| *** aquilo has quit IRC | 00:02 | |
| Lullen | If I make a program that makes you add stuff, for example a exercise. Should it be in a dialog or in a window to be "maemo-style" | 00:06 |
|---|---|---|
| SpeedEvil | Well - if you take the example of adding a new alarm, it should pop up into a new deeply annoying window with stupid scrolly bits. | 00:11 |
| Lullen | okey then I'll keep it like that | 00:28 |
| Lullen | Hmm I can not even install my own app, what a failure | 00:29 |
| *** SpeedEvil has quit IRC | 00:29 | |
| *** SpeedEvil has joined #maemo-devel | 00:29 | |
| Lullen | how does the autobuilder works? When going to my project page to download the .deb file I get a 110kb file, when creating the deb file on my computer it is ~7MB. It is the same code, pro file, debian folder so what can make this problem? | 00:37 |
| lcuk | Lullen, the command you are giving to dpkg_buildpackage will be different | 01:00 |
| lcuk | and the most likely scenario is you are leaving your .git folder inside the package | 01:02 |
| lcuk | ;) | 01:02 |
| Lullen | wich mean? :D | 01:03 |
| lcuk | well, you tell me how you build a package | 01:03 |
| lcuk | on your machine, what command do you give | 01:03 |
| Lullen | dpkg_buildpackage | 01:03 |
| Lullen | to upload i use dpkg_buildpackage -sa -S | 01:04 |
| lcuk | have you tried extracting the package to see what extra is inside it.. | 01:06 |
| Lullen | no I did not even know that I could do that, but why I am asking is when I download the .deb file from the project page it won't install as it is incombitable | 01:08 |
| *** DocScrutinizer is now known as joerg42 | 01:11 | |
| azakai | Hello all, is there a good way to measure how much power is being used? (to check how efficient a program is, how much it will affect battery life) I can't find something clear in the wiki or forums. And things like Kill-A-Watt don't have milliwatt resolution which would be needed here | 01:13 |
| SpeedEvil | Several issues. | 01:15 |
| *** joerg42 is now known as DocScrutinizer51 | 01:15 | |
| SpeedEvil | http://wiki.maemo.org/N900_Hardware_Power_Consumption is an overview of consumption. | 01:16 |
| SpeedEvil | I need to write a software power consuption page. | 01:16 |
| SpeedEvil | I'm in hte middle of writing a power nanalyser - but it's very much proto code as yet. | 01:16 |
| SpeedEvil | Firstly. | 01:16 |
| SpeedEvil | When idle, be truly idle. | 01:16 |
| SpeedEvil | strace is plenty for this - your app should not be making any calls when idle, unless it truly has to. | 01:17 |
| azakai | SpeedEvil: well, I am now in the process of reducing wakeups in our code (mobile firefox), and I keep getting asked "how much more battery life" will your work give. So I'd like actual battery figures, and not just # of wakeups or something else completely in software | 01:17 |
| SpeedEvil | Secondly, powertop gives an idea of wakeups - your app should not increase from a baseline level wakeups, when it is running but idle. | 01:17 |
| SpeedEvil | ah - fair enough | 01:18 |
| SpeedEvil | At this stage, the easiest way to do this is to install the 'power kernel' from extras-devel - and then modprobe the bq27x00_battery (or whatever it's called) module. | 01:18 |
| SpeedEvil | Then look at /sys/wherever/current_now | 01:19 |
| azakai | is that reliable? hal-device bme results seem to show only numbers at resolution 10% or so | 01:19 |
| SpeedEvil | yes. | 01:19 |
| SpeedEvil | hangon | 01:19 |
| azakai | ok | 01:19 |
| SpeedEvil | You sound like not a noob. This is some code stskeeps wrote to probe the battery meter. | 01:20 |
| DocScrutinizer51 | anyway not directly correlatable to particular process | 01:20 |
| azakai | ah, that sounds useful | 01:20 |
| SpeedEvil | This is some very prototype code - bq27200.sh probes the battery meter directly - hi.sh reads it out in a more high resolution fashion - and i2cget is used by it to read the meter. | 01:23 |
| SpeedEvil | www.mauve.plus.com/bat.tar.gz | 01:23 |
| SpeedEvil | http://wiki.maemo.org/N900_Hardware_Charge_Meter has links to the datasheet for the chip | 01:24 |
| SpeedEvil | briefly | 01:24 |
| azakai | SpeedEvil: thanks! checking it out | 01:24 |
| SpeedEvil | ./bq27220.sh -r | 01:24 |
| SpeedEvil | LOOPMODE=5 | 01:24 |
| SpeedEvil | mv RSOC CSOC mA NAC CACD CACT TTF TTE TEMP EDV1 | 01:24 |
| SpeedEvil | 23:24 4115 100 100 180 1221 1221 1221 0 65535 303 0 | 01:24 |
| azakai | hmm, the page says I need a replacement BME? | 01:25 |
| SpeedEvil | it's being run in loopmode - it polls every 5 secs. Eventually, that's not relevant to this | 01:25 |
| DocScrutinizer51 | only usefull values are mAh, current and voltage. forget the rest | 01:25 |
| SpeedEvil | mv = current voltage. RSOC/CSOC = state of charge in % - the difffference isn't really important. mA - positive is charging. | 01:25 |
| azakai | ok, cool | 01:25 |
| SpeedEvil | The other ones are probably not very usefull. | 01:26 |
| azakai | do I need the 'power kernel' mentioned before | 01:26 |
| azakai | ? | 01:26 |
| DocScrutinizer51 | charge% is largely BS | 01:26 |
| SpeedEvil | no | 01:26 |
| azakai | ok | 01:26 |
| SpeedEvil | DocScrutinizer: Actually - once it's learned it's not. | 01:26 |
| SpeedEvil | But yes, as it probably won't initially learn, it's largely bs. | 01:27 |
| DocScrutinizer51 | but you harly ever get learn complete | 01:27 |
| SpeedEvil | But - when idle - it will get around 10-12mA | 01:27 |
| SpeedEvil | The battery capacity is around 1200mAh or so - so it lasts around a hundred hours fairly easily. | 01:27 |
| DocScrutinizer51 | current now in mA is highly accurate avrg over last Ts | 01:28 |
| DocScrutinizer51 | 5s | 01:28 |
| SpeedEvil | Reading this every 5s does not affect the power use that much. It adds a small constant, but it is fairly small. | 01:28 |
| DocScrutinizer51 | given you have correct RS | 01:28 |
| azakai | hmm | 01:29 |
| azakai | mAh would be the battery remaining, but that isn't listed? | 01:29 |
| DocScrutinizer51 | which regrettably Nokia fails to disclose | 01:29 |
| SpeedEvil | NAC = normally available capacity in mAh | 01:29 |
| SpeedEvil | This will be inaccurate. | 01:30 |
| SpeedEvil | However. | 01:30 |
| DocScrutinizer51 | might be considerably off | 01:30 |
| SpeedEvil | When discharging, it will decrease monotonically at the correct rate | 01:30 |
| DocScrutinizer51 | diff in mAh is accurate | 01:30 |
| DocScrutinizer51 | abs value is not | 01:30 |
| azakai | hmm those numbers look similar to what I saw before with hal-device bme | grep reporting, is that the same, or is this more accurate? | 01:30 |
| SpeedEvil | So if it says NAC=1200 - then half an hour later - 1100 - you've used 100mAh in 1/2 hour | 01:30 |
| SpeedEvil | These are more accurate in some ways. | 01:31 |
| DocScrutinizer51 | yep | 01:31 |
| azakai | ok, that sounds good :) | 01:31 |
| SpeedEvil | If you force several learn cycles, the battery charge meter can be really very accurate | 01:31 |
| SpeedEvil | But that currently involves stopping BME, as it won't do a learn cycle without it. | 01:31 |
| DocScrutinizer51 | btw the script isn't from stskeeps | 01:31 |
| SpeedEvil | I suggest skimming the datasheet on the charge meter page. | 01:32 |
| SpeedEvil | It's not? | 01:32 |
| SpeedEvil | the hi.sh one is the bq27200 script hacked a bit by me to just do current charge level reporting, and standby current | 01:32 |
| * SpeedEvil wishes the charge counter was 32 bits - but meh. | 01:32 | |
| DocScrutinizer51 | shadowjk | 01:33 |
| SpeedEvil | err | 01:33 |
| SpeedEvil | yes | 01:33 |
| DocScrutinizer51 | and mine is much older still :-P | 01:34 |
| SpeedEvil | In short - do ./bq27200 -r >log | 01:34 |
| * SpeedEvil needs to get on with his energy profiler script. | 01:35 | |
| azakai | ok, I will try that. Thanks guys! | 01:35 |
| DocScrutinizer51 | yw | 01:35 |
| SpeedEvil | Basically poll most twiddles I can find, and try to associate wakeups and energy use to a process, and store that value. | 01:35 |
| DocScrutinizer51 | and don't worry bout bme | 01:35 |
| SpeedEvil | Good luck! | 01:35 |
| DocScrutinizer51 | SpeedEvil: timeslice*=100 might help | 01:36 |
| SpeedEvil | ? | 01:36 |
| DocScrutinizer51 | scheduler timeslice | 01:37 |
| SpeedEvil | Sorry - not following | 01:37 |
| DocScrutinizer51 | prolly 20ms now | 01:37 |
| SpeedEvil | ah | 01:37 |
| SpeedEvil | Perhaps. | 01:37 |
| SpeedEvil | I'm not yet at that stage yet | 01:37 |
| SpeedEvil | I'm trying to work out what kernel structures it's sane to poll every 5s. | 01:37 |
| SpeedEvil | And what ones are too heavyweight | 01:38 |
| SpeedEvil | For example - /proc/*/smaps might be nice, but... | 01:38 |
| DocScrutinizer51 | basically proc is heavy | 01:38 |
| SpeedEvil | yes | 01:38 |
| DocScrutinizer51 | man ps | 01:38 |
| SpeedEvil | I know that much. | 01:39 |
| SpeedEvil | My questions are more what bits of proc are unacceptably heavy. | 01:39 |
| DocScrutinizer51 | hmm difficult | 01:39 |
| SpeedEvil | For example, it would be nice if I could readout per-proceess net-meters. | 01:39 |
| SpeedEvil | But I think that would at least require iptables hackery | 01:40 |
| SpeedEvil | and a recent kernel | 01:40 |
| DocScrutinizer51 | I honestly read in a older szsop guide you shouldn't use ps too lightly | 01:40 |
| SpeedEvil | Well - yes - but... | 01:40 |
| SpeedEvil | In some ways the processor speed has made that not so important. | 01:40 |
| SpeedEvil | For example - if a wakeup takes 20ms - and polling /proc takes 10ms - you're not going to save much energy by shaving that to 1ms. | 01:41 |
| DocScrutinizer51 | (netmeter) I wonder how cpu usage is done | 01:41 |
| DocScrutinizer51 | probably net usage could work similar | 01:41 |
| SpeedEvil | That, and it would be nice to have a more detailed breakdown of MHz/state | 01:42 |
| SpeedEvil | For example - I see 34 'wakeups' | 01:42 |
| SpeedEvil | How many of those were clustered in the same millisecond. | 01:42 |
| SpeedEvil | In which case, they're really lightweight. | 01:42 |
| SpeedEvil | But if the processor was woken out of C4... | 01:43 |
| DocScrutinizer2 | what you want is transitions, not states | 01:43 |
| * DocScrutinizer2 pops beer bottle to start the more relaxed part of evening | 01:43 | |
| SpeedEvil | yes | 01:45 |
| SpeedEvil | What initiated the series is also more interesting thatn the series of course. For example - sms-widget polling the SMS database every 5 seconds. | 01:46 |
| SpeedEvil | most of the wakeups are in other stuff. | 01:46 |
| DocScrutinizer2 | maemo is doing really interesting stuff, like grouping alarms | 01:47 |
| SpeedEvil | yes - there is some really clever stuff in bits. | 01:48 |
| SpeedEvil | Trying to work out where it's being clever, and where it's being dumb is going to take more reading. | 01:48 |
| DocScrutinizer2 | all that is prolly SmartReflex(TM)_the_sw_part | 01:49 |
| *** DocScrutinizer2 is now known as DocScrutinizer | 01:49 | |
| SpeedEvil | I guess that's what much of the 'i2c_omap' when idle is. | 01:50 |
| DocScrutinizer | i2c_omap is SmartReflex - regulator stuff | 01:51 |
| DocScrutinizer | I'd guess | 01:51 |
| SpeedEvil | Currently also wondering if you can configure the FM radio to go direct to the phones | 01:51 |
| DocScrutinizer | direct? | 01:52 |
| SpeedEvil | As at the moment FMradio (a bit of software that is at least workable in its latest incarnation) goes through a/d, and d/a to get to the phones | 01:52 |
| DocScrutinizer | aiui it's inside BT chip | 01:52 |
| SpeedEvil | So it can keep it awake | 01:52 |
| SpeedEvil | yeah | 01:52 |
| SpeedEvil | but it comes out as L+R audio, that goes into the audio codec | 01:52 |
| DocScrutinizer | really? | 01:52 |
| DocScrutinizer | lemme check | 01:53 |
| SpeedEvil | http://wiki.maemo.org/N900_Hardware_Schematic | 01:54 |
| SpeedEvil | pages 8 and 9 looks like | 01:55 |
| DocScrutinizer | LINE2x_PA | 01:55 |
| SpeedEvil | It goes into one of the line in ports of the codec - but I haven't looked through the internal diagram of the codec to see about it properly. | 01:55 |
| DocScrutinizer | yep | 01:55 |
| DocScrutinizer | it's 2 codecs back2back | 01:55 |
| DocScrutinizer | weird shit | 01:56 |
| DocScrutinizer | L3/4 has some details | 01:56 |
| SpeedEvil | And lots of fun routers. | 01:56 |
| SpeedEvil | Was playing earlier with audio minimum power - got it down by some 30-40% over normal case - but still seeing 100 wakeups/second - even on aplay /dev/zero | 01:58 |
| SpeedEvil | Which I guess is due to the small audio buffer. | 01:58 |
| DocScrutinizer | --buffer-time=5000000 (us!) | 01:59 |
| DocScrutinizer | even 10000000 | 01:59 |
| DocScrutinizer | or | 02:00 |
| SpeedEvil | yes | 02:00 |
| SpeedEvil | diddn't work | 02:00 |
| DocScrutinizer | --period-time=1000000 | 02:00 |
| SpeedEvil | '' | 02:00 |
| DocScrutinizer | :-/ | 02:00 |
| DocScrutinizer | PA crap | 02:00 |
| SpeedEvil | no | 02:00 |
| SpeedEvil | PA dead | 02:00 |
| DocScrutinizer | ooh | 02:00 |
| SpeedEvil | Played audio through headphones, captured state | 02:00 |
| DocScrutinizer | weird | 02:00 |
| SpeedEvil | killed pulse, reset state, then aplay | 02:01 |
| SpeedEvil | worked fine | 02:01 |
| DocScrutinizer | I guess the crappy codec just has next to zero buffer | 02:01 |
| SpeedEvil | the codec has exactly zero buffer | 02:02 |
| SpeedEvil | it's in the SoC | 02:03 |
| DocScrutinizer | so basically things are double buffered. one is the app buffer which you config with --period-time= and which is passing its segments to the driver. the other is a 4 words buffer in the codec and what you see are the wakeups when this one gets empty and driver needs to refill via I2S | 02:03 |
| DocScrutinizer | MCBSP2(4:0) | 02:06 |
| DocScrutinizer | CPU > OMAP3430 | 02:06 |
| DocScrutinizer | SoC -> codec digital audio | 02:06 |
| DocScrutinizer | . | 02:07 |
| DocScrutinizer | MONO_LOM_A | 02:08 |
| DocScrutinizer | MONO_LOP_A | 02:08 |
| DocScrutinizer | LINE2RM_B | 02:08 |
| DocScrutinizer | LINE2RP_B | 02:08 |
| DocScrutinizer | is the bridge between the 2 half codecs | 02:08 |
| *** swc|666 has quit IRC | 02:11 | |
| DocScrutinizer | SpeedEvil: "audio concept" 6-18 p.178 | 02:12 |
| DocScrutinizer | in L3/4 | 02:12 |
| SpeedEvil | hmm | 02:13 |
| SpeedEvil | The wakeups are not 10000 times a second - but 100 | 02:14 |
| SpeedEvil | So I'm assuming they are the internal SOC 'soundcard' | 02:14 |
| SpeedEvil | Hmm - so if the two halves can be bridged under programmatic control - it should be possible to simply with the right mixer config - play radio with no CPU? | 02:15 |
| DocScrutinizer | err, lemme check | 02:16 |
| DocScrutinizer | got a ds for AIC34? | 02:18 |
| SpeedEvil | yes | 02:18 |
| SpeedEvil | http://focus.ti.com/docs/prod/folders/print/tlv320aic34.html | 02:19 |
| DocScrutinizer | ty | 02:19 |
| SpeedEvil | I've started linking the schematic pages. | 02:21 |
| DocScrutinizer | LINE2xX_A is routed to A block Mixing/Muxing Volume Controls and then to LEFT/RIGHTxOx_A and HP-AMP | 02:26 |
| DocScrutinizer | so the answer is: positively yes, FMRX to HP is possible without any CPU | 02:27 |
| *** csaavedra has quit IRC | 02:27 | |
| SpeedEvil | neat! | 02:29 |
| SpeedEvil | Now - what's the betting you can't actually achieve this with the stock config :) | 02:29 |
| DocScrutinizer | hard to find anybody holding the bet | 02:29 |
| DocScrutinizer | well, we got I2Ctools :-P | 02:30 |
| SpeedEvil | :) | 02:30 |
| SpeedEvil | I suspect alsamixer will actually do just fine | 02:30 |
| DocScrutinizer | I2C_2 | 02:30 |
| DocScrutinizer | should | 02:31 |
| SpeedEvil | you need to add pcm.real { | 02:31 |
| SpeedEvil | type hw | 02:31 |
| SpeedEvil | card 0 | 02:31 |
| SpeedEvil | } | 02:31 |
| SpeedEvil | /... to /etc/asound.conf though | 02:31 |
| DocScrutinizer | LOL | 02:31 |
| *** t7g_ has joined #maemo-devel | 02:32 | |
| SpeedEvil | otherwise the default goes to pulseaudio | 02:32 |
| DocScrutinizer | those FR****NG F****NG bastards | 02:32 |
| DocScrutinizer | I guess I should sanitize the whole ALSA config | 02:33 |
| DocScrutinizer | then *maybe* let in PA as a guest, later on :-P | 02:34 |
| DocScrutinizer | not to PCM though, maybe to DMIX | 02:34 |
| SpeedEvil | I wonder how much of the stock software actually talks to pulse, and how muhc just uses alsa | 02:35 |
| DocScrutinizer | that's the nice thing in ALSA, each plugin's ABI is identical, and so is the PCM | 02:35 |
| *** T7g has quit IRC | 02:36 | |
| DocScrutinizer | a lot uses ALSA, alas that'S the ALSA->PA comaptibility implementation (I don't want to call that a plugin) | 02:36 |
| DocScrutinizer | nothing is using generic ALSA | 02:37 |
| DocScrutinizer | except PA | 02:37 |
| DocScrutinizer | and probably that's a PITA as all the policy enforcing engine is a PA bastard | 02:38 |
| DocScrutinizer | so none of the "scenario" switching will work without PA | 02:39 |
| DocScrutinizer | afk, scientific TV | 02:39 |
| SpeedEvil | ah | 02:40 |
| SpeedEvil | wave | 02:40 |
| DocScrutinizer51 | forgot this one :-P | 02:40 |
| *** macbeth8c has quit IRC | 02:52 | |
| *** njsf_ has joined #maemo-devel | 03:00 | |
| *** njsf_ has quit IRC | 03:00 | |
| *** njsf_ has joined #maemo-devel | 03:00 | |
| *** SpeedEvil has quit IRC | 03:29 | |
| *** SpeedEvil has joined #maemo-devel | 03:29 | |
| *** rcampbell has joined #maemo-devel | 03:34 | |
| *** Aranel has joined #maemo-devel | 03:38 | |
| *** Aranel has quit IRC | 03:38 | |
| *** Aranel has joined #maemo-devel | 03:38 | |
| *** swc|666 has joined #maemo-devel | 03:38 | |
| *** swc|666 has joined #maemo-devel | 03:39 | |
| *** Aranel has quit IRC | 03:39 | |
| *** Aranel has joined #maemo-devel | 03:57 | |
| *** ptl has quit IRC | 04:09 | |
| *** ptl has joined #maemo-devel | 04:20 | |
| *** ptl has quit IRC | 04:20 | |
| *** ptl has joined #maemo-devel | 04:20 | |
| *** njsf_ has quit IRC | 04:26 | |
| *** ptl has quit IRC | 04:39 | |
| *** azakai has quit IRC | 04:40 | |
| *** GeneralAntilles has joined #maemo-devel | 04:44 | |
| *** ptl has joined #maemo-devel | 04:55 | |
| *** ptl has quit IRC | 04:55 | |
| *** ptl has joined #maemo-devel | 04:55 | |
| *** GeneralAntilles has quit IRC | 05:20 | |
| *** njsf_ has joined #maemo-devel | 05:26 | |
| *** swc|666 has quit IRC | 05:37 | |
| *** GeneralAntilles has joined #maemo-devel | 05:39 | |
| *** GeneralAntilles has quit IRC | 06:19 | |
| *** swc|666 has joined #maemo-devel | 06:20 | |
| *** GeneralAntilles has joined #maemo-devel | 06:26 | |
| *** uv__ has joined #maemo-devel | 06:54 | |
| *** uv__ is now known as uvatbc | 06:54 | |
| *** uv has joined #maemo-devel | 06:57 | |
| uv | Hi everyone, I'm trying to make an app that uses telepathy-qt4. I have it compiled and running on my linux machine and inside scratchbox | 06:58 |
| uv | I want to be able to compile it using the Nokia SDK. Here's what I've thought of trying: | 06:58 |
| uv | download tp-qt4 and extract it. Load it as a project in qtcreator, compile it, link it against my application (also loaded) in qtcreator | 06:59 |
| uv | and compile everything | 06:59 |
| uv | The gotcha in this is tp-qt4 requires to run ./configure (I could make that a build step) | 07:00 |
| uv | but is there any cleaner way? | 07:00 |
| uv | Ideally I'd like to be able to setup the environment to cross compile using the Nokia SDK, then compile tp-qt4 and install the compiled binaries into the Nokia SDK directories. | 07:01 |
| *** GeneralAntilles has quit IRC | 07:06 | |
| *** Aranel has quit IRC | 07:07 | |
| *** GeneralAntilles has joined #maemo-devel | 07:12 | |
| *** njsf_ has quit IRC | 07:23 | |
| *** DocScrutinizer is now known as DocScrutinizer2 | 07:24 | |
| *** DocScrutinizer2 has quit IRC | 07:30 | |
| *** DocScrutinizer has joined #maemo-devel | 07:30 | |
| *** rcampbell has quit IRC | 07:51 | |
| *** kamui__ has joined #maemo-devel | 08:03 | |
| *** shinkamui has quit IRC | 08:07 | |
| *** timoph|AFK is now known as timoph | 08:15 | |
| *** tealbird has joined #maemo-devel | 08:22 | |
| *** kamui__ is now known as shinkamui | 08:33 | |
| *** uv has quit IRC | 09:04 | |
| *** uvatbc has quit IRC | 09:04 | |
| *** bricks has joined #maemo-devel | 09:11 | |
| *** DocScrutinizer51 has quit IRC | 09:17 | |
| *** DocAvalance has joined #maemo-devel | 09:24 | |
| *** DocAvalance is now known as DocScrutinizer51 | 09:25 | |
| *** DocScrutinizer51 has joined #maemo-devel | 09:25 | |
| *** DocScrutinizer51 has quit IRC | 09:34 | |
| *** DocAvalance has joined #maemo-devel | 09:36 | |
| *** pgas has quit IRC | 09:39 | |
| *** frals_ has joined #maemo-devel | 09:51 | |
| *** frals_ has quit IRC | 10:05 | |
| *** frals_ has joined #maemo-devel | 10:09 | |
| *** pgas has joined #maemo-devel | 10:18 | |
| *** amigadave has joined #maemo-devel | 10:21 | |
| *** DocAvalance is now known as DocScrutinizer51 | 10:27 | |
| *** DocScrutinizer51 has quit IRC | 10:27 | |
| *** DocScrutinizer51 has joined #maemo-devel | 10:27 | |
| *** achipa has joined #maemo-devel | 10:34 | |
| *** swc|666 has quit IRC | 10:34 | |
| *** csaavedra has joined #maemo-devel | 10:35 | |
| *** tealbird has quit IRC | 11:57 | |
| *** tealbird has joined #maemo-devel | 12:09 | |
| Lullen | Hmm... I just succeeded to upload a package to the autobuilder(wich succeeded) that is 0kb and have no dependency... How can this be? http://maemo.org/packages/package_instance/view/fremantle_extras-devel_free_i386/qexercise/0.1-2/ | 12:25 |
| Lullen | All I done from 0.1-1 is adding a icon, desktop file and edit the section in the control file | 12:25 |
| lcuk | Lullen, i guess the change i nthe control file did something unexpected! | 12:27 |
| Lullen | i can post it | 12:33 |
| Lullen | http://pastebin.ca/1885654 | 12:34 |
| lcuk | Lullen, and what was it before.. | 12:36 |
| lcuk | (it might not be this but from the mods you have made its the only one that has potential to effect things | 12:36 |
| Lullen | hmm | 12:37 |
| Lullen | it was unknown at the section I guess | 12:38 |
| Lullen | and I removed the homepage line | 12:38 |
| Lullen | and editied the last line | 12:39 |
| Lullen | (the description) | 12:39 |
| frals_ | Lullen: it takes one import cycle before you can see your package in packages interface | 13:11 |
| frals_ | ie, if you check that link now its fine | 13:12 |
| Lullen | now it seems okey only that the homepage is not removed but anyhow... It have been imported and the section is user/utilities. I can not find it in application manager and when I try to use "apg-get install qexercise" I get the following error: "Package qexercise is not available, but is referred to by an other package. This may mean that the package is missing, has been obsoleted, or is only available from another source" | 13:23 |
| frals_ | did you check that its imported in to -devel on maemo.org/packages/view/qexercise? | 13:37 |
| *** lizardo has joined #maemo-devel | 13:41 | |
| Lullen | RepositoryLatest version | 13:41 |
| Lullen | Fremantle Extras-devel free armelqexercise 0.1-3 | 13:41 |
| Lullen | it should be right | 13:41 |
| frals_ | sounds about right | 13:42 |
| flux | is there some maemo application around that makes use of libnotify and has sources available? | 13:42 |
| frals_ | and you did apt-get update before trying to install? | 13:42 |
| Lullen | yes | 13:42 |
| flux | I'm trying to replicate what happens when I get new sms messages: however many messages I get, only one notification bubble is visible (I get a new one for each message) | 13:43 |
| flux | also, clicking the notification window opens the application that sent the notification; in my use, it opens a new window (which goes away when opened) | 13:43 |
| frals_ | Lullen: apt-get install qexercise works fine for me | 13:44 |
| Lullen | really? | 13:45 |
| Lullen | strange... | 13:45 |
| Lullen | can you open up the program too? | 13:45 |
| frals_ | flux: i think you might have to create a new "notification" group in /etc/something/rtcomgroups or something like that to get the same behavior | 13:46 |
| frals_ | Lullen: it opened yeah | 13:47 |
| flux | frals_, ah, thanks. I found /etc/hildon-desktop/notification-groups.conf | 13:47 |
| Lullen | Thanks I'll try to reflash my device | 13:47 |
| frals_ | flux: fwiw i tried messing with that at some point but never really got it to working despite setting the category to my own and restarting hildon-desktop | 13:47 |
| frals_ | flux: but i was just trying to hack it quick, if you manage to solve it id love to hear about it :) | 13:48 |
| flux | frals_, I'll make a TODO note about that :-). | 13:48 |
| flux | ah, hildon-desktop has its sources available | 13:50 |
| frals_ | yeah so does libhildonnotify and err | 13:50 |
| frals_ | i think hildonsvnotificationdaemon or such | 13:50 |
| flux | hmph, except hildon-desktop apparently has nothing to do with that | 13:50 |
| frals_ | which might be interesting as well | 13:50 |
| flux | you mean hildon-sv-notification-daemon | 13:52 |
| flux | apparently it's only for sound and vibra | 13:52 |
| flux | but maye the description is misleading :) | 13:52 |
| flux | Unable to find a source package for hildon-plugins-notify-sv :( | 13:52 |
| flux | hildon-home knows about notification groups | 13:53 |
| *** csaavedra has quit IRC | 13:53 | |
| flux | frals_, so maybe you were on the right track, but you should've restarted hildon-home instead? | 14:01 |
| *** bricks has quit IRC | 14:10 | |
| frals_ | might have been, but i think i even rebooted device without it working :) | 14:16 |
| *** tealbird has quit IRC | 14:26 | |
| *** csaavedra has joined #maemo-devel | 14:47 | |
| *** ktzqbp has joined #maemo-devel | 14:54 | |
| *** GeneralAntilles has quit IRC | 15:42 | |
| *** GeneralAntilles has joined #maemo-devel | 15:49 | |
| *** fiferboy has joined #maemo-devel | 15:58 | |
| *** fiferboy has quit IRC | 15:58 | |
| *** fiferboy has joined #maemo-devel | 15:58 | |
| Lullen | Anyone knows why labels don't update when changing the text when doing it in a function before(or directly after) I call window.show() ? | 16:15 |
| Lullen | I have tried to update & repaint the label & the whole window directly after but nothing updates. If I make a button that calls label.update when pushed everything is right | 16:21 |
| *** madsy has quit IRC | 16:22 | |
| w00t_ | Lullen: got a source snippet? | 16:24 |
| w00t_ | (http://pastebin.com) | 16:24 |
| *** edisson has joined #maemo-devel | 16:39 | |
| *** madsy has joined #maemo-devel | 16:54 | |
| *** achipa has quit IRC | 16:58 | |
| *** frals_ has quit IRC | 17:07 | |
| *** amigadave has quit IRC | 17:36 | |
| *** hellodave has joined #maemo-devel | 17:54 | |
| *** hellodave has left #maemo-devel | 17:56 | |
| *** macbeth8c has joined #maemo-devel | 18:20 | |
| macbeth8c | hi SpeedEvil | 18:20 |
| macbeth8c | how to set the code fragment to compile only on device? | 18:20 |
| SpeedEvil | ? | 18:21 |
| SpeedEvil | Dunno | 18:21 |
| SpeedEvil | what gode fragment | 18:21 |
| SpeedEvil | code | 18:21 |
| macbeth8c | I want to use accelerometer | 18:21 |
| macbeth8c | if on device | 18:21 |
| macbeth8c | #ifdef something | 18:21 |
| SpeedEvil | Why? | 18:21 |
| SpeedEvil | If(exist("/sys/whatever")) | 18:22 |
| macbeth8c | #ifdef ARM ? | 18:22 |
| lcuk | macbeth8c, many arm devices exist with and without accelerometer | 18:25 |
| lcuk | just do as SpeedEvil suggests and test existence? | 18:25 |
| macbeth8c | in my case the app will be used only on N900 and PC | 18:27 |
| fragment | mmee | 18:27 |
| macbeth8c | lcuk, I am reading your post in the forum in accelerometers connected thread :D | 18:33 |
| lcuk | macbeth8c, i make lots of posts - link? | 18:33 |
| macbeth8c | http://talk.maemo.org/showthread.php?p=288990 | 18:33 |
| lcuk | heh that was ages ago | 18:34 |
| macbeth8c | lcuk, can I setup the accelerometer to work with scratchbox? | 18:39 |
| macbeth8c | (with Ubuntu) | 18:39 |
| lcuk | whenever i do that i find it a bit heavy and unweildy to tilt my desktop | 18:39 |
| SpeedEvil | My laptop has an accel. :) | 18:40 |
| lcuk | yeah i know :p | 18:40 |
| lcuk | they dont have the same branch tho | 18:40 |
| macbeth8c | lcuk, I mean to setup some configuration what should it return | 18:40 |
| lcuk | macbeth8c, it just says file not found.. | 18:41 |
| lcuk | because hte filesystem doesnt have it - what we should have done with accelerometer is kernel driver to forward the i2c to the proper normal expected location of accelerometer stuffs | 18:41 |
| lcuk | then same apps would work on different machines (like laptops) without any fuding | 18:42 |
| lcuk | fudging | 18:42 |
| macbeth8c | has someone done it? | 18:43 |
| lcuk | well i just had a function to read from the alternative location if the first one failed | 18:43 |
| lcuk | tho i havent used it in a while | 18:43 |
| macbeth8c | ok | 18:43 |
| macbeth8c | /sys/class/i2c-adapter/i2c-3/3-001d/coord | 18:47 |
| macbeth8c | is it regular text file on the device? | 18:48 |
| lcuk | yes | 18:49 |
| lcuk | rs=fscanf((FILE*) fd,"%i %i %i",ax,ay,az); | 18:49 |
| macbeth8c | yea I see the example | 18:50 |
| macbeth8c | nice | 18:50 |
| macbeth8c | :) | 18:50 |
| macbeth8c | but why they have those oa? variables? | 18:52 |
| macbeth8c | static int ocnt=0; | 18:53 |
| macbeth8c | static int oax=0; | 18:53 |
| macbeth8c | static int oay=0; | 18:53 |
| macbeth8c | static int oaz=0; | 18:53 |
| lcuk | macbeth8c, original versions, reading the values raw - theres a lot of jitter | 18:55 |
| lcuk | so the smoothing only moves a fraction towards the reported value | 18:56 |
| lcuk | and so after ~10 calls to the function its nice and smooth :) | 18:56 |
| macbeth8c | ok | 18:56 |
| *** azakai has joined #maemo-devel | 19:02 | |
| macbeth8c | lcuk, thats good, I need to allign the picture so that it is easier for me to detect the face :) | 19:04 |
| lcuk | cool | 19:05 |
| *** timoph is now known as timoph|AFK | 19:08 | |
| macbeth8c | lcuk, beautiful I have the code for rotation. I will test it this evening on the device :-) | 19:15 |
| lcuk | macbeth8c, what are you writing it in? | 19:15 |
| DocScrutinizer51 | lcuk: actually smoothing should be done in LIS302 | 19:24 |
| lcuk | DocScrutinizer51, many things should be there - is there an api for it (and calibration too) | 19:25 |
| DocScrutinizer51 | lcuk: it's not like we want to poll LIS302 with 400Hz. Rather it should be set so it generates an IRQ whenever smoothed values change | 19:26 |
| lcuk | examples of it happening elsewhere? | 19:26 |
| lcuk | and that would be worse - it only gets read per frame | 19:26 |
| lcuk | and does change | 19:27 |
| lcuk | mazimum is 58hz at fullres | 19:27 |
| DocScrutinizer51 | lis302 has an adjustable highpass filter and theshold for that purpose | 19:27 |
| lcuk | maximum | 19:27 |
| lcuk | so how is it spoken to | 19:27 |
| *** azakai has quit IRC | 19:29 | |
| *** azakai has joined #maemo-devel | 19:30 | |
| DocScrutinizer51 | damn disconnects | 19:30 |
| DocScrutinizer51 | lcuk: it's not like we want to poll LIS302 with 400Hz. Rather it should be set so it generates an IRQ whenever smoothed values change | 19:31 |
| DocScrutinizer51 | lis302 has an adjustable highpass filter and theshold for that purpose | 19:31 |
| lcuk | <lcuk> examples of it happening elsewhere? | 19:32 |
| lcuk | <lcuk> and that would be worse - it only gets read per frame | 19:32 |
| lcuk | <lcuk> and does change | 19:32 |
| lcuk | <lcuk> mazimum is 58hz at fullres | 19:32 |
| lcuk | <lcuk> maximum | 19:32 |
| lcuk | <lcuk> so how is it spoken to | 19:32 |
| *** kamui__ has joined #maemo-devel | 20:03 | |
| *** amigadave has joined #maemo-devel | 20:03 | |
| *** shinkamui has quit IRC | 20:07 | |
| *** janu has joined #maemo-devel | 20:14 | |
| janu | can anyone tell how to get SIM card info ??? which API ? | 20:15 |
| SpeedEvil | What sort of info? | 20:17 |
| janu | SIM unique info | 20:18 |
| janu | my application should be able to guess if the SIM has been changed | 20:18 |
| SpeedEvil | gconftool-2 --dump / |less | 20:29 |
| SpeedEvil | search on imsi in lower-case | 20:29 |
| SpeedEvil | It's in there | 20:29 |
| *** Venemo has joined #maemo-devel | 20:49 | |
| DocScrutinizer | lcuk: what got frames to do with lis302? | 20:53 |
| DocScrutinizer | I mean, the whole API is fsckd up. You'd have to specify a minimum delta and a upper transit freq for a lowpass filter to smooth out "jitter", and then the LIS302 shouldn't wake CPU until there's actually a new value tuple to report, within bounds of granularity the app specified | 20:56 |
| DocScrutinizer | btw almost identical considerations apply to libts | 20:57 |
| DocScrutinizer | as well as virtually every analog sensor you could think of | 20:58 |
| lcuk | DocScrutinizer, i mean the orientation is read as part of the frame refresh code - just like any game etc | 21:04 |
| lcuk | i personally dont care whether they have changed, they are just plugged into the equation | 21:05 |
| DocScrutinizer | hmm, for an app doing a high load sheduled workthread like video playback per-frame that's for sure ok. But what with daemons like the one that rises the dialer on turning to portrait. I don't think you'd want that one poll a sysnode on a regular basis. Rather you open a blocking read and wait for the data dropping in, so the whole proces is idle on central wait until sysnode delivers trigger in form of new data | 21:08 |
| lcuk | DocScrutinizer, and that doesnt happen | 21:09 |
| lcuk | theres events for that afaik | 21:09 |
| DocScrutinizer | what's an 'event' here? | 21:09 |
| DocScrutinizer | event like readf() returning with a buffer full of data? | 21:10 |
| lcuk | DocScrutinizer, not sure, never looked into how the rotation support was implemented | 21:13 |
| lcuk | but afaik it comes in as either a dbus event | 21:13 |
| lcuk | or through gtk whatnots | 21:14 |
| lcuk | DocScrutinizer, can you get rid of the "WARNING: Trolls will be booted on first troll" from the top | 21:14 |
| DocScrutinizer | lcuk: if there's no other redundant way to handle e.g. accelerometer, then the coords sysnode is lowest level, and no matter which process/app/daemon/handler is dealing with it, it simply mustn't be polled | 21:14 |
| DocScrutinizer | me? nope | 21:15 |
| lcuk | theres no need at all for such a crappy warning - i dont think anyone has ever even opped up in here | 21:15 |
| DocScrutinizer | no chanop here | 21:15 |
| *** dazol has quit IRC | 21:15 | |
| lcuk | and it just sits there talking about trolls to everyone whos here | 21:15 |
| DocScrutinizer | i'm^ | 21:17 |
| DocScrutinizer | ping one of the ops for it | 21:17 |
| lcuk | wouldnt know who | 21:17 |
| lcuk | only op i know is you | 21:18 |
| DocScrutinizer | seems stskeeps has resigned on that role - thought he's been the original owner of this chan | 21:18 |
| DocScrutinizer | /msg chanserv access #maemo-devel list | 21:18 |
| lcuk | ta DocScrutinizer | 21:19 |
| DocScrutinizer | np | 21:20 |
| *** Mek has quit IRC | 21:20 | |
| *** swc|666 has joined #maemo-devel | 21:22 | |
| *** Aranel has joined #maemo-devel | 21:23 | |
| *** Aranel has quit IRC | 21:23 | |
| *** Aranel has joined #maemo-devel | 21:23 | |
| *** Mek has joined #maemo-devel | 21:23 | |
| janu | can anyone tell how to get SIM card info(SIM unique) ??? which API ? | 21:27 |
| Venemo | perhaps Mobility API? | 21:28 |
| Venemo | but I don't know if it is possible, because of obvious security reasons | 21:28 |
| *** edisson has quit IRC | 21:29 | |
| lcuk | GeneralAntilles, are you at home | 21:33 |
| *** amigadave has quit IRC | 21:42 | |
| *** tealbird has joined #maemo-devel | 21:42 | |
| *** GAN900 has joined #maemo-devel | 21:45 | |
| *** ChanServ sets mode: +o GAN900 | 21:45 | |
| *** GAN900 changes topic to "Welcome to #maemo-devel | Logs: http://mg.pov.lt/maemo-devel-irclog/ | http://maemo.org/development/" | 21:45 | |
| *** GAN900 sets mode: -o GAN900 | 21:46 | |
| lcuk | \o/ | 21:46 |
| lcuk | kickass thanks GAN900 | 21:46 |
| *** amigadave has joined #maemo-devel | 21:48 | |
| *** amigadave has quit IRC | 21:49 | |
| *** ktzqbp has quit IRC | 21:56 | |
| *** tealbird has quit IRC | 22:01 | |
| Venemo | hi people | 22:03 |
| Venemo | I have a question for those who are familiar with gconf | 22:03 |
| Venemo | I have this line of code: | 22:04 |
| Venemo | gconf_value_get_int(gconf_entry_get_value(gconf_client_get_entry(gconfClient, "/apps/osso/hildon-desktop/views/current", NULL, true, NULL))); | 22:04 |
| Venemo | does it introduce a memory leak? | 22:04 |
| Venemo | I mean, should I put each of those function calls in pointers and then delete them afterwards? | 22:05 |
| *** kW has joined #maemo-devel | 22:07 | |
| *** GAN900 has left #maemo-devel | 22:07 | |
| lcuk | Venemo, i think you should - it looks to me like it would | 22:09 |
| lcuk | not necessarily delete - but release afterwards | 22:09 |
| Venemo | what do you mean? | 22:09 |
| Venemo | to me, release means the delete operator in C++ | 22:09 |
| Venemo | or free() in C | 22:09 |
| Venemo | lcuk: so, what do you mean by "not delete but release"? | 22:11 |
| *** kW_ has quit IRC | 22:11 | |
| lcuk | gconf_entry_unref (applet_entry); | 22:11 |
| Venemo | oh. | 22:11 |
| Venemo | okay, thank you | 22:13 |
| lcuk | need to double check, but looking at code which uses the get_entry() seems to have associated entry_unref() after finishing | 22:13 |
| lcuk | np Venemo good luck | 22:13 |
| *** BluesLee has joined #maemo-devel | 22:27 | |
| macbeth8c | Venemo, hi :-) | 22:31 |
| macbeth8c | Venemo, have you solved pkg config issues? | 22:31 |
| *** BluesLee has quit IRC | 22:31 | |
| lcuk | macbeth8c, you never said what you were writing your stuff in | 22:35 |
| macbeth8c | lcuk, you mean the lang? | 22:37 |
| lcuk | yeah | 22:37 |
| macbeth8c | C++ | 22:37 |
| lcuk | in qt or just native c++ on gtk? | 22:37 |
| macbeth8c | qt | 22:38 |
| lcuk | cool beans | 22:38 |
| macbeth8c | lcuk, I hope the application will work on other smartphones as well | 22:39 |
| macbeth8c | like IPhone OS or Android | 22:39 |
| lcuk | ihpone wont - theres no qt allowed is there? | 22:39 |
| lcuk | iphone rather | 22:39 |
| lcuk | and android is written in java | 22:40 |
| macbeth8c | oh | 22:40 |
| lcuk | symbian devices will work tho :) | 22:40 |
| w00t_ | supposedly, static compilation on iPhone might be allowed. but as Qt isn't yet ported there, that's a bit academic | 22:40 |
| w00t_ | and there is a Qt port to android | 22:40 |
| lcuk | qt on android? | 22:40 |
| lcuk | as in compiled to java? | 22:40 |
| w00t_ | http://code.google.com/p/android-lighthouse/ | 22:41 |
| w00t_ | "Before we start let's clarify what is an android application? Android applications are java apps packed into a package. In this package you can also embed one or more (shared) libs which can be loaded from java code and can interact with java using JNI. Currently you can't have a binary application in the package so, qt application, embedded in this bundle, are in fact shared libs." | 22:41 |
| w00t_ | it's a pretty awesome hack | 22:41 |
| lcuk | not really as the qt side would also need to interact back wit hthe java stack | 22:42 |
| Venemo | machbeth8c: yes, I did | 22:43 |
| Venemo | machbeth8c: it seems that I unnecessarily added some packages - after removing those, I don't get a timeout from sh_libdeps | 22:43 |
| lcuk | w00t_, static compilation of X framework on iphone was pretty much mooted since you have to write xcode dont you? | 22:43 |
| w00t_ | lcuk: http://blip.tv/file/3232378 is one of the earlier examples | 22:43 |
| w00t_ | lcuk: no idea, I don't bother myself with i*.. just repeating some discussion that I came across the other week from people who *do* care about it :-p | 22:44 |
| * lcuk nods | 22:46 | |
| macbeth8c | w00t_, my supervisor not gonna be very happy - he is an iMan ;-) | 22:46 |
| w00t_ | macbeth8c: well, they're a nice enough package.. i don't dislike them for the sake of it.. I just don't appreciate the lockdown and some other practices there | 22:47 |
| *** silbo_ has joined #maemo-devel | 22:53 | |
| *** njsf_ has joined #maemo-devel | 23:03 | |
| macbeth8c | w00t_, I must say Maemo is a good platform. TI can find only one disadvantage with the weight of device (N900) | 23:03 |
| w00t_ | :) | 23:03 |
| macbeth8c | hope they gonna improve weight in next releases | 23:03 |
| SpeedEvil | I personally don't have a problem bench pressing 180 grams. | 23:04 |
| SpeedEvil | Have you considered investing in a gym membership? | 23:05 |
| macbeth8c | SpeedEvil, you read mu mind ;-) | 23:05 |
| SpeedEvil | Well - the primary way to reduce weight would be by nuking the keyboard. | 23:05 |
| SpeedEvil | Which I'm not really a fan of. | 23:05 |
| macbeth8c | *my | 23:06 |
| macbeth8c | SpeedEvil, do you really think the keyboard is the major factor. I had an impression that most of the weight was placed in the upper part of the device where the screen was | 23:11 |
| SpeedEvil | macbeth8c: yes. | 23:11 |
| SpeedEvil | macbeth8c: the keyboard itself weighs almost nothing. | 23:11 |
| SpeedEvil | But that's missing the point. | 23:11 |
| SpeedEvil | If you lose the keyboard, you lose the mass of the keyboard. | 23:12 |
| SpeedEvil | But more importantly, you lose the mass of the slider. | 23:12 |
| SpeedEvil | You lose the mass of the top of the keyboard and the bottom of the screen | 23:12 |
| macbeth8c | yea | 23:12 |
| SpeedEvil | And you can slim down the sides, as they are now uninterrupted, and you can get the same stiffness with less plastic. | 23:12 |
| macbeth8c | true | 23:12 |
| SpeedEvil | I would guess that would slice off at least 35-45g | 23:13 |
| *** Venemo has left #maemo-devel | 23:48 | |
| *** lizardo has quit IRC | 23:52 | |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!