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 2004-07-26

00:08:41 Quit unionpeak (Read error: 110 (Connection timed out))
00:10:09***Saving seen data "./dancer.seen"
00:10:29 Quit uski ("Leaving")
00:31:40 Quit NibbIer (Read error: 104 (Connection reset by peer))
00:31:41 Quit Threshold (Read error: 104 (Connection reset by peer))
01:24:56 Quit gromit`` (Read error: 110 (Connection timed out))
01:25:05 Part amiconn
01:31:30 Join unionpeak [0] (
01:38:02 Join dstar5 [0] (
02:10:11***Saving seen data "./dancer.seen"
02:15:41 Join NibbIer [0] (
02:20:42dstar5anyone have some rockbox movies to share?
02:34:19 Quit unionpeak (Read error: 110 (Connection timed out))
02:49:17 Quit elinenbe (" HydraIRC -> <- s0 d4Mn l33t |t'z 5c4rY!")
02:49:45 Join elinenbe [0] (
02:57:09 Join midk_ [0] (
02:57:47 Quit midk_ (Remote closed the connection)
02:57:49 Quit scott666_ (Read error: 60 (Operation timed out))
02:57:50 Join midk_ [0] (
02:58:53 Join scott666_ [0] (
03:01:48 Quit midk (Nick collision from services.)
03:01:53 Nick midk_ is now known as midk (
03:07:42dstar5lol <−− midk has quit (Nick collision from services.) those always get to me :)
03:08:17midkget to you?
03:08:32dstar5they are funy.
03:09:16dstar5midk, How do crazy people go through the forest?
03:09:22midknot sure.
03:09:31dstar5They take the psycho path.
03:09:40dstar5How do you get holy water?
03:10:01dstar5Boil the hell out of it.
03:10:49dstar5midk, i have a joke for you
03:10:56dstar5Two gay guys were in the shower together when one looked down and saw a puddle of white liquid.
03:11:02dstar5He said to the other man What did I tell you about farting in the shower?
03:11:11midkthat's just sick
03:11:48dstar5nasty nasty!
03:13:13dstar5time to go home
03:13:30 Quit dstar5 ("Leaving")
03:30:37 Quit NibbIer (Read error: 104 (Connection reset by peer))
03:39:09 Quit scott666_ (Read error: 60 (Operation timed out))
03:39:48 Join scott666_ [0] (
04:10:14***Saving seen data "./dancer.seen"
04:14:15 Join Treyqae [0] (
04:14:43 Part Treyqae
04:20:54 Join elinenbe_ [0] (
04:24:18 Quit midk ("Leaving")
04:38:43 Quit elinenbe (Read error: 110 (Connection timed out))
04:38:43 Nick elinenbe_ is now known as elinenbe (
05:09:23 Quit AciD ("")
05:14:18 Join NibbIer [0] (
05:37:39 Quit scott666_ ("i'll be back...eventually...")
05:37:39 Quit NibbIer (Read error: 54 (Connection reset by peer))
06:10:18***Saving seen data "./dancer.seen"
06:37:50 Join elinenbe_ [0] (
06:41:37 Join midk [0] (
06:45:47 Quit elinenbe (Read error: 60 (Operation timed out))
06:45:48 Nick elinenbe_ is now known as elinenbe (
07:15:01 Join NibbIer [0] (
07:31:44 Join LinusN [200] (
07:31:45 Quit NibbIer (Read error: 54 (Connection reset by peer))
07:33:45midkhey LinusN
07:33:54midki noticed a small error in your cross-compiler doc
07:35:00midkearly on you say you need "gcc-3.4.1.tar.gz (find it at your closest GNU FTP site)", but under "Build GCC" you say "/home/linus/gcc> ../../gcc-3.0.4/configure −−target=sh-elf −−prefix=/home/linus/sh1 −−enable-languages=c"
07:35:45midkapparently, you made a version typo with gdb too.
07:40:47LinusNthanks for noticing
07:40:59midki forgive you. :)
07:41:15midki'll have to try it again - it didn't work for me the first time around
07:42:01LinusNbuilding the cross compiler?
07:42:25LinusNwhat went wrong?
07:42:47midkeverything seemed to go just fine.. rockbox just didn't build. iirc, it kept re-making apps
07:43:07midki tried it again, and that time there was no sh-elf-gcc to be found
07:43:29LinusNyou are aware that you have to chagne the PATH?
07:43:35midkyeah.. 'leaving directory 'apps'' then right afterwards, 'entering directory 'apps''
07:43:48midkyeah, i did..
07:44:00midkwith your export PATH line, no?
07:44:17LinusNyeah, or add in in .bash_profile
07:44:23midkdid both.
07:44:39midkin the /home/midk/sh1 folder i found a folder 'sh-elf'
07:44:58midkbut there was only 'gcc' in there,
07:45:59LinusNthere should be a "bin" directory too
07:46:21midkyep, there was
07:46:32midkalong with about 6 others
07:46:42LinusNthe PATH should point to the bin dir
07:47:01midkexport PATH=/home/midk/sh1/bin:$PATH ?
07:47:08LinusNperhaps you forgot to "make install" gcc?
07:47:15midkno, i did that too.
07:47:36midkoh well. i'll try it again soon
07:49:48LinusNdid you use the same prefix for both binutils and gcc?
07:50:18midkyeah. could have been a typo but i'm quite sure i did it right
07:50:28midki'll try it later tonight or tomorrow
07:53:00midkall right. time to reboot and, hopefully, successfully resize some partitions. be back sooner or later.
07:53:14 Join Guest [0] (
07:53:34LinusNmidk: may the source be with you
07:54:04midk(it's an iso) :)
07:54:13 Quit midk ("Leaving")
08:02:29 Join midk [0] (
08:05:09GuestHi midk - got to play with the stuff we did last night. Works great.
08:06:04GuestGot ID3v2.3 reading and writing working today, so that piece of the puzzle is a no-brainer.
08:06:35midkoh, good
08:08:57*LinusN is curious
08:09:47midkwe were talking about a feature in f2 or f3 that would add your currently playing file to one of five playlists. the we could have a 5-star rating system like the iPod.
08:10:06 Join amiconn [0] (
08:10:19***Saving seen data "./dancer.seen"
08:10:57amiconnG'morning everyone
08:11:04LinusNamiconn: hola, amigo
08:11:17GuestThanks midk
08:11:43GuestDownloaded the devkit setup from Cyborg systems - looks like an easier way to set up the dev environment
08:11:43midkGuest, for?
08:11:51LinusNmidk: how about extending the "add-to-playlist" plugin so that you get to choose the categories/playlists via settings?
08:11:52amiconnLinusN: Although I like the new soft shitdown feature, I have to tell you about 2 quirks I found
08:12:07LinusNshitdown, thanks a lot :-)
08:12:09GuestResponding to LinusN faster than I could..
08:12:35amiconnoops, shutdown of course :-)
08:12:50midkGuest, if you were serious, np :)
08:13:00GuestAnyway, is Cyborg's a good way to set up the dev?
08:13:10LinusNit works, afaik
08:13:42amiconn(1) While the sutdown is correctly suppressed from OFF-OFF if the charger is connected, it is not if triggered from the menu
08:13:50amiconn*shutdown, grr
08:13:56LinusNamiconn: not?
08:14:36amiconnIt says "Shutting down...", then goes back to the menu
08:14:39LinusNamiconn: ah, my bad
08:15:05LinusNi put the charger check too late in the code
08:15:22LinusNhmmm, i had a reason to do that, but i can't remember what it was...
08:15:30amiconn(2) Why calling "Shutdown" triggers a disk spinup first if the disk was already quiet is beyond me
08:15:45LinusNbecause it saves the settings
08:15:55LinusNwhat else would a clean shutdown do?
08:17:02LinusNi could of course add a check for changes settings...
08:17:12amiconnIt does that also when I just switch on the unit, wait until the disk spins down, then hit OFF-OFF. No setting changed in between...
08:17:22LinusNsee above
08:17:40GuestDifferent topic: I was playing with the voice feature, got the Crystal voice working for menus. I used joerg hohensohn's vbs script to generate directory names. It worked but it used Microsoft's Sam voice. What's the right way to get some other voice to work with the script?
08:17:41amiconnOk, thanks.
08:18:29amiconnGuest: The script uses whatever defaulkt voice is set in your windows speech control panel (as long as it is an Englsih one)
08:18:31midkLinusN, so holding off isn't really supported anymore?
08:18:56midkor is safe shutdown "in addition to" holding stop/off?
08:19:11LinusNmidk: we can't do anything about holding off, it's hardware controlled
08:19:26LinusNthat's the whole thing
08:19:37midkright. but i'm saying, you used to detect an off hold before you merged safe shutoff, right?
08:19:54LinusNotherwise i wouldn't have to goo all this way to put a clean shutdown on the double-off click
08:20:21LinusNwe detect a held off key on the fm, yes, and that is still there
08:20:25midkthe hardware off on the FM is five seconds. but if you held it and rockbox was operating fine it was only 2-3 seconds, doesn't that mean that rockbox controls poweroff?
08:20:56midkyou should show the logo and "shutting off" text as the disk spins and saves settings. if it doesn't already.
08:21:10LinusNit does
08:21:18LinusNbut not when you hold off
08:21:32midkof course not. but also that would be a good feature on the FMs.
08:21:42midksince we do control poweroff there
08:21:55LinusNyes, but it requires MAJOR changes to the code
08:21:58amiconnLinusN: Another small one for the safe poweroff: For holding off, it says "Battery: charging" if the charger is connected. It does give no feedback at all if you click OFF-OFF
08:22:19LinusNi should probably say the same thing
08:22:36midkoh, i once wrote a patch that, when you hold off in the dir browser with charger plugged it went to the charger screen - imo this was a good idea. do you agree?
08:23:43LinusNwhy is it a good idea?
08:24:17midki just thought it was.. although it'd appear that when i see a good idea it's usually a not-so-good one
08:24:27Guestamiconn, that's what I thought. I loaded some other voices and the SAPI control panel add-on (spchapi.exe). The other voices show up as SAPI 4 voices in the "Other->SAPI 4 Control Panel" pane, but there's no way to change the default voice. I'm on XP - what am I doing wrong?
08:24:31amiconnI'd agree to midk, both shutdown methods should go to the charging screen if you use them with the charger connected. It would look much more logical to the average user
08:24:54amiconnGuest: The script is only able to use SAPI 5 voices
08:24:57LinusNmidk: there is no difference between charging in the charging screen than in the "normal" mode
08:25:01midkyou could even extend that idea to idle-poweroff. if the idle time reaches the set poweroff time it would "power down" to the charging screen.
08:25:11midkLinusN, it's just, like amiconn said, a cosmetic thing.
08:26:25midki think holding off should result in the charging screen instead of "battery charging" - it just seems more logical
08:27:16amiconnIn adition, it would protect from accidentally pressed keys (other than ON)
08:27:29*LinusN is experiencing a serious deja-vu
08:27:59LinusNwho pushes keys by accident when charging?
08:28:27*midk looks at amiconn
08:29:58GuestOK - that explains a lot. Is the only way to get other SAPI 5 voices to download the whole SAPI 5 SDK (132MB)? I downloaded the Mike and Mary redistributable, but the exe download only contains an msm installer merge module, no msi install. What a pain.
08:30:01amiconnI didn't say that it happens to me (I even don't use keylock at all), but it may happen to others. Of course the "risk" is low :-)
08:30:45amiconnGuest: Iirc, unfortunately you have to download the whole SAPI 5 SDK
08:30:53LinusNwe once rebooted when off/stop was held
08:31:01midki simply see "charging screen" replacing "powered off" status when charger is plugged.
08:31:12midkidle poweroff or shutdown would lead to it.
08:31:22midknot necessarily idle poweroff, though.
08:31:27midkactually, no, not idle poweroff.
08:31:32LinusNabsolutely not
08:31:54midkquiet, it was just a stroke of brilliance :)
08:32:14LinusNfelt good, didn't it?
08:32:14GuestOK, thanks. Don't mind doing it, so long as I know it's worthwhile and not going to mess up the voice stuff that already works. D/Ling now.
08:32:45midkLinusN, yeah. but then i realized it'd be rejected because some people may actually use this thing to play music.
08:33:50Guestbrb - D/L hogging my router...
08:35:08amiconnGuest: I *could* extend the script to be able to use SAPI 4 as well the same way as my voice file generation script does, but SAPI 4 is slooow from a script, has to do everything in realtime
08:37:07amiconnLinusN: Still I think going to the charging screen if a poweroff event is triggered is most logical to the user: If he hits OFF, he wants the unit to shut off, so going to the charging screen instead if the charger is connected is "as OFF as possible" in that case
08:38:18amiconnYou have to check for that anyway (saying battery: charging" atm), so code size increase should be neglectible
08:40:36midkright, that reminds me, why do you have a seperate function for simply splashing "Battery: Charging"?
08:42:57LinusNthat's because we don't want to enter the charging screen, but people kep asking why it didn't "shut off" when the charger is attached
08:43:31midkright, but why a seperate function?
08:43:39LinusNso the "charging" splash informs the user that it is charging and there is no need to go to the charging screen
08:43:42midkyou call it, instead you could just call splash instead
08:44:04LinusNyes, but i think there were plans to extend it
08:44:25LinusNand it might be called from several places
08:45:12midkwell i think when a user holds off with charger plugged they probably want it to turn off.. isn't the charging screen as off as possible?
08:45:45LinusNi guess it is
08:49:31LinusNdamn, the exit-from-viewer poweroff bug is hard to fix
08:50:45midki have an idea. let's give up!
08:50:53*LinusN has a better idea
08:51:06midklet's FIX IT!
08:51:26amiconnLinusN: Empty the button queue when a plugin has exited? (Of course this would prevent plugins from deliberately injecting button events on exit, but no plugin does that so far)
08:52:41LinusNit already does
08:52:43GuestGot the 5.1SDK, it works - thanks, amiconn.
08:52:56LinusNthe problem is that the plugin exits before the user has released the key
08:53:26amiconnAh oops, yes
08:53:46LinusNi have solved it now
08:54:03LinusNi don't have to act on the release event
08:55:05 Quit c0utta (Read error: 60 (Operation timed out))
08:56:32LinusNhmmm, i had a reason to do that in the beginning of the development of this patch...
08:56:33amiconnLinusN: I thought this is necessary to distinguish a short press from a long one?
08:57:38LinusNyes, but i wonder why i wanted that
08:59:07LinusNhmm, i have an idea
08:59:18amiconnHmm, imho it shouldn't be necessary. If you react on keydown, the poweroff? question would also kick in when the user holds down OFF, but that should make no difference, since the hw powerdown catches it anyway
08:59:58LinusNi should just ignore the OFF repeat event, so the splash isn't cancelled
09:00:23LinusNand on the FM, i could save the settings when holding OFF as well...hmmm...
09:00:30amiconnPerhaps it is even better to react on keydown, because the pop-up question may remind a user to use OFF-OFF now instead of holding OFF
09:01:10amiconnOn the v2/fm, holding OFF should also trigger the safe poweroff
09:01:19LinusNyes, but that's hard to implement
09:01:21midkthat's what i suggested!
09:01:27midkand that's the response i got.
09:01:31midkcarry on.
09:01:47LinusNthe poweroff is triggered by the button driver, in the interrupt context
09:02:58midkyes. the interrupt. and the context. triggered. driver. yeah.
09:03:30amiconnPerhaps convert it to a special button event (BUTTON_POWEROFF) and place it into the button queue?
09:03:43LinusNof course
09:04:08LinusNbut that will force us to change all button_get() loops
09:04:17LinusNlike for the USB connection
09:04:54LinusNbut that may be a Good Thing in the end
09:05:33LinusNbecause then i finally have a reason to implement the system_handler() idea
09:07:19LinusNin all button loops, add a call to system_handler() in the default: case
09:07:35 Join c0utta [0] (
09:07:36LinusNwhich checks for USB and for the OFF event
09:07:55amiconnThat could even save some code space...
09:08:03LinusNprobably, yes
09:08:10amiconnSounds like a good idea
09:08:17LinusNi'll do that
09:10:13amiconnAnother option would be to implement something like an event system
09:10:53amiconnEach thread/ plugin may "register" for those events it wants to handle itself.
09:11:27amiconnAll other events would be handled by the thread that registered it earlier
09:12:10LinusNproblem is that the usb event should be handled by the same thread
09:12:45amiconnEvents/ event groups could be EVENT_BUTTON, EVENT_USB, EVENT_POWEROFF, EVENT_TIMERTICK etc.
09:13:19amiconnYes, occuring every 1/100 s
09:14:33amiconnIt may even be possible to generate some "soft" events to register for plugins that are interested in them. EVENT_MPEG_TRACKCHANGE comes to mind
09:14:46LinusNwhy timertick?
09:15:45 Join NibbIer [0] (
09:16:02amiconnTimertick may be of interest for plugins.
09:16:37LinusNmaybe we should export tick_add_task()
09:16:37LinusN instead?
09:19:02amiconnMaybe (didn't know about this function until now!). Imho an event system would be an clean & elegant replacement for a number of api functions
09:20:11LinusNevents, as in pseudo button codes?
09:20:16LinusN(i hope)
09:21:20amiconnIt would work in a similar way as the button codes do now.
09:21:23LinusNhaving timer ticks posted to the button queue will probably cause problems with overflown button queues
09:21:45amiconnIt could work like this:
09:21:58amiconnWe have a global event queue.
09:22:27amiconnAny thread, irq routine etc. that generates events posts them there
09:22:38LinusNin addition to the button queue?
09:23:02amiconnNo, instead of it, since the button events would be part of the event system then
09:23:41LinusNwhat if noone reads the event queue?
09:23:58amiconnAny thread that is interested in them would "register" for them (e.g. with a callback function)
09:24:08LinusNbutton events too?
09:24:35amiconnThe receiving of events could be destructive or non-destructive
09:24:46amiconn(decided by the thread that wants them)
09:25:10LinusNdestructive, as in removing it from the queue?
09:25:49LinusNlet's say, for the sake of argument, that we "register" for the timer_tick event
09:26:02amiconnThe default registration (by the initial thread) for all events would be to discard them all
09:26:06LinusN(and the obvious button event)
09:26:34LinusNthen we handle a button event which triggers a disk read
09:26:36amiconnThen the usual threads of rockbox would register for those events that they want to receive
09:27:05LinusNduring the disk spinup we receive between 200-500 timer events
09:27:49LinusNwhich are stored in the queue because we are busy waiting for the drive
09:28:30LinusNmost events are only interesting for the main thread
09:28:31amiconnTriggering a disk spinup doesn't cause a yield()?
09:29:21LinusNof course it does, but should the main thread throw away the timer tick events?
09:29:38LinusNin every sleep(), yield() etc?
09:29:50LinusNthen you can't depend on them for timing
09:30:04LinusNmy point is that the timer tick events are useless
09:30:56amiconnMaybe that timer events are useless, this was only one suggestion of an event
09:31:04LinusNand i don't like the registering scheme, because it clutters up the code
09:31:25amiconn(Sorry, gotta hurry)
09:31:33LinusNok, cu around
09:31:48LinusNTRACK_CHANGE is a good event, though
09:32:15LinusNwe wanted to make an event of it in the beginning
09:32:33LinusNbut code structure issues made it a little difficult
09:32:57LinusNso i like the event idea
09:33:12LinusNbut i don't want the registering scheme
09:34:50LinusNi should document the kernel in wiki...
09:40:30 Join Doehni_ [0] (
10:10:21***Saving seen data "./dancer.seen"
10:15:57 Quit NibbIer (Read error: 104 (Connection reset by peer))
10:15:59 Join Ich|bett [0] (
10:16:14 Nick Ich|bett is now known as NibbIer (
10:34:26LinusNi'm implementing the default_button_handler() as we speak
10:34:32amiconnLinusN: Iirc the mpeg thread uses something like an event system internally
10:34:52LinusNyes it used a message queue
10:35:15LinusNi think we should use the existing button queue for events
10:35:33 Quit Guest (Read error: 110 (Connection timed out))
10:36:06amiconnThe registering scheme would have the advantage that it is extendable without touching things that wouldn't have use for the new events
10:36:38LinusNif the button switch doesn't handle it, it ends up in the default case
10:37:07LinusNso you won't have to change anything
10:37:10amiconnThe system_handler approach does this as well, yes
10:37:23LinusNit works even without the default handler
10:37:33 Join uski [0] (
10:38:21amiconnThe current button queue isn't exactly a pure button queue anyway, as it passes e.g. SYS_USB_CONNECTED
10:38:56LinusNin fact, every thread is supposed to have a queue
10:39:08LinusNand the main thread queue is the button queue
10:39:30LinusN(well, every thread that is interested in the SYS_USB_CONNECTED event, that is)
10:40:19LinusNthe SYS_USB_CONNECTED event is broadcast to all queues
10:41:07amiconnThe registration scheme would have the advantage that a thread/ plugin doesn't receive events it is not interested in
10:41:23LinusNwhy is it important that it doesn't receive them?
10:41:58amiconnThere wouldn't have to be a default case calling the system_handler
10:42:19LinusNit works even without the system handler
10:43:20LinusNi see what you mean
10:43:33LinusNyou expect some magic to take care of the USB_CONNECTED event?
10:44:01LinusNthe main thread must handle the USB event
10:44:12LinusNbe it in the default handler or anywhere else
10:44:50LinusNand the plugin runs in the main thread
10:45:16amiconnYes, of course one thread has to handle it. But if a plugin doesn't use the hd, it could ignore the USB_CONNECTED event, and go on executing.
10:45:18LinusNall threads must respond to the event if they have a message queue
10:45:46amiconnOf course in order to exploit the advantages of an event system, plugins should always run in an own thread
10:46:10LinusNand what does the main thread do in the mean time?
10:46:52LinusNin your example, you would enter USB mode with the plugin still running?
10:47:10LinusNkind of confusing to the user, don't you think?
10:48:01LinusNi think you over-complicate things
10:51:26amiconnMaybe that a full event system isn't necessary atm, but it may be the case if Rockbox gets ported to another player one day
10:51:44LinusNwhy would it?
10:52:50LinusNimho, the only case when a registration would be necessary is when there is a risk of flooding the queue
10:53:17LinusNlook at the Windows message system for example
10:53:47LinusN(i can't believe i said that, using windows as a model example)
10:54:05LinusNyou get tons of messages that you don't "want"
10:54:17LinusNand they are all taken care of by the default message handler
10:55:04amiconnI didn't deal much with Windows messaging yet
10:55:56LinusNwell, it works much like rockbox today
10:56:14amiconnOf course you are right if there is always a default handler handling all events you do not want to deal with yourself
10:56:29LinusNyou get messages in a queue, and you deal with those you are interested in, and pass the rest back to the system default handler
10:56:46amiconn..only needing to register for events that may otherwise flood the queue (e.g. timertick)
10:57:03LinusNmuch simpler code
10:57:20LinusNand you don't need to bother with unregistering all the time
10:57:27amiconnBut a full event system may give you some advantages
10:57:41LinusN"full" == register/unregister?
10:58:20amiconnUsing one global queue (well, it would be more complex than a simple queue then)
10:58:46amiconn- The ability to handle events with or without removing them from the queue
10:59:20LinusN... with lots of problems when several threads decide to to a "destructive read"
10:59:24amiconn- The ability to inject events as well (possibly using this system for the mpeg thread as well)
11:00:03amiconn- The ability to define priorities, i.e. with thread should receive an event first
11:00:21LinusNwhat is it that you gain with this approach?
11:00:36LinusNi see nothing but compilcated code
11:03:41LinusNfor example, which thread is responsible for removing the events?
11:04:16LinusNyou would probably need to have a separate thread with the lowest priority, to mop up the unhandled events
11:04:34LinusNand when should it do it?
11:05:04LinusNa thread could be in sleep() at the time, and miss the event
11:10:39amiconnIf an event occurs that is registered by no thread, it should simply be discarded
11:11:32amiconnOr, if there is no registration scheme for simple events, it has to be removed at the lowest priority, yes
11:12:18LinusNso with the global queue, registration will probably be a necessity
11:12:59LinusNok, so which thread is responsible for removing the event from the queue?
11:13:03amiconnThe sleep() approach wouldn't work anymore like it does now. It could be replaced by a one-time registration for a wake up event.
11:13:33uskior maybe you could use some kiond of "interrupt" system ?
11:13:52amiconnSo, either this or another event for that thread would break the sleep()
11:13:52uskisleep() would be running, but a event handler would be called in the thread, as an interrupt vectore
11:14:09LinusNback to the main question, what does this approach buy us?
11:17:26amiconnImho it may provide more flexibility
11:17:35LinusNfor what?
11:17:46LinusNwhat flexibility are we missing today?
11:17:51 Quit NibbIer ()
11:20:10 Join [IDC]Dragon [0] (
11:20:50amiconnImho it may make extending the system easier. For instance, if you introduce a BUTTON_POWEROFF now, and a plugin doesn't handle that in its default case, what does happen?
11:20:52[IDC]Dragonhi guys
11:21:12LinusN[IDC]Dragon: hi
11:21:31[IDC]DragonI peeked in the logs a bit
11:21:38LinusNamiconn: it will not save the settings before shutting down
11:21:54[IDC]Dragonprobably I come too late
11:22:00midkhm, probably a horrid idea.. but... what about a function for plugins that saves the screenbuffer to a char so you can recall it later - sort of the opposite of lcd_bitmap?
11:22:30LinusNmidk: do you need it?
11:22:35[IDC]Dragonare you trying to get rid of the USB check everywhere, or what?
11:23:06midkLinusN, it'd prove useful in my opinion. and yeah, i'd use it in my rockblox update more than likely.
11:23:12midkfor a confirmation to restart screen
11:23:12LinusNwell, amiconns idea might head in that direction
11:23:42midkvoid lcd_savebitmap(unsigned char *location); ?
11:23:46LinusNmidk: but can't you just redraw the screen?
11:24:05midki have to clear it to show my screen, and rockblox doesn't remember the position of the blocks
11:24:21LinusNthen have it remember them
11:24:39midkor i could just forget the confirmation screen.
11:24:40LinusNhow does it check the collisions then?
11:24:57LinusNfor a complete row
11:25:06midkerm, it doesn't keep them drawn.
11:25:22LinusNyes, but how does it know that it has a full row?
11:25:54midki'm actually not sure, i never understood "to_virtual" and "from_virtual" which is involved in the process
11:26:14midkand i don't feel the need to learn it. i really didn't expect anything more than "no"
11:26:28midki was just looking to be rejected i guess :)
11:26:29LinusNimho, if you need to save and restore the screen, you might need to rework the code
11:26:37midkright. i don't want to do that
11:26:42midksaving a bitmap would do fine but..
11:26:47midkthere is no such function.
11:27:19LinusNi sound like a jedi:-) "you don't want to sell me death sticks"
11:29:18LinusNamiconn: which problem can't be solved with the current event approach?
11:37:14[IDC]Dragonor is this some poweroff concept?
11:37:24*[IDC]Dragon is off track
11:39:06LinusNwell, i am restructuring the SYS_USB_CONNECTED handling
11:39:39LinusNso all button switch loops call default_button_handler() in the default case
11:39:52LinusNinstead of handling the SYS_USB_CONNECTED event
11:40:18LinusNthis will later on allow for a SYS_POWER_OFF event
11:40:40midkit would seem my shout outs are not needed. so i will stop shouting out.
11:40:43LinusNthe discussion kind of spun off from there
11:41:07LinusNmidk: the project depends on your shout outs :-)
11:41:32 Join HenrikB [0] (
11:41:44LinusNwooo, HenrikB!
11:41:58LinusNwhat an honour
11:42:18LinusNif you are Henrik Backe that is :-)
11:42:20[IDC]Dragonimho it would be great if this USB "exception" is transparent to the application world, but this perhaps has been discussed to death
11:42:53LinusNyes, but it is hard to hide alltogether
11:42:53*[IDC]Dragon should make myself more rare, to be greeted like such
11:43:09midkLinusN, :) ok i'll continue
11:43:22HenrikBSure am, I'm on vacation so know I'm able to chat during the day.
11:43:30midkyeah. stay with us HenrikB!
11:43:36LinusN[IDC]Dragon: there are two issues with USB mode:
11:43:55midkyeah.. usb mode.
11:44:09LinusN1) we must respond to the event and show the USB screen
11:44:28LinusN2) we should reload the browser buffer afterwards
11:44:48LinusNissue 1) is hidden in the default handler
11:44:56LinusNissue 2) is not
11:44:59midkyeah. handle that.. handler.
11:45:02midkyou go linus
11:45:16[IDC]Dragon2) could be a callback, event, whatever
11:45:44LinusNyes, today all functions return a SYS_USB_CONNECTED value, or "true"
11:46:01LinusNmaybe the dir browser could check a flag instead
11:46:52[IDC]Dragonperhaps 3) messing with a file/dir which is in use by the box is more of a problem
11:47:38[IDC]Dragonwhat if I delete the file/dir currently playing?
11:48:26LinusNwell, then it resumes with the next file in the playlist
11:49:51[IDC]DragonI was just dreaming: in USB mode, all threads could carry on until they do disk I/O, which would cause the OS to suspend them. After USB removal, they can proceed and won't know
11:52:23LinusNinteresting idea
11:53:11LinusNalthough the main thread would still have to act on the event, showing the usb screen
11:53:24midkyay, usb mode.
11:53:34[IDC]Dragonscreen ownership is an issue, because of the USB logo
11:53:57LinusNyes, that's why the main thread does it
11:53:58[IDC]Dragonif we reqally need one, that is
11:54:27[IDC]Dragonit could be a little status bar icon like the plug
11:54:28LinusNwe need a good confirmation that rockbox has released the drive
11:55:14LinusNand what do you want to do in rockbox while in USB mode?
11:55:19LinusNplay a game?
11:56:21[IDC]Dragonbut my main thinking is if we can make application land simpler
11:56:46midkis it possible to flash the red led on hd access even in usb mode?
11:56:47midkit should be.
11:57:13LinusN[IDC]Dragon: actually, most of the USB application problems are because the dir browser wants to know that the buffer needs reloading
11:57:21LinusNmidk: no
11:57:46 Join NibbIer [0] (
11:58:03LinusNand because we want to go back to the initial state after USB mode, to be safe
11:59:09LinusNso all functions are expected to return after being in usb mode
12:01:39LinusNand most of the time, we want to know if it returned because of USB mode
12:05:15amiconnThere could be an event SYS_USB_DISCONNECTED the same way as there is SYS_USB_CONNECTED now, triggering the dir browser to reload its buffer
12:06:25HenrikBI've got a question, is mkdir("/.rockbox") added to the startup yet?
12:06:30amiconnAnd, iirc a plugin doesn't need to return if usb is connected, it could continue after the usb_screen() call returns
12:06:43amiconn(Off to lunch)
12:07:15LinusNHenrikB: no
12:07:25LinusNand i think it shouldn't do that
12:07:45LinusNit should instead warn the user that he hasn't installed rockbox properly
12:09:12LinusNamiconn: hmmm, interesting idea
12:10:22***Saving seen data "./dancer.seen"
12:12:34uskia question: anyone is used to go to club ? if yes, do you know how much they sell glowsticks ?
12:13:03HenrikBIf there is no .rockbox the playlist returns some nasty errors,
12:13:40LinusNHenrikB: yes, that's why we should warn
12:14:08LinusNthere are more things that won't work, like the viewers, rocks etc
12:14:18HenrikBalso if there is an ampty .rockbox dir all the essential function would work
12:15:09LinusNwhy would the user want a crippled installation?
12:17:09HenrikBMy thinking is that a crippled installation is better than non working installation
12:17:54LinusNperhaps, but i still think the warning is in place
12:18:09LinusNit could do both, but then the warning will not reappear
12:19:02HenrikBSure is, and a big red one it should be, but if we check for the existence of .rockbox
12:19:08LinusNuski: i have no idea what glowsticks are, so my answer is no :-)
12:19:11HenrikBwe might as well create it.
12:19:37LinusNthat's probably the best
12:19:49LinusNHenrikB: you fix?
12:20:20LinusNa warning + create the dir
12:20:53HenrikBI'll do it, but not until next week, going to denmark for the rest of the week.
12:21:04LinusNok, i'll do it then
12:21:05 Join lImbus [0] (
12:21:36uskiLinusN, glowsticks are the small "tubes" that emit light when you bend them
12:22:08midkyou bend them to fracture the capsule inside
12:22:09uskidid u see the movie Ocean Eleven ?
12:22:25uskimidk, do u know how much they usually sell them ?
12:22:40midkhmm, not sure. maybe 25 cents?
12:23:01uskiwow, damn cheap
12:23:09midkcould be 50 cents
12:23:13midki haven't used any lately :D
12:23:22midkit shouldn't be any more than that.
12:23:29uskiok :\
12:23:39midki should write a glowsticks manpage-type thing
12:23:44midkHow Glowsticks Work.
12:23:49HenrikBTime for lunch, I'll be back!
12:23:57midksee you, HenrikB
12:24:10uskimidk, why ? it would be interesting if you have some chemicals equations :)
12:24:14 Quit HenrikB ("Lämnar")
12:24:41midkuski: i'd just say something like "you bend them and then magic happens to light the thing up colory!" :]]
12:24:50midki considered it.
12:25:15midki watched their "what happens if you shoot your computer monitor" video
12:25:18midkit was quite good.
12:26:02uskihmmm i think i'll open my glowsticks and put the liquid in a transparent box
12:26:04midkgood, good video.
12:26:10midkonce i put it on my face
12:26:12midkit buuuuurned.
12:26:15midki mean...
12:33:31 Part lImbus
12:34:36uskiblah i opened a glowstick :D
12:34:51uskithe liquid in it is extremely bright
12:35:21uskinow, i'll "paint" a paper sheet with it (it's written on the box: non toxic liquid ;))
12:37:56midkit's fun
12:37:57midkI EAT IT
12:37:59midki mean..
12:38:24midkhmm you should make a big bowl full of glow liquid and dump it on a cat. then the cat glows evilly
12:38:37midkand you can run around screaming "MY CAT IS RADIOACTIVE!!! HELP!!!!"
12:40:03midkbring him into the pet place
12:40:05midkthe vet's
12:40:08uskii now have a glowing paper sheet :)
12:40:13midkand be like.. my cat is radioactive can you fix him
12:46:18 Join lImbus [0] (
12:53:37lImbushi mikd, err mitg, err imkd, err imdk, err whatever. not talinkg about the permutations of midk2k3 and midknight2003
13:15:19 Quit midk (Read error: 110 (Connection timed out))
13:35:40uskitime to break a glowstick ! :D
13:40:44[IDC]Dragonhi amiconn, u there?
13:48:41[IDC]DragonROMbox results are not tooo exiting
13:48:58[IDC]Dragonbut at least, it doesn't need more power
13:49:37amiconnNo, but this was expectable. First, the upper limit would be the 10% more free RAM, as you said in your mail.
13:51:05amiconnSecond, the harddisk isn't the only part drawing power. The results show that the gained runtime percentage gets smaller with longer runtimes
13:54:21[IDC]Dragonwhere the results of your 2 rounds somewhat close?
13:54:43[IDC]Dragon(for the same setup, I mean)
13:56:08amiconnYes, they were, but the uncertainty was in the same range as the gained runtime (~4 %)
13:56:33[IDC]Dragonhmm, this sortof invalidates it
13:57:28amiconnFurthermore, the 10:03 runtime I measured in the very first cycle was obviously a glitch (either caused by the elevated temperature or by the fact that the cells didn't undergo a full discharge/charge cycle in the box before)
13:59:01amiconnAs I wrote, the results are somewhat uncertain (and doing statistics with only 2 rounds is neither)
13:59:58[IDC]DragonI know, but thanks a lot still
14:00:38[IDC]Dragonthat's why you've been so quiet last days, your box was occupied by the tests ;-)
14:00:40amiconnDoing these test the proper way would either need a larger number of boxes, or much much time...
14:01:09[IDC]Dragonor some equipment to measure the energy consumption over time
14:01:39 Join Guest [0] (
14:01:46 Part Guest
14:02:50amiconn[IDC]Dragon: Yes, my box was occupied by the tests for some days. Plus, I was in a place where I did have only dialup available, which is no real fun ;-)
14:02:59[IDC]DragonLinusN is/was working on gcc 3.4.1
14:03:28 Join _AciD [0] (
14:03:42[IDC]DragonI even fail to install gcc 3.3.1 :-(
14:04:10[IDC]Dragonsomehow incompatible with my cygwin
14:04:29LinusNsince the battery runtime can change between charges, it is probably better to measure the actual current than the runtime
14:04:53LinusNthe 3.3 series is a bitch to compile
14:05:17LinusNit has a bug where it requires libc headers for no reason
14:05:26[IDC]DragonI didn't even copile it, was trying to use a binary
14:05:43amiconnLinusN: Yes, but this had to be done with a high time resolution, then integrated
14:05:52LinusNamiconn: yes
14:06:11[IDC]DragonI can check for suitable equipment here
14:06:19 Join AciD` [0] (
14:06:56amiconn[IDC]Dragon: For me, installing gcc 3.3.1 as described on the rockbox site worked without problems. Of course you have to update your cygwin installation first especially if your cygwin1.dll is rather old
14:07:17amiconnUpdating cygwin with the cygwin installer is rather easy imho
14:07:58[IDC]DragonI haven't found an installer in my setup
14:09:00amiconnHow did you install cygwin without the installer?
14:09:25amiconnBtw: the installer is _not_ placed within the cygwin directory, and doesn't need to be
14:09:32[IDC]Dragonit was long ago...
14:09:45[IDC]DragonI think it was the installer
14:10:01[IDC]Dragonit downloaded tons of stuff
14:10:25***Saving seen data "./dancer.seen"
14:10:27amiconnYes, and it keeps track of what it downloaded and installed
14:10:38amiconnThat way, updating an installation is simple
14:10:47[IDC]Dragonbut my cygwin group shows no installer now, and I didn't find an executable in the files which looks like so
14:11:45LinusNthe installer is the program you used to install cygwin in the first place
14:12:00LinusNit isn't moved anywhere, it stays where you put it
14:12:28[IDC]Dragonso, it doesn't get into the installation?
14:12:47LinusNyou can download it again if you wish
14:12:59[IDC]DragonI'll try that
14:13:29LinusNbut you should also try to find the directory with the installation info that setup.exe created when you installed it
14:13:41LinusNin the same dir as the instellr
14:13:47 Join midk [0] (
14:13:51LinusNbut i don't think it's necessary
14:14:11[IDC]Dragonfor the versioning, it may
14:15:08[IDC]Dragonis gcc 3.4.1 "officially" supported now?
14:15:27amiconnLinusN: I found a glitch in the wiki web statistics when updating the voice files last night: The wiki "credits" me two changes for each file upload
14:15:42LinusN[IDC]Dragon: in rockbox? yes
14:16:02LinusNamiconn: cool! now i know how to boost my stats! :-)
14:16:13amiconnLinusN: Does gcc 3.4.1 generate better code than 3.3.1 ?
14:16:30LinusNi don't know, i have never succeeded in compiling 3.3.1 :-)
14:16:37[IDC]DragonI think you two have to try a gcc "shootout"
14:17:34amiconnThat requires me to compile 3.4.1 myself, or is there a cygwin package available now?
14:17:44LinusNi don't know
14:18:04LinusNcheck with eric lassauge
14:18:22LinusNnope, only 3.3.1
14:18:26LinusNjust checked
14:19:34LinusNi have now completed my first step in the default handler project
14:20:24LinusNall (or most) "case SYS_USB_CONNECTED:" have been moved to the default case instead, with a call to default_event_handler()
14:21:05[IDC]DragonLinus, Jens, you could compile todays tarball, Linus with 3.4.1, Jens with 3.3.1, and compare
14:21:36 Quit _AciD (Connection timed out)
14:21:44LinusNi can compile the current cvs
14:22:19 Quit midk ("Leaving")
14:22:20amiconn[IDC]Dragon: Iirc, gcc 3.3.1 is also what you get when you install BlueChip's devkit
14:23:57LinusNwhich one should i build?
14:24:00LinusNfm recorder?
14:24:20 Join midk [0] (
14:25:02amiconnLinusN: As you like, just tell me
14:25:16LinusNtook 26 seconds to build
14:25:20LinusNfm recorder
14:25:39LinusNajbrec.ajz: 188108 bytes
14:29:38amiconnfm recorder, current cvs, 2 MB, normal, english: took 70 seconds to build, ajbrec.ajz: 185884 bytes
14:30:12amiconn(Of course the build times are in no way comparable)
14:31:14LinusNyour is a lot smaller
14:32:30LinusNi wonder what 3.3.1 does to make it that small
14:32:31[IDC]Dragonhmm, lesson is don't use 3.4 ?
14:33:02LinusNamiconn: are you sure that you built a clean current cvs?
14:33:17 Quit midk (Remote closed the connection)
14:33:26[IDC]Dragonwithout geek bitswap?
14:33:35LinusNa clean current cvs
14:33:44LinusNno changes at all
14:33:47amiconn[IDC]Dragon: That lesson is definitely valid on other platforms, as reported by numerous users. E.g. gcc 3.4.x has a number of "new" bugs at least on ppc
14:33:54 Join midk [0] (
14:34:45amiconnLinusN: Yes, that was plain current cvs. I have two local cvs copies, one of them being a plain unmodified one.
14:34:53 Quit midk (Remote closed the connection)
14:34:55 Join midk [0] (
14:39:48amiconnIf I compile with oldish gcc 3.0.4, fm recorder, same settings otherwise: build time 51 seconds, ajbrec.ajz: 190824 bytes
14:43:17LinusNi wonder what -Os would yield...
14:44:26 Quit [IDC]Dragon ("no fate but what we make (EOF)")
14:46:33lImbuswill my rockbox use rombox if I flash a todays build ?
14:47:08lImbusI plan to record this evening and the next days, but want to update (for sorting and soft-off).
14:47:28LinusNlImbus: rombox is not official
14:49:39lImbusbut in cvs, isn't it ?
15:01:07amiconnlImbus: For building rombox, you need to link with a special linker script after compiling, then create an .ucl from rockbox.bin with a special version of uclpack with understands the −−none option
15:02:01lImbuscould there be any interference that get's my ucl-file 5 KB small ?
15:02:05amiconnI updated the rombox build last night, and there is now a build for the v2 which _should_ work
15:02:09lImbusi won't try to flash with that
15:03:21amiconnThe .ucl should be a few bytes larger than the rockbox.bin, around 180 KB for v1, 184 KB for v2
15:04:33lImbusyeah, but it's not, so where could be the problem. havn't played/tampered with my local cvs for a while.
15:06:41amiconnA normal .ucl (not rombox) is usually around 110 KB
15:12:05lImbusmine is 5 bytes long
15:12:18lImbusit contains the word "fake" in simple ascii chars
15:12:54elinenbelImbus: you need to have the uclpack binary in your path.
15:13:02lImbuseven with a new build-dir and ../tools/configure
15:13:06lImbusI had all the time
15:13:21lImbusBASH: ucl: command not found
15:13:33lImbuswhat the hell went wrong ?
15:17:35LinusNsimple, you don't have ucl.exe
15:19:05lImbuswith windows explorer, I see about 25 files with touch, uclpack, patch, cvs and so on.
15:19:24lImbusw:\rockbox\usr\bin that is
15:19:53lImbusin cygwin I see a lot files more, but nothing like ucl in /usr/bin>
15:20:20lImbusand it worked before !
15:21:58 Quit midk ("Leaving")
15:22:04 Join midk [0] (
15:25:19 Quit midk (Remote closed the connection)
15:25:27 Join midk [0] (
15:25:44 Quit midk (Read error: 104 (Connection reset by peer))
15:26:43LinusNi guess uclpack.exe is what you want
15:27:15LinusNthe special −−none version, that is
15:27:20 Join midk [0] (
15:27:31lImbusI think so. but I can't get access to it. can't call it with ./uclpack if I sit in the directory where windows says it's in.
15:28:33lImbusI just verified it's installed. cygwin-setups says I got version 1.01 or something like that.
15:29:00LinusNso you cd to the directory and type ./uclpack.exe and nothing happens?
15:29:45midkam i the only one that finds this line *really* funny?
15:29:47midk"Make sure you have sh-elf-gcc and siblings in the PATH."
15:30:04LinusNyou mean "siblings"?
15:30:25midkit's just funny.
15:30:26lImbusLinusN: yes. it says BASH: uclpack.exe not found
15:30:45LinusNand if you type "ls" it is there?
15:32:04lImbusno. it's not there, but windows says it is, along with other files
15:32:38lImbushow to see what's inside %path% in cygwin ?
15:35:00 Quit midk (Remote closed the connection)
15:35:05 Join midk [0] (
15:35:29lImbusmidk: are you ok ?
15:35:39midka bit tired.
15:35:47midkbut yes.
15:38:20 Nick midk is now known as midk|sleep (
15:38:31*midk|sleep is away: sleep.
15:39:08lImbusLinusN: ls in /usr/bin shows the same content as ls in /bin
15:39:39lImbusbut /usr/bin is really existing on my windows-fs and there is no indication for a symlink
15:40:01lImbusshould I copy the contect of /usr/bin to /bin ? *wonder*
15:42:25LinusNlImbus: as long as it's in your PATH, everything should be ok
15:42:54LinusNlImbus: don't rely on what Windows says
15:43:19lImbushey, I am running cygwin in windows, not the other way round
15:43:25lImbuscopyied uclpack to /bin and to /bin/ucl.exe to be sure.
15:43:25LinusNput the executable somewhere in your cygwin path
15:43:54lImbusnow I can call ucl and uclpack from the ~/build-dir/ but make does not pack it yet.
15:44:03lImbusit's still 5 bytes long
15:44:12LinusNis it in the PATH?
15:44:22lImbusImbushow to see what's inside %path% in cygwin ?
15:44:35LinusNecho $PATH
15:45:07lImbuspath seems to be very empty. but I can call ucl from the build-dir ?!?
15:45:21LinusNremove rockbox.ucl and make again
15:45:49lImbusahh. well.
15:45:55lImbusnice. now I got one
15:46:10lImbusnobody knows how that could work if my $path is empty
15:46:36LinusNdid you type "echo $path" or "echo $PATH"?
15:46:49lImbusargl. case sensitive
15:47:03LinusNwhy do you think i wrote echo $PATH?
15:47:33 Join HenrikB [0] (
15:47:35LinusNi don't write anything else in upper case when i'm in irc
15:47:41lImbushehe. sorry, not used to that thingy.
15:48:02HenrikBHi again
15:48:14lImbus$PATH contains pretty much everything needed. as well as both /usr/bin and /bin
15:48:52LinusNhi HenrikB
15:49:02 Join quelsaruk [0] (
15:49:11LinusNquelsaruk: welcome!
15:49:13quelsarukhi good morning
15:49:18quelsarukor good night here in japan
15:49:35LinusNwhere in japan?
15:49:42quelsarukjust wanted to say hello
15:49:54quelsarukright now in Kyushu
15:50:20quelsaruktomorrow im going to Kyoto
15:50:34quelsaruksee you linus... and everyone
15:54:02 Part quelsaruk
15:55:10HenrikBLinusN, I've got a small codesaver, I've replaced the id3_browse() with the id3edit rock patch.
15:56:11HenrikBbut the patch must have my onplay() patch for wps applied first.
16:00:05LinusNHenrikB: yeah, lifting out the id3 browser might be a good idea
16:09:16HenrikBLinusN, it saves ~750 bytes with gcc 3.3.1
16:10:26***Saving seen data "./dancer.seen"
16:20:20 Join [IDC]Dragon [0] (
16:20:43[IDC]Dragonah, CGIIRC works again
16:20:52[IDC]DragonI got kicked out
16:23:24 Quit AciD` (Read error: 104 (Connection reset by peer))
16:40:19 Join elinenbe_ [0] (
16:45:59 Join dstar5 [0] (
16:50:27 Join Ericgoh [0] (
16:58:22 Quit elinenbe (Read error: 110 (Connection timed out))
16:58:23 Nick elinenbe_ is now known as elinenbe (
16:58:28dstar5LinusN: compiling for target is giving me an error at linking
16:59:35LinusNdstar5: and the error is...?
17:00:31dstar5i amg getting to it
17:00:34dstar5it is about ld
17:02:40dstar5collect2: cannot find `ld'
17:02:40DBUGEnqueued KICK dstar5
17:02:40dstar5make[1]: *** [/home/Lee/rockbox-devel/build/rockbox.elf] Error 1
17:02:40dstar5make[1]: Leaving directory `/home/Lee/rockbox-devel/apps'
17:02:40***Alert Mode level 1
17:02:40dstar5make: *** [apps] Error 2
17:02:41***Alert Mode level 2
17:02:41dstar5Lee@home ~/rockbox-devel/build
17:02:47dstar5$ ld
17:02:49dstar5ld: no input files
17:03:44 Join mecraw___ [0] (~lmarlow@
17:04:52dstar5LinusN: any ideas?
17:08:27LinusNdstar5: are you using bc:s devkit or what?
17:08:49dstar5standard cygwin install
17:09:16dstar5with gcc 3.4 and binutils 1.15 for sh-elf built myself
17:09:37LinusNi find it odd that it tries to use ld and not sh-elf-ld
17:10:19LinusNdid you use the same prefix for both gcc and binutils?
17:12:42***Alert Mode OFF
17:13:17LinusNtype this:
17:13:23LinusNsh-elf-gcc −−version
17:13:30LinusNsh-elf-ld −−version
17:14:19dstar5umm for some reason sh-elf-ld says command not found
17:14:33dstar5sh-elf-gcc (GCC) 3.4.0
17:15:52dstar5thats strange...
17:16:27dstar5i have all the other binutils tools i think
17:16:32dstar5ar, as...
17:18:33LinusNif sh-elf-ld is missing, your installation of binutils must have failed
17:18:58dstar5GNU ar 2.15...
17:20:07dstar5ill just rebuild
17:24:10dstar5i get patch for gcc 3.4.1 also...
17:28:07[IDC]Dragoncu, folks
17:28:14 Quit [IDC]Dragon ("no fate but what we make")
17:32:42 Quit Ericgoh (Read error: 60 (Operation timed out))
18:08:55 Part LinusN
18:10:30***Saving seen data "./dancer.seen"
18:13:24 Part HenrikB ("Lämnar")
18:24:08 Join _AciD [0] (~acid@
19:09:09 Quit _AciD (Read error: 110 (Connection timed out))
19:09:09 Quit NibbIer (Read error: 54 (Connection reset by peer))
19:28:56 Quit dstar5 ("Leaving")
19:53:27 Join scott666_ [0] (
20:10:31***Saving seen data "./dancer.seen"
20:25:27 Join _AciD [0] (
20:32:37 Join NibbIer [0] (
20:43:03 Join Tunsnask [0] (
20:47:24 Quit Tunsnask (Client Quit)
20:57:49 Join zeekoe [0] (
20:59:56 Quit zeekoe ("aAA44AArrgGGHhh... my computer just did something _really_ REALLY wrong... cu later")
21:00:27 Join zeekoe [0] (
21:00:31 Quit _AciD (Connection timed out)
21:04:05 Join _AciD [0] (
21:14:40 Join gromit`` [0] (
21:15:11 Join Smooth [0] (
21:21:34SmoothWhy the "Shutdown" option when holding the OFF button does the same job?
21:22:53 Join LinusN [200] (
21:23:23LinusNSmooth: u there?
21:24:00Smoothyea linus
21:24:02LinusNthe shutdown option saves the settings and spins down the hard drive before turning off
21:24:08Smoothoh ok
21:24:15Smoothjust was wondering
21:24:53LinusNamiconn: u there?
21:25:07Smoothman that gmini220 wipes the floor with the ipod
21:25:13 Join Tunsnask [0] (
21:25:14amiconnLinusN: I'm here
21:25:39LinusNi'm changing the SYS_USB_CONNECTED behaviour of the plugins
21:25:54LinusNi can't decide if the plugins should exit or not
21:26:07Smoothlinus whats this new "First step in revamping the USB event handling, paving the way for the upcoming SYS_POWER_OFF event"?
21:26:31LinusNi think they could continue running, but have to return PLUGIN_USB_CONNECTED when they eventually return
21:27:09LinusNSmooth: it's mainly for the FM and V2 recorders, that can trap the Hold-OFF and save the settings when the user holds OFF
21:27:46Smootharchos rules
21:28:10amiconnLinusN: I'd say this depends on the plugin, if it makes sense to keep it running when usb is connected
21:28:34amiconnWould the plugin be "interrupted", displaying the usb screen in that case?
21:28:43LinusNamiconn: no, it should show the usb screen as usual, but they may continue running afterwards
21:29:12LinusNinstead if immediately returning PLUGIN_USB_CONNECTED
21:29:24amiconnThis may lead to at least 2 problems:
21:29:50LinusNthe plugin calls the default handler, like the normal rockbox code, but it may continue running after the USB screen si finished
21:29:52amiconn(1) The plugin has to handle the usb event, because it has to redraw the screen afterwards
21:30:51amiconn(2) Same problem arises for all plugins using grayscale (not that hard to handle there, as long as _only_ grayscale is displayed)
21:31:19LinusNall plugins handle the USB event today
21:31:29Smootharchos must be the only company that supports simple drag-n-drop to its HD rather than using DRM software
21:31:33LinusNthey just handle it differently, most of them just return
21:31:42amiconnYes, but they usually don't redraw the screen, as they exit afterwards
21:31:48LinusNSmooth: there are a few others as well
21:31:57Smoothcan't think of any others
21:32:02Smoothapart from iriver
21:32:32LinusNamiconn: the calendar continues
21:32:56amiconnOops, wasn't aware of that one (I don't use it)
21:32:56LinusNi was just thinking of what should be the requirement
21:33:18LinusNi don't think we should force the plugins to exit
21:33:48amiconnI think it should be handled as it is now, let the plugin decide whether to exit or not
21:33:59LinusNthen we agree
21:34:11CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
21:34:11*LinusN goes back to editing all plugins
21:34:52lImbusbye guys, I'm off for home
21:34:57amiconnThere will still be a problem with the grayscale plugins, since in case an usb event is detected, they have to execute code _before_ the usb screen is displayed
21:34:58LinusNlImbus: bye
21:35:15LinusNthen they can easily trap the USB event themselves
21:35:20 Part lImbus
21:35:34LinusNand then call default_event_handler(SYS_USB_CONNECTED);
21:35:50LinusNthe recording code does that, for example
21:37:14amiconnPerhaps I could integrate a function to simplify that for the developer into the grayscale framework
21:38:06amiconnyup, something like that
21:39:25amiconnThere could also be another plugin lib function that does screen backup/restore into a memory area supplied by the plugin
21:39:37amiconn...for b&w graphics
21:39:52 Nick midk|sleep is now known as midk (
21:40:16amiconnmidk: evening ;)
21:40:23LinusNamiconn: to restore the screen?
21:40:29amiconnLinusN: yes
21:40:43LinusNi think we should increase the plugin memory space
21:40:56zeekoeevening :)
21:41:25amiconnLinusN: I would do this only if it is absolutely necessary for a plugin.
21:41:27zeekoemidk: it's been midday at you place
21:41:44midkzeekoe, true, but i was up til 6:50am
21:41:57LinusNpoor bluechip is struggling to fit his stuff in the 32K
21:42:21amiconnLinusN: I it a real challenge (in a positive sense) to fit all functions into 32K
21:43:07amiconnJörg managed to fit a complete jpeg decoder _and_ the grayscale framework into 32K
21:43:21amiconnThere is even some room left for extension
21:43:38zeekoemidk: sick...
21:43:42LinusNmidk: your dream may come true :-)
21:43:46zeekoemidk: don't you have to work...
21:44:29amiconnLinusN: Yes, but if I think about it, this would only be a few lines, probably not worth putting into functions.
21:44:41LinusNis the frame buffer exported?
21:44:50amiconnAfter all, these are simple memcpy() calls
21:45:09amiconn896 bytes
21:45:22midkLinusN, i dreamed i was saving some lady from lots of i,robots
21:45:27LinusNactually, usb_screen() could do it
21:45:30midkit probably will.
21:45:51 Join Strath [0] (
21:46:12amiconnLinusN: Yes, lcd_framebuffer is exported
21:46:20LinusNit may not work that well on the player, though
21:47:18Smootheveing then
21:47:20amiconnLinusN: Ooops, I forgot the player.
21:47:31 Part Strath
21:47:43LinusNwell, i think the player has a "frame buffer" too now
21:47:54LinusNused by the rocklatin framework
21:48:00Smoothdo players still exist?
21:48:06LinusNi have two
21:48:19Smooththought they would be obselete by now
21:48:22 Quit _AciD ("'%20radi)
21:48:40LinusNthere are quite a few of them out there
21:49:03zeekoeLinusN talked me out of buying a player :-)
21:49:05Smooth2 lines of character cells? thats just GAY!
21:49:08zeekoeThanks, Linus!!! :-)
21:49:45Smoothmidk talks people out of buying ipods
21:50:38zeekoeSmooth: and of buying what things do you talk people out?
21:50:55Smoothtalk people out of buying anything made by Creative Labs
21:52:05zeekoei thought about buying a creative dap jukebox, about 2 years ago
21:52:15zeekoeglad i didnt buy it, it's huge
21:52:18Smoothused to have one myself before the archos
21:52:24Smoothand the battery life is piss-poor
21:52:44zeekoeis it?
21:52:53Smoothonly 4 hours
21:53:08zeekoei really like the long playing time of the archos
21:53:19Smoothand the software required to send music to the DAP is bad, bad, bad
21:53:26zeekoei had a lennox power one, it only had 2-3 hours on 2 batts...
21:53:41Smoothyou have to log in all your MP3 files and send them only in small spoonfuls
21:53:49zeekoeFAT support r00lz
21:53:49Smoothif you try and send the whole lot in one go, it crashes
21:54:02amiconnActually, the Archos was the first hd mp3 player that fulfilled my primary requirement: putting music on it must not require special software
21:54:21LinusNsame here
21:54:30amiconn(especially as I did not own an x86 pc at that time)
21:54:33LinusNthat's the main reason i chose archos
21:54:34Smoothim sure that pleased the record-companies
21:54:39zeekoeamiconn: what did you own?
21:54:50zeekoeamiga r00lz
21:55:02LinusNi have one A500 and one A4000
21:55:03zeekoeamiga with usb?
21:55:12amiconnMy amiga is still up and running 24/7
21:55:20amiconnYes, with usb
21:55:22LinusNnot mine :-(
21:55:35zeekoeusb isnt that long available on amiga, is it
21:55:53zeekoe2 years or so, now i think of it
21:56:06*zeekoe translated some stuff for OS4 to Dutch
21:56:24amiconnI bought a Highway card (4 ports usb 1.1 for the Amiga Zorro bus) a little more than 2 years ago, the same time when I bough my Archos
21:56:25LinusNamiconn: i'm thinking of adding a lib function for saving the lcd frame buffer
21:56:48Smoothhardly change their size when you further compress them using GZIP
21:57:02LinusNseems a little clearer in the code
21:57:34LinusNcomplete with a buffer for storage
21:57:49LinusNwon't be linked if not used
21:58:37amiconnLinusN: If you want to make this work for the player as well, you would have to export whatever lcd buffer is used there
21:58:52LinusNand the plugin wouldn't have to care about the player issues
21:59:03Smoothwho changed the progress bar from the slider to the Ipod-style ba?
21:59:27LinusNi didn't know ipod had one
21:59:42LinusNthen i wouldn't have done it :-)
21:59:44Smooththe progbar on the ipod slowly fills rather having a slide
22:00:08LinusNi think the current one looks better
22:00:11Smoothbasically the new progbar works the same as the ipods
22:00:31Smoothhey i have a new idea for a progbar
22:00:45Smoothwhy not have the screen invert from left to right like the invert on the clock plugin
22:01:06zeekoethe clock plugin has cool new features?
22:01:08LinusNlooks awful imho
22:01:09*zeekoe checks out
22:01:23Smoothu know what i mean linus?
22:01:31Smooththe left-to-right invert on the clock plugin?
22:01:55LinusNi know
22:01:59Smoothim sure mikeholden thought it was a good idea when he revamped the plugin
22:02:15Smoothits his plugin
22:02:27LinusNid midk and mikeholden the same person?
22:02:43Smoothwell both start with mi and have a k
22:02:58midktrack's in his little .. world.. again. :)
22:03:35amiconnImho the clock plugin shows what happens if you don't care for code size (sorry midk). It is the second largest plugin code-wise, much larger than what I would expect concerning its functionality
22:03:52Smoothwe could have a world-time plugin
22:03:57midkamiconn: wait till i add atomic synchronization :)
22:03:58Smoothlike on some digital watches
22:04:07midkthat's when i'll hit that barrier.
22:04:19zeekoeyou need a receiver for that... or a computer
22:04:26Smoothmoody mikeholden
22:04:35midkhonestly, i'm going to finish fuzzy mode then optimize what i have
22:04:45midki can take out some old code. the logos can also go
22:04:53zeekoesome guys at my school made a thing to _send_ dcf-77 (atomic clock) signals... they were able to change the time :-P
22:04:56midkduplicate code could be moved to a different function
22:05:41Smoothfuzzy logic?
22:05:49midkno, fuzzy.
22:09:04amiconnLinusN: Could you please send me an example of rockbox, compiled with gcc 3.4.1, along with the corresponding .map file?
22:09:30amiconnI'm curious _why_ gcc 3.4.1 produces larger binaries than gcc 3.3.1
22:09:50amiconnI'd prefer something that I know well, e.g. the grayscale.rock
22:10:34***Saving seen data "./dancer.seen"
22:11:48amiconnI already have a suspicion what may cause this...
22:12:29 Quit midk ("Leaving")
22:14:09 Join midk [0] (
22:14:11 Join AciD` [0] (
22:15:26zeekoespeaking of grayscale...
22:15:36zeekoei tried converting some movie about 2 weeks ago
22:15:41zeekoeit's a 25 fps movie
22:15:52zeekoebut it ran way too fast (2 times or so)
22:16:16zeekoeso i tried setting the input frame rate to 25, and the output frame rate to 25
22:16:38zeekoegot a really weird movie, not really what one could call greyscale.
22:16:54zeekoetried again, but then using 25 fps in, 67 fps out
22:16:56midkdon't change the output frame rate. or don't put it that low..
22:17:00zeekoesame, weird result
22:17:25zeekoeit _should_ be possible to change it, isnt it?
22:17:29amiconnzeekoe: Using the command line tool suite or the DirectShow filter?
22:17:42zeekoedirectshow filter?
22:17:55amiconnIf it is the latter, I'm afraid I can't help, since I never used that
22:17:56zeekoejust the command line thingy
22:18:07zeekoewhat's the ds filter?
22:19:09amiconnThe DirectShow filter is a windows thingy, usable with all video processing tools that use Microsoft's DirectShow architecture, that is able to produce .rvf files, with sound
22:19:19amiconnIf the so
22:19:42zeekoewhoops... empty batteries...
22:20:00amiconn..urce sound format is not mp2 or mp3, it does additionally require the lame DirectShow encoder
22:20:19zeekoei think i have that
22:26:01zeekoehey, i see the directshow thing on the site
22:26:15zeekoei'll try that first
22:32:28zeekoehmm... it's 1x real time :-/
22:38:31Tunsnaskhow do i set my recorder on HOLD?
22:42:08midkat the wps, F1+down
22:42:51zeekoecool! didnt know that :)
22:45:07 Quit Tunsnask ()
22:45:25zeekoehmm... i think i did something wrong last time, now the video is fine
22:45:31zeekoestill way out of sync though
22:45:53zeekoedoes it matter the audio is 22khz, stereo?
22:46:59amiconnIt shouldn't. I converted a number of movies, even very long ones, and always got the audio in sync.
22:47:18midkreboot brb
22:47:20amiconnBut then I use the command line tool suite to do the conversion...
22:47:20 Quit midk ("Leaving")
22:48:42zeekoehmm weird
22:48:51zeekoeyeah, i'm using the command line stuff too, again
22:49:03zeekoethe directshow thingy did really weird
22:49:23zeekoeit continued to play even a minute after the movie ended
22:49:24amiconnOf course, using the command line tools requires quite a number of steps to convert a movie though
22:49:39zeekoefirst convert to 112x64
22:49:47zeekoethen avi2yuv
22:49:50zeekoethen halftone
22:49:53amiconnThis is what I do when converting a dvd:
22:49:54zeekoethen rvf_mux
22:50:23amiconn(1) Rip the dvd and demux into an .m2v and an .ac3 file
22:51:16 Join midk [0] (
22:51:18amiconn(2) Convert the .m2v into an 112x64 avi (cropping & scaling to get the aspect right and get rid of the black bars)
22:51:34amiconnI'm using Avisynth and VirtualDub for that
22:52:13amiconnIf the resulting Avi would be larger than 2 GB (I use Intel YUV format), it has to be splitted into 2 GB segments
22:52:55amiconn(3) Convert the avi to yuv with avi2yuv. Size will be almost the same after that if you use Intel yuv in the avi
22:53:15amiconnThis has to be done for every segment
22:53:30zeekoe(sorry for interrupting...) wheeeeeeee! totally in sync now! i converted 22khz, 8bit to 44k1, 16bit
22:54:14amiconnOk, good. Should I stop here?
22:54:23zeekoeyou may continue :)
22:54:35zeekoeyou forgot the audio part :P
22:54:46amiconn(4) Use halftone to convert to video-only .rvf format
22:55:20 Quit Smooth (Read error: 104 (Connection reset by peer))
22:55:38amiconn(5) if you had to segment the video (both avi2yuv and halftone can't handle files >2 GB), concatenate the .rvf segments together
22:55:54amiconnThis is possible because video-only .rvfs have no header
22:56:24amiconn(5) Use headac3he to decode the .ac3 and downmix it to a stereo wav
22:56:37amiconnErr, this was (6)
22:57:14zeekoeok, i already expected headac3he here :)
22:57:29amiconn(7) Encode this .wav with lame into an .mp3 (I use −−preset medium for normal movies, −−preset standard for music videos/movies)
22:58:04zeekoeso you _really_ watch movies on the recorder?
22:58:09amiconnFortunately lame is able to handle wavs > 2 GB, it only shows weird encoding times/ frame counts in that case
22:58:31amiconn(8) rvf_mux video and audio together
22:58:55LinusNamiconn: is that so on linux as well?
22:59:03LinusNthe 2Gb flaw?
22:59:19LinusNand on NTFS?
22:59:21amiconnzeekoe: Yes, although I don't do that often.
22:59:23zeekoeamiconn's guide to rvf in less than 10 steps :)
22:59:48zeekoei wonder, how do you get 2 GB+ files?
23:00:02zeekoemy files are only 4 MByte per minute
23:00:27amiconnConverting e.g. LOTR. LOTR3 does break the 2 GB barrier even for the audio part
23:01:01amiconnLinusN: For lame on Linux, I don't know.
23:01:06zeekoebut not for the video i guess
23:01:51amiconnzeekoe: Haha. Of course it does, as I said _even_. The 2 GB barrier for the video part is it much earlier
23:01:58amiconn*is hit
23:02:33zeekoeof course
23:02:53zeekoethe input files are color 112x64, so the output files are much smaller
23:03:10zeekoehow many greys are in the output file? 67?
23:03:32 Join Tunsnask [0] (
23:03:40amiconnLinusN: The 2 GB barrier in avi2yuv has nothing to do with the file system, but with the file handling code in it. Unfortunately we are not able to change it, since avi2yuv is not open source, and hence not available for Linux
23:05:07amiconnzeekoe: There are no "real" grays in the output file, but only black and white pixels. .rvf uses a higher frame rate (standard is 67 fps) to simulate gray
23:05:27zeekoeso theoretically it's 67 greys
23:05:34 Quit Tunsnask (Client Quit)
23:06:16amiconnNo, the "theoretical" number of grays is unlimited for .rvfs, since it does not use a repeated sequence, but a continuous stream of pixels
23:06:48zeekoehmm... cool :)
23:07:16zeekoeDirectStreamDigital (R) (TM) :-P
23:07:30scott666_whatever happened to the gui?
23:07:35 Nick scott666_ is now known as scott666 (
23:07:51amiconnscott666: what gui?
23:08:19scott666the one jörg (i think) was working on
23:08:37zeekoeit's in the tools dir, just downloaded it
23:09:03 Join Tunsnask [0] (
23:09:05scott666LinusN: any idea when 2.3 is going to be released?
23:10:07LinusNscott666: some time in september i hope
23:10:59scott666not until september? is there something big in the works?
23:11:20LinusNthings are happening, and the core team are on vacation
23:12:16scott666so 2.3 will probably be around when rombox is finalized?
23:15:39LinusNhas nothing to do with rombox
23:16:21amiconnLinusN: Got your mail. Actually, grayscale.rock looks like a bad example, since for that the gcc 3.4.1 result is smaller than what gcc 3.3.1 produces (by 36 bytes)
23:16:46zeekoehow nice of microsoft...
23:17:04gromit``in case you dunno how to do :)
23:17:19zeekoe"you're sick of a system that's rock-stable? want your good ol' viruses back? here's the guide"
23:17:50zeekoenot that windows xp is that unstable
23:18:04LinusNamiconn: can you do sh-elf-objdump -h rockbox.elf?
23:18:51 Join [IDC]Dragon [0] (
23:18:55LinusNamiconn: never mind, it won't give us anything
23:19:11amiconnLinusN: yes. Where should I put the result?
23:20:39Tunsnaskchould any one send my a guide to patch my recorder?
23:20:55LinusNTunsnask: ???????
23:21:02zeekoewhat patch
23:21:43LinusNamiconn: email me the results from the objdump
23:22:01Tunsnaskhow i sply a patch
23:22:39Tunsnaskbut that site dont really make sense to me...
23:22:54Tunsnaskanyway I'll try
23:23:31LinusNif it doesn't make sense for you, well, then patching isn't for you i guess
23:24:02Tunsnaskit gotta be
23:26:02LinusNTunsnask: i don't want to offend you or be rude, but the rockbox patches are for developers
23:26:24zeekoeTunsnask: what patch do you actually want to apply?
23:26:44Tunsnaski dont really know
23:26:53zeekoewhy not
23:27:03zeekoewhat do you want from your rockbox
23:27:33Tunsnaski dont know
23:27:50zeekoeso, where did you get the idea of applying a patch?
23:28:12Tunsnasksome friends
23:28:18zeekoeor is it just so you can tell your friends at school "hey,look, i patched my...
23:28:22zeekoeheh, okay
23:28:31LinusNamiconn: the text section is much smaller in your dump
23:29:11Tunsnaskmy search dont work...
23:29:33amiconnLinusN: I wonder _why_ that is. would be an interesting thing to read & compare
23:29:57Tunsnaskall my files in my .rockbox folder dont work...
23:30:11LinusN"don't work"?
23:30:20zeekoecopy new ones :)
23:31:05LinusNTunsnask: when you install rockbox, you must unzip the entire zip file onto your jukebox
23:31:24zeekoeor use the windows installer :)
23:31:37amiconnFor grayscale.rock, the text section is the only one that is larger with gcc 3.3.1, all others are identical (apart from .comment, but that doesn't count)
23:32:24LinusNamiconn: you want my
23:32:49LinusNthen pick a module that differs a lot in size
23:32:57LinusNand i'll send it to you
23:32:58amiconnYes, that may be interesting (along with rockbox.bin)
23:33:18Tunsnaskstrange it work now.. i had tried it before
23:33:32LinusNhey, i'll send you all of it (you have DLS, right?)
23:33:51LinusNhang on
23:35:34Tunsnaskmy search still dont work :-(
23:35:53Tunsnaskplugin returned error
23:36:11zeekoeyou need to open m3u's
23:36:15zeekoeyou do that, do you?
23:36:21 Quit [IDC]Dragon ()
23:36:32zeekoego to an m3u, press on+play
23:36:40zeekoeselect open with -> search
23:36:45zeekoeand off you go
23:37:11Tunsnaskerh... where do I find a m3u?
23:37:53zeekoemake one
23:38:05amiconnLinusN: Perhaps the search plugin could tell that you have to use open with on an .m3u to use it if there is no parameter
23:38:40Tunsnaskhow do i make a m3u?
23:38:41zeekoeTunsnask: something like menu -> playlists -> create playlist or so
23:38:45amiconnCurrently, it is possible to start it and even enter a search string. If you then hit "Ok" -> Plugin returned error
23:38:48zeekoemy recorder is off now
23:38:55zeekoei mean, plugged into usb
23:39:47amiconnAlthough it isn't possible to accidentally start the plugin directly if you have a correct installation unless you set "show files" to "all"
23:41:54amiconnThanks, got it
23:42:17Tunsnaskerhm... it still dont work :-/
23:42:30zeekoeTunsnask: what do you do, exactly?
23:42:49Tunsnaskok wait i got it
23:42:59zeekoenice :)
23:43:08zeekoeLinusN: is it already possible for us earthlings to build rombox?
23:43:17Tunsnaskthere i go :-)
23:44:23LinusNzeekoe: ask amiconn
23:44:50zeekoeok, right, you were the rombox guy
23:45:02zeekoeamiconn: is it already possible for us earthlings to build rombox?
23:45:31zeekoes/you/amiconn: you
23:46:32amiconnIf you are prepared to fiddle a bit with scripts, and you are either using cygwin or are able to build the required special version of uclpack, yes
23:47:47zeekoei saw that somewhere on the rockbox site :)
23:47:50*zeekoe looking
23:48:03amiconnDon't look, it isn't found there
23:48:39zeekoeso where is it found?
23:48:52Tunsnaskok is it normal that a search takes really long time??
23:48:53amiconn[IDC]Dragon, who originally "invented" rombox, prepared a special version of uclpack which is able to generate an uncompressed .ucl file
23:49:09LinusNTunsnask: no
23:49:46amiconnHe send it directly to me. Although uclpack is open source, it is not found on the rockbox page, as it is not a simple single-source-file tool
23:50:04amiconnzeekoe: I could send it to you if you want
23:50:23Tunsnaskthere just stands searching.... searching.... searching... and so for 10 minutes...
23:50:39zeekoeamiconn: well.. i dont think i really need it anyway
23:51:15zeekoebtw, how's the version management in rockbox? when will a plugin be incompatible?
23:54:11LinusNTunsnask: what were you searching for?
23:54:53LinusNzeekoe: it is incompatible if the plugin API has changed
23:55:21Tunsnaska song
23:56:11amiconnLinusN: This is funny: comparing files show that gcc 3.4.1 is better in condensing data (both .rodata and .bss are smaller, the latter only by 4 bytes), but worse in condensing code (both .text and .icode are larger) compared to gcc 3.3.1
23:56:26LinusNTunsnask: come on, if you want our help, try to be helpful back!!!!!!
23:58:04LinusNi know you were searching for a song, that's the whole purpose of the search plugin
23:58:13amiconnThere is a bug in the linker of gcc 3.4.1: The size reported for *fill* elements looks rather funny
23:58:24Tunsnaskthen why do you ask...?
23:58:33LinusNyou mean binutils 2.15
23:58:42amiconnYes of course
23:58:49zeekoeTunsnask: what _exactly_ are you searching for
23:58:52LinusNTunsnask: you just told me that it searches forever

Previous day | Next day