kugelyou should probably said that :)
20:49:22Blue_DudeCrud, I can't remember.
20:49:33pixelmamt: audio works now! :D but the playback time jumping is still there (makes the progressbar look... interesting too)
20:50:33kugelBlue_Dude: any idea what's the speed hit if this? I think we could live with 1.6K if it doesn't have a huge speed hit
20:51:05Blue_DudeThe speed should be fine. That's one reason I streamlined the exp10 routine.
20:51:09mtpixelma: Thanks ! :) \o/
20:51:19pixelmamt: guess that also has to do with missing seeking report?
20:51:47kugelsadly, there's no easy way to detect this :(
20:51:55gevaertsAnyone with an e200r around?
20:51:57Blue_Dudekugel: Also, fp_factor just isn't called that often.
20:52:04 Join stripwax [0] (n=dave@87-194-34-169.bethere.co.uk)
20:52:31kugelwhat's the gain of double precision here?
20:52:48***Saving seen data "./dancer.seen"
20:53:30mtpixelma: I don't think so .. I don't have a jumping progress bar like what you're mentioning .. does happen all the time or when playback is approaching the end ?
20:53:33Blue_Dudekugel: the old routine was locked into 12 bits of fractional precision. If that was raised much, it would overflow very easily. Exp generates bigs numbers.
20:53:47pixelmamt: all tj
20:53:56pixelmas/tj/the time
20:54:05saratoga_homeBlue_Dude: does anything need more precision?
20:55:30Blue_DudeI dunno. My original goal was to consolidate a library. The replaygain stuff was very useful, but locked into a specific implementation. I widened it some.
20:55:33mtpixelma: I guess I would be able to see than on a sim build, so I'll work on that later. Thanks a lot for your help.
20:56:21kugelpixelma: I figure you can test the PLA rework now? :D
20:56:45Blue_DudeI suppose I could go back and make it a fixed 12 bit routine to get the size back down. I'm not sure it would help much.
20:57:27pixelmamt: you're welcome. Hopefully this is really a problem that shows also in the sim... :)
20:58:02mtsaratoga_home: since wma's bswap.h is already optimised, should I point cook's bitstream to it or have another copy for cook ?
20:58:02Blue_DudeOne problem I found was that even with 12 bits, the routine would break - overflow with some inputs and give wildly incorrect results. Going to long long integers internally fixed that.
20:58:07mtpixelma: hopfully :)
20:58:09CIA-71New commit by mcuelenaere (r21671): Onda VX747 backlight: use a higher frequency to reduce flickering
20:58:29Blue_DudeIt's possible that we've been getting occasional bad numbers all along.
20:58:46saratoga_homemt: for now I would duplicate it
20:59:02pixelmakugel: trying out some plugins take a little bit more time to test though...
20:59:12Blue_DudeI don't believe I have to do this two days ina row, but I gotta bail to my paying job.
21:00
21:00:13kugelso it actually was calculating wrong regularly (not only rarely)?
21:00:17Blue_DudeIf there are other issues, I'll check the logs when I get home.
21:00:19mtpixelma: What's your target again ?
21:00:27saratoga_homemt: do you think they are called frequently enough to need further optimization? i admit i haven't looked at that file since we commited it
21:00:58pixelmamt: Iaudio M5 (my target from the coldfire family)
21:01:07saratoga_homekugel: hopefully plugin developers were aware their operations can overflow if they give very large values
21:01:31Blue_Dudekugel: yeah. Even though you thought you were giving it numbers in range, it would break internally.
21:01:32kugelreplaygain is a core thing
21:01:45 Quit r0b- (Read error: 110 (Connection timed out))
21:02:03kugelCorrect code is worth it I guess
21:02:04Blue_DudeGive exp10 2.0 and it would work. 2.4 would spit out a negative number.
21:02:07Blue_Dudeetc.
21:02:32kugelbut I think it should've been a seperate patch
21:02:34saratoga_homekugel: replaygain checks for overflow and clamps values to prevent it
21:02:53Blue_DudeNo it would break on in range numbers too.
21:03:09Blue_DudeCrud. No time. Sorry guys.
21:03:12 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
21:04:34 Nick hd is now known as HellDragon (i=jd@Wikipedia/HellDragon)
21:05:18mtsaratoga_home: I don't think further optimizations would make a significant change. I'm a bit in a hurry though, so I'll dig deeper later.
21:05:26saratoga_homekugel: see the convert_gain function in replaygain.c
21:05:33saratoga_homeok
21:05:34kugelsaratoga: the ams installers are on the download server now. I and funman planned to make a forum post in the "Official test build" forum (ie not put it under support for now)"
21:05:37saratoga_homejust curious
21:06:16saratoga_homeis swap32 assemblerized on arm?
21:07:32pixelmakugel: I agree with Llorean on the repeat for the quit button one. Some plugins on my Ondio (and on the c200) use it at least for an additional exit without menu (e.g. sudoku) and I don't find it inconvenient at all, easier than going through the menu or a button combo and you need the other buttons, can't remember accidentally shutting down because of it (hard power off is only after 10..15 seconds on both)
21:07:51kugel....
21:08:09kugelsoftware poweroff isn't always disabled
21:09:17CIA-71New commit by bertrik (r21672): S5L8700: implement kernel timer
21:09:49kugelok, then let's rate convenience over unexpected behavior.
21:09:55pixelmastill, no accidental shutdown for me and it's better than the alternatives. I don't know which current PLA plugins could be affected now though
21:10:13pixelmaI don't think it's unexpected
21:10:33kugelI can't say long power is expected to do anything but power of on a e200
21:11:27pixelmawell, the Ondio has a lot less buttons - and e.g. a stop in the WPS or the Radio is a long Off too (short Off for pause)
21:11:44pixelmathere
21:11:45kugelit has enough for my PLA
21:12:02gevaertsI've uploaded Portalplayer RC bootloaders to my webspace, and started a thread in Official Test Builds
21:12:28pixelmawhat do you need - four directions, select, quit (or a menu too?)
21:12:46kugelthese 6, yes
21:12:56pixelmaso no menu?
21:13:06kugelyes, no menu
21:13:23pixelmawhere would you put it if you need one?
21:13:39kugelput it onto quit, and have a exit item
21:14:17 Quit BHSPitLappy (Remote closed the connection)
21:14:38pixelmaas I said, an additional way to exit is quite convenient there and works nicely in some plugins
21:14:49CIA-71New commit by bertrik (r21673): Update initialisation and display of debug info in meizu M3 bootloader
21:15:11pixelmawell maybe that's non-PLA then
21:15:16kugelNo, it's ok, I'll do it. Still better than the current pla
21:15:27kugelWe'll get reports if it's such an issue
21:15:52pixelmayou don't have to use the repeat action, do you?
21:16:57kugelnot necessarily, no
21:21:14saratoga_homegevaerts: can you do sansapatcher too?
21:21:15kugelgevaerts: you don't plan to commit to the branch?
21:21:38gevaertssaratoga_home: sansapatcher should be in there
21:21:38gevaertskugel: commit what?
21:21:59kugelversion string maybe, I don't know
21:22:11gevaertsAll changes to svn code are committed. I put the version numbers in build/Makefile
21:22:34kugelhm
21:22:49kugelwe won't get a tagged release resulting in the same bootloader then, though
21:23:10gevaertsno. That's a problem...
21:23:20saratoga_homecan't we just hardcode the bootloader file?
21:23:30 Quit toffe82 (Remote closed the connection)
21:23:41kugelthat's what I did. I don't think it's a big problem
21:23:43Lloreankugel: On what target is software poweroff happening in plugins? I asked you this earlier and you told me it's "not relevant" but it clearly is if you're citing it./
21:23:44gevaertshm, since I now made all PP bootloaders 6.0, I could hardcode them
21:24:17JdGordonoh crud... does the manualk still have the config values listing?
21:24:37LloreanAnd why can't we make the quit shortcut work on those by having it quit, then if continued to be held for another long press, soft poweroff?
21:24:49kugelLlorean: run bubbles on the e200, and keep pressing stop ingame, for example
21:25:06CIA-71New commit by gevaerts (r21674): bump version for new release. No code changes
21:25:23pixelmaJdGordon: what?
21:25:24 Part stripwax
21:25:26saratoga_homespeaking of which, could we make the poweroff time a little shorter on the e200/e200v2?
21:25:34saratoga_homei often find myself letting go too soon
21:25:56Lloreankugel: And it can be disabled on that target without any serious repercussions. Is there anywhere it's actually *necessary* that it not exit the plugin first, like i suggested?
21:26:02saratoga_homeeven a half second shorter would be really nice i think
21:26:12kugelLlorean: it's fine, you don't need to discuss
21:26:21kugelI will do the CANCEL_REPEAT thing
21:26:21pixelmasaratoga_home: I'd say that's better than accidentally shutting down, and people are different
21:26:35pixelmasaratoga_home: I admit that I never tried an e200 though
21:26:40JdGordonpixelma: is this change ok? http://pastebin.com/m7b73966e
21:26:50saratoga_homeit seems worse to me on the e200v2 for some reason
21:27:03saratoga_homeas if the button sensitivity where different
21:27:23saratoga_homepixelma: its a fairly long delay on the e200, so i've never accidently shutdown
21:27:36saratoga_homemuch longer then i would naturally press a button for in the UI
21:28:11JdGordoncan someone reword the statusbar text in the manual for the behaviour please?
21:28:41CIA-71New commit by gevaerts (r21675): Set PP bootloader version to 6.0
21:28:54pixelmaJdGordon: just use "renote" for the opt, that's parsed from the languages' features.txt and is only valid for lcd remotes (as that's what it's used in the languages for)
21:28:59JdGordonhmm, actually I can maybe do it
21:29:00saratoga_homeyahhh new bootloaders
21:29:11gevaertskugel: better? ;)
21:29:17kugelhehe
21:29:47kugelI just don't think we need a super sophisticated system if we're doing a mere release in two years
21:30:09kugelbut JdGordon's idea was a good one, imo.
21:30:11saratoga_homewith any luck we'll only need one more release, and that will be if we ever get USB in the PP bootloaders
21:30:41gevaertskugel: we need a sophisticated system just because we only release every few years. If not, things *will* go wrong
21:31:20kugelMaybe JdGordon should extend his patch to bootloaders
21:31:34JdGordonI have a patch somewhere on the tracker to add a RB_RELEASE define which can be passed to configure...
21:32:05 Quit saratoga_home ("CGI:IRC (EOF)")
21:32:38kugelgevaerts: the patcher also have a version number, btw :)
21:33:08 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
21:33:13JdGordonpixelma: http://pastebin.com/m1e0175c0 ?
21:33:18kugelsansapatcher 0.8 is due
21:33:26gevaertskugel: I know :)
21:33:55gevaertsThat's why I updated those as well!
21:34:12kugelat least we got the stone rolling, it was kinda annyoing that we didn't manage to release new bootloaders
21:34:15JdGordonFS #10278 is my changes to the release scripts which might be useful for the bootloader
21:35:00CIA-71New commit by mcuelenaere (r21676): Lua: use rb->screens[] to do painting
21:37:24pixelmaJdGordon: seems better with the opt, just a typo in the item: Rremote - btw. I don't think I would put the current statusbar at the bottom...
21:37:27kugelsoo, let's wait for FlynDice's opinion and then post the AMS thread
21:37:53 Nick n00b81|afk is now known as n00b81 (n=taylor@unaffiliated/n00b81)
21:38:12JdGordonpixelma: some people might.... I figured it would be fun to add anyway seen as I was splitting the setting up
21:38:26 Join Lear [0] (i=chatzill@rockbox/developer/lear)
21:38:32JdGordonthat diff is OK to commit after the typo fix? I cant build the manual so have no idea if its correct
21:38:52kugelwe will see tomorrow
21:38:58pixelmaturning it off on the remote only xseems nice though
21:39:50pixelmaor a smaller font on the remote...
21:41:40pixelmaJdGordon: haven't tried compiling but don't see any obvious. It could be that the period would be after a space for non-lcd-remote targets
21:42:45JdGordonthere is no space after the word statusbar
21:44:02 Join lifeless [0] (n=lifeless@188.16.87.166)
21:44:07pixelmaI mean at the end of the description where " on the remote display" is opted
21:45:03JdGordonme too
21:45:36JdGordonshould i commit that?
21:45:55pixelmabut there is a line break and I think tex makes va space out of it (just one)
21:46:07pixelmas/va/a
21:46:13JdGordonoh
21:46:28gevaertsIs there a full diff of these manual changes somewhere? I can test-build if needed
21:47:03 Quit lifeless (Remote closed the connection)
21:47:16pixelmahttp://pastebin.com/m1e0175c0 (except the small typo)
21:47:19 Join lifeless [0] (n=lifeless@188.16.87.166)
21:47:24 Nick fxb is now known as fxb__ (n=felixbru@h1252615.stratoserver.net)
21:47:24 Quit lifeless (Read error: 54 (Connection reset by peer))
21:47:25JdGordonhttp://pastebin.com/pastebin.php?dl=m5cca685
21:47:26 Nick fxb__ is now known as fxb (n=felixbru@h1252615.stratoserver.net)
21:47:38JdGordonthat has the tpo fixed and hopefully no space
21:48:28 Quit lyngaas ("Leaving.")
21:49:00 Join lyngaas [0] (n=staale@19.81-167-149.customer.lyse.net)
21:49:14pixelmaguess the space between }{ is too much
21:49:33pixelmathat looks odd
21:50:37gevaertsthe pdf seems to look fine
21:51:16JdGordonpixelma: which one?
21:51:27*JdGordon is happy to leave this to you if you want it :)
21:51:40pixelmaon the line you just changed
21:51:47JdGordonoh remove the space?
21:52:36pixelmajust between \opt{remote} and the next {
21:53:09pixelmanot sure whether that's important though
21:53:11JdGordonok, done
21:53:16JdGordoncommiting....
21:53:40gevaertsJdGordon: gevaerts/e200.pdf">http://www.evonet.be/~gevaerts/e200.pdf (no remote) and http://www.evonet.be/~gevaerts/h120.pdf
21:54:14JdGordonta
21:54:24CIA-71New commit by jdgordon (r21677): fix the manual for the statusbar changes
21:57:07pixelmagevaerts: could you also try for a non-lcd-remote target?
21:57:30gevaertsgigabeat f ok?
21:57:41pixelmashould be
21:57:47kugelgevaerts: the ipod nano also doesn't have usb?
21:57:58 Join stoffel [0] (n=quassel@p57B4FBD8.dip.t-dialin.net)
21:59:29Mikachuwhat does "have usb" mean here?
22:00
22:01:02gevaertspixelma: gevaerts/gigabeat.pdf">http://www.evonet.be/~gevaerts/gigabeat.pdf
22:01:08gevaertslooks OK to me
22:01:24gevaertskugel: rockbox usb is disabled on all ipods
22:01:38CIA-71New commit by bertrik (r21678): s5l8700: fix off-by-one error in DMA count
22:01:44kugelah ok
22:01:59Mikachuonly in the releases
22:02:11gevaertsbut that doesn't matter anyway, since ipods handle usb-on-boot differently
22:04:39pixelmagevaerts: yes, although it made me notice that it's sometimes "statusbar" and then "status bar" :\
22:06:42 Quit Silicium ("leaving")
22:08:01 Quit _lifeless (Read error: 113 (No route to host))
22:08:55 Join toffe82 [0] (n=chatzill@ppp-71-138-19-215.dsl.frs2ca.pacbell.net)
22:09:07 Quit toffe82 (Client Quit)
22:10:41CIA-71New commit by bertrik (r21679): S35390a RTC: fix duplicate i2c initialisation
22:16:27markunbertrik: great work on the meizus!
22:16:35markunI'll get your M6SL tomorrow :)
22:20:10 Join AfterDeath [0] (n=icxcnika@freenode/weird-exception/network-troll/afterdeath)
22:23:33 Quit stoffel (Remote closed the connection)
22:31:26 Join thegeek [0] (n=nnscript@s243b.studby.ntnu.no)
22:37:10 Join petur [0] (n=peter@rockbox/developer/petur)
22:41:47 Quit KBH (Read error: 110 (Connection timed out))
22:45:41 Quit Lear ("ChatZilla 0.9.85 [Firefox 3.5/20090624025744]")
22:50:55pixelmahas anyone with an X5 ever had a problem that the tuner wouldn't work and which was fixed by a restart? It sounded like it actually was fixed at the lowest frequency no matter whether I went through the presets etc.)
22:52:51***Saving seen data "./dancer.seen"
22:55:40 Quit aaron424 (Remote closed the connection)
22:59:13kugelJdGordon: can you explain your objections against plugin-goto-wps now?
23:00
23:04:09bluebrotherurgh. The ability to put the statusbar on the bottom eats up 1k?
23:04:37 Join HBK [0] (n=hbk@pool-71-96-74-73.dfw.dsl-w.verizon.net)
23:04:50kugelno
23:05:16 Quit Rondom (Nick collision from services.)
23:05:32 Join Rondom [0] (n=Rondom@dslb-084-057-130-041.pools.arcor-ip.net)
23:05:38linuxstbbluebrother: It looks like the build was combined with the fixed-point lib...
23:05:52*linuxstb wonders how that can happen
23:06:10Unhelpfulis the "new" build system doing the builds yet?
23:06:16Bagderno
23:06:17 Join webguest89 [0] (n=566b3421@gateway/web/cgi-irc/labb.contactor.se/x-4fb9d4f30faa2b56)
23:07:04Unhelpfulthen they can still be combined if commits are close enough together.
23:07:46Mikachuthere was a commit at 17:59, both the other commits were done before that round finished
23:07:55Mikachuat 18:06 and 18:07
23:08:09webguest89uh, sorry to bother (and new to this IRC) but I've got a problem with the latest release. My iPod nano light doesn't turn on anymore after it turns off. And my blackglass theme is broken :( help.
23:08:19*bluebrother would expect combining the fixed point math routines in a lib would not create a red delta
23:08:37Unhelpfulbluebrother: putting them in core would...
23:08:37linuxstbBagder: Does the new system deal with rapid-fire commits better (perfectly?) ?
23:08:39kugelbluebrother: read the logs, it explains the delta
23:08:52*bluebrother puts some help in the channel
23:09:02bluebrotherUnhelpful: oh, they weren't in the core before?
23:09:03Bagderlinuxstb: so what is "perfect treatment" in that sense you mean?
23:09:11linuxstbBagder: I mean one build per commit...
23:09:18BagderI wouldn't like that
23:09:19linuxstbs/commit/revision/ (but same thing?)
23:09:37Bagderimagine 7 fast commits, it'll take ages until the guys get the results for the last
23:09:56linuxstbI would prefer that to not getting all the build info for each revision.
23:10:12Mikachu<insane idea>you could queue up the old commits, so when the latest finishes, you do the old commits and fill in the table when nothing more important is going on</insane idea>
23:10:53linuxstbBagder: So I guess you didn't design that feature into the new system (as you don't want it!) ?
23:10:53BagderMikachu: yeah, that _could_ of course be done, but I wouldn't consider that very fun to work on get going
23:10:59gevaertswebguest89: you mean the backlight?
23:11:11Bagderlinuxstb: correct, but it wouldn't be very hard to make it act like that
23:12:03webguest89yep, the backlight. sorry for ambiguous posting. it doesn't work properly.
23:12:18linuxstbBagder: It's not a big deal, it's just that I would have expected our build system to always do one build per revision...
23:12:22 Quit petur ("Zzzzz")
23:12:37*linuxstb stops assuming
23:12:41Bagder:-)
23:13:01Bagderwhen we get build times down to less than a minute I'll join that camp ;-)
23:13:24linuxstbEasy, drop support for swcodec...
23:13:24gevaertswebguest89: the hold switch isn't on?
23:13:40 Quit kugel (Nick collision from services.)
23:13:48 Join kugel [0] (n=kugel@e178109164.adsl.alicedsl.de)
23:14:21webguest89doesn't matter. ipod works fine. light doesnt until I reboot. Of course hold isn't on :)
23:14:37gevaertshm, weird.
23:14:40 Join wincent [0] (n=wincent@host-091-097-067-213.ewe-ip-backbone.de)
23:14:47webguest89It's a bug, I bet it is.
23:15:08MikachuBagder: i got another crazy idea, most build clients use ccache right? you could send all builds to all clients after the round finishes, just to prime their caches :)
23:15:15*bluebrother is really fond of "read the logs" if the question can be answered with a single sentence :<
23:15:20gevaertswell, yes, but I haven't heard about it before. This isn't the sort of bug that people don't notice
23:15:54BagderMikachu: haha. but seriously, this concept favors the same build to the same client on repeated builds anyway, to be ccache friendly
23:16:03Mikachuokay :)
23:16:27Bagderand all the killed builds will actually help with feeding ccache
23:16:36linuxstbwebguest89: Before you installed the latest release, when did you last install/update Rockbox?
23:17:19UnhelpfulMikachu: ccache can easily have poor performance anyway... since the key for matching files is the the preprocessed source, you just have to change a widely-used header file to get nearly everything rebuilt.
23:17:22pixelmawebguest89: also, did you check the backlight settings?
23:17:22webguest89uh... kinda half a year ago or more. But I checked. All new files are installed properly
23:17:36webguest89of course I did
23:17:52MikachuUnhelpful: yeah, that's why after such a change it helps to make subsequent builds faster if all clients did a build in between commits with the change
23:18:06*Mikachu sprinkles some grammar on that sentence
23:18:52webguest89in fact, when I turn it off from the settings and turn it back on again absolutely nothing happens. only solution is to leave it on all the time
23:19:17 Quit n00b81 ("Leaving")
23:19:21webguest89but the battery won't last long
23:20:21webguest89and I don't have a backup at hand (but I have home one). What do I do now? Leave it as it is?
23:20:59gevaertswebguest89: can you upload the contents of your .rockbox/config.cfg somewhere (e.g. http://pastie.org/)
23:21:32 Quit Rondom (Read error: 104 (Connection reset by peer))
23:21:36Unhelpfulhe could rename .rockbox/config.cfg and see if that fixes it...
23:22:28kugelbluebrother: I thought the logs explain it better, as it shows Blue_Dude explaining why he did what
23:23:02webguest89grrrr dunno, I could upload them on rapidshare. Or I could drag and drop them in my apache web server (don't ask). But then I'll have to tell you my IP and make sure I don't sturn off the computer until you grab them
23:23:18linuxstbwebguest89: Just upload it to http://pastebin.com
23:23:31webguest89doesn't work. tried playing around with that
23:23:32Unhelpfulor you can paste them into a web form and be done...
23:23:33Unhelpfulit's text
23:23:47linuxstbwebguest89: Then try a different pastebin - e.g. pastebin.ca
23:24:20webguest89wait a sec. I never used that before
23:25:47kugelUnhelpful: "mkamsboot Version unknown-090705", after make clean && make in the master branc
23:27:05kugelUnhelpful: I didn't know I need to pass something to it
23:27:22kugelit worked for svn, so I thought it was a git issue maybe
23:27:53pixelmaAlexP: at least the shifting around IRIVER_RC_H100_PAD made me find a bug in sliding_puzzle.tex... I also decided to make that two commits (shifting and then adding the Iaudio one)
23:28:27pixelmasorry, only called "sliding.tex"
23:29:14 Quit kugel (Nick collision from services.)
23:29:20 Join kugel [0] (n=kugel@e178120096.adsl.alicedsl.de)
23:29:55webguest89http://pastebin.com/d47965e1a okay, here are some. Might have gotten mixed up while I was panicking about my iPod being broken, but none work
23:29:56 Join stripwax [0] (n=dave@87-194-34-169.bethere.co.uk)
23:30:44pixelmaAlexP: I wonder though if and how we could make the switch between columns more visible
23:31:23gevaertswebguest89: so if you set the backlight timeout to e.g. 10 seconds, it turns off after 10 seconds but pressing a button doesn't make it turn on again?
23:32:01webguest89no. it doesn't turn on _anymore_ no matter what I try
23:32:56gevaertsHave you tried resetting the settings?
23:33:03pixelmawhat if you switch off the backlight fade in and out?
23:33:11webguest89but it stays on while it's plugged in my usb and appears as a disk drive. Yes, I've tried.
23:33:50webguest89none work. tried all settings. just doesn't turn on until I reboot
23:34:51 Quit bmbl ("Bye!")
23:37:12gevaertswebguest89: the "backlight fade in" and "backlight fade out" settings look weird. They should be "on" or "off", not a number
23:37:50webguest89I think it might be the firmware. And what I was trying to say above was, that perviously, when I plugged it in, the usb sign would show for a sec and then it would reboot in "file mode" (like all ipods do) And not with that stupid light on.
23:38:31 Join safetydan [0] (n=deverton@rockbox/developer/safetydan)
23:38:55webguest89fade in and fade out can have numerical settings since it's a matter of "in how many seconds will it fade out" I guess. I could be wrong, tho
23:39:16gevaertsWhich exact version are you running?
23:39:49webguest89latest. downloaded it just an hour ago from the site
23:39:54kugelgevaerts: they're numbers for ipods
23:40:09kugelfor all pwm fading targets, acutally
23:40:11gevaertskugel: ah. I didn't know that...
23:40:20kugelgevaerts: rtfm :D
23:40:58linuxstbwebguest89: Latest official release (i.e. v3.3), latest "current build" or latest "daily build" ?
23:41:17Lloreanwebguest89: "latest" isn't an exact version anyway. It changes every time a developer commits something, so can change sometimes a few dozen times in an hour during active periods.
23:41:51webguest89ummm... don't ask me, I dunno. I downloaded it manually from that page with many shiny players on it
23:41:58 Join Zagor [242] (n=bjst@46.35.227.87.static.tab.siw.siwnet.net)
23:42:09gevaertshave a look at System->Rockbox Info
23:42:47Lloreanwebguest89: Unfortunately, none of us can come over and look at your iPod. That means you're going to get asked a lot of questions that need specific and clear answers, because nobody else is having this problem, so we need to know a lot of details about your specific case and what might make it different.
23:43:09webguest89yup. r21672-090705
23:43:58webguest89I know, I know. Just wasn't sure where what "version" he meant
23:44:56 Quit bluebrother ("leaving")
23:44:57JdGordonkugel: i dont have objectsions... it just doesnt sit right with me
23:46:30kugelok, then I'll go for it
23:48:59Dhraakellianhmm... rockboxdev.sh is failing on me because "/usr/include/linux/errno.h:4:23: error: asm/errno.h: No such file or directory"
23:49:18Bagderwow
23:49:18*Unhelpful thinks that "make go-to-wps work in parts of rockbox implemented as plugins" is one good reason
23:49:47*Dhraakellian notices a bugzilla link in his google results that he hadn't previously seen
23:50:01webguest89oh, and btw, the fade out option doesn't work at all. forgot to mention. it just turns off suddenly. It went just fine before I did the update
23:52:23linuxstbwebguest89: I don't know if you've answered this question already, but have you tried resetting your settings? (there's a menu option in Rockbox to do that).
23:54:26webguest89In what rockbox? The utility? Though I did manually put a clean version and still the same stuff.
23:54:38Bagderhttp://daniel.haxx.se/blog/2009/07/05/concepts-of-a-new-distributed-build/
23:54:52linuxstbwebguest89: Yes, in Rockbox itself.

Previous day | Next day