**** BEGIN LOGGING AT Sat Jan 16 02:59:58 2016 Jan 16 03:32:12 ceene: >> The Valid Discharge Flag (VDQ) in the FLAGS register is set when this condition occurs and remains set until the learning discharge cycle completes or an event occurs that disqualifies the learning cycle.<< and >>A learning cycle can be disqualified by any of the following conditions: 1. Cold temperature: Temperature ≤ TCOMP[3:0] (°C) when the EDV1 threshold voltage is reached. 2. Light load: A capacity learning cycle is Jan 16 03:32:14 disqualified if average current is less than or equal to 2 times the initial standby load when the EDV1 threshold voltage is reached. 3. Fast voltage drop: VOLT ≤ (EDV1 – 256 mV) before EDV1 is set. 4. Excessive charging: Cumulative Charge > 255 NAC counts (910 μVh) during a learning discharge cycle (alternating discharge/charge/discharge before EDV1 is set).<< Jan 16 03:34:43 this boils down to: don't charge during learning cycle (self evident), do discharge with a moderate load that's not too low but also not too high Jan 16 03:35:42 particularly don't cause load bursts when batt voltage is about to reach EDV1 Jan 16 03:36:29 ideally you keep backlight lit, this will warrant a proper calibration Jan 16 03:39:26 ~bq-calibrate Jan 16 03:39:27 methinks bq-calibrate is http://maemo.cloud-7.de/maemo5/patches_n_tools/calibrate-bq27k.sh, or http://talk.maemo.org/showpost.php?p=1053809&postcount=846 Jan 16 03:41:22 "Ho ever" :-P Jan 16 03:46:26 actually there's a lot of cargo cult action suggested in that tmo post. I suggest to rather ignore it all together, it's confusing at best Jan 16 03:52:19 infobot: no, bq-calibrate is http://maemo.cloud-7.de/maemo5/patches_n_tools/calibrate-bq27k.sh, or http://talk.maemo.org/showthread.php?p=1247595 Jan 16 03:52:20 DocScrutinizer05: okay Jan 16 04:12:56 DocScrutinizer05: there are a few other cases listed. I'm not entirely sure, but I think the "Excessive self-discharge" one might be relevant. Jan 16 04:14:29 when I got a new battery (polarcell), the LMD was at something like 1100 mAh, and after fully charging the new one and letting it discharge, it stayed at 6% with NAC = 68 for a few hours, then eventually NAC went to 67 and VDQ changed to 0 at the same time. Jan 16 04:15:24 so from there the NAC and RSOC dropped as I would expect, but VDQ had already been reset, presumably because it was discharging far more than it had expected to given the existing LMD. Jan 16 04:16:48 iirc, I might've observed that twice before just trying to discharge it less efficiently by leaving the light on for the last few hours. Jan 16 04:17:32 I have the log of it happing at least once. Jan 16 04:20:16 https://gist.github.com/Maxdamantus/fc0c0089eedbb5a00982 Jan 16 04:21:52 18:15 3760 6 6 -40 71 71 71 65535 105 27 1 0 Jan 16 04:22:10 20:39 3702 6 6 -127 68 68 68 65535 32 21 1 0 Jan 16 04:22:41 6% at 3.7 V for two and a half hours. Jan 16 04:24:17 Right, 68*16 = 1088, which was my LMD at the time. Jan 16 04:30:36 * Maxdamantus wonders why there's a big red "Not Recommended For New Designs" line at the top of every page in the bq27200 document. Jan 16 04:31:20 Because it was a bad system in the first place, or because it's simply been superceded (and whatever that is will end up with the same line eventually)? Jan 16 04:32:46 * Maxdamantus suspects the former. Jan 16 04:33:08 Actually, probably the latter Jan 16 04:33:17 I've seen that in a lot of 'old chip' pages Jan 16 04:33:33 not sure how other hardware does it, but I imagine it usually doesn't require typical people to do full discharges before telling them anything useful about the battery. Jan 16 04:33:37 Because they want you to use a new design chip(which is probably better supported) Jan 16 04:34:28 unless the devices have a couple of batteries that they can switch between as they calibrate each one. Jan 16 04:35:10 Y'know, I'm still not sure how accurate any of these devices battery meters are. I just know that, say, my Nexus 7 tablet charges to "100%" and shuts down at like 3%. Whether there's another 50% capacity in the battery, I'll never know. Jan 16 04:36:01 On the other hand, my old Iriver mp3 player with Rockbox on it had an even worse gauge; you'd do something stressful and it would drop by like 30% in one minute; let it sit there and it would come back up just as quickly. Jan 16 04:36:03 Yes, but whatever it does, it probably doesn't expect people like my mum to make sure she does a full discharge at least once every 30 cycles. Jan 16 04:36:50 That Rockbox thing probably just directly translates voltage to estimated charge. Jan 16 04:37:01 Yeah Jan 16 04:37:14 I've never done a full discharge(intentionally) on my N900, but I've never expected it to be 100% accurate, either, Jan 16 04:37:18 So you're just seeing the voltage drop. Jan 16 04:37:47 If you've never done a full discharge you're probably not relying on bq27200. Jan 16 04:38:11 Maemo doesn't use it by default. It's just some piece of hardware that is largely unused but wasn't removed. Jan 16 04:38:29 No, I guess not. I'm not sure how the stock calculator works either Jan 16 04:38:50 I think it works similarly. Jan 16 04:39:07 (just by looking at the voltage, and possibly some extra guesswork in software) Jan 16 04:40:00 The difference with bq27200 is that it also monitors current, so can more-or-less measure how much power is actually passing through. Jan 16 04:41:27 The caveat is that with that device in particular the user has to do a full discharge at least every 30 cycles. Jan 16 04:41:44 which is probably unreasonable to expect of most consumers. Jan 16 04:54:57 Cycle Count since Learning: 26 Total Cycle Count since last full reset: 247 Jan 16 08:00:52 hmm, shouldnt proper battery meter rely on more than one source of information? ie. charge amount, discharge amount, battery history Jan 16 08:01:08 and counting self discharge into it too Jan 16 08:21:44 What is "charge amount" and "discharge amount"? Jan 16 08:23:33 Batteries are weird. They don't exactly store a particular amount of charge. Jan 16 08:24:08 The amount of charge you get out depends on things like the discharge rate and ambient temperature. Jan 16 08:24:33 maxd: during charge one can measure roughly how much charge went in Jan 16 08:24:48 *roughly* Jan 16 08:25:16 That's what bq27200 does, though it makes its own mind up as to what information it provides. Jan 16 08:25:43 i wonder if bme-replacement does its own calculations Jan 16 08:25:58 own calculations based on what? Jan 16 08:26:05 on the data above Jan 16 08:26:24 I think bme-replacement is just a reimplementation of bme, which doesn't use bq27200. Jan 16 08:26:40 then it has to calculate charge state by its own Jan 16 08:27:02 If you want to use bq27200, you basically have to do what I said earlier—intentionally do a full discharge at least once every 30 cycles. Jan 16 08:27:27 It doesn't tell you how much charge has passed through it since the last query. Jan 16 08:27:46 It tells you what it thinks about the state of the battery based on its own calculations. Jan 16 08:28:34 offtopic: http://git.linaro.org/people/arnd.bergmann/flashbench.git Jan 16 08:30:07 SD cards are weird too. Jan 16 08:30:54 The speed of transfer depends on things like the datastructure the controller has built up over time for your block device. Jan 16 08:32:11 Maxdamantus: bme-replacement uses bq27xxx Jan 16 08:32:18 Ah, okay. Jan 16 08:32:36 That's probably what I'm using then. Jan 16 08:33:12 So there wasn't something that simply reimplemented bme (producing similar behaviour)? Jan 16 08:33:49 Maxdamantus: none I am aware of Jan 16 08:47:47 http://www.ti.com/product/BQ27200 Jan 16 08:47:54 “Requires No User Calibration” .. hmm Jan 16 08:48:37 Automatic Capacity Reduction With Age Jan 16 10:02:54 y Jan 16 10:03:01 xvv Jan 16 10:56:45 i have slight problem with yappari. If i type fast and press twice the same letter it makes the symbol from long press. Also somtimes single presses are registered as long presses for symbols. Does anyone else have this? And maybe it's more of a inconvenience than problem actually. Jan 16 10:57:14 vajb, this sounds more like general problem than app specific Jan 16 10:57:54 happens only in yappari Jan 16 10:58:44 it feels like app freezes for some microseconds and then continues and misses or wrongly prints my typing Jan 16 10:59:17 start iostat or vmmstat in another window and observe swap activity? Jan 16 11:01:56 i have cpu and memory monitor next to battery symbol and memory shows about half way and cpu peaks when wrong letters appear Jan 16 11:07:47 seems i don't have iostats nor vmmstat installed. Jan 16 11:08:06 its in sysstat pkg Jan 16 11:08:20 and its vmstat not vmm Jan 16 11:08:34 and iostat not iostats Jan 16 11:09:54 Vajb: what about in the terminal? Jan 16 11:10:17 I suspect removing/disabling HIM will fix it. Jan 16 11:10:35 if you don't need the VKB or "symbols" UI anyway. Jan 16 11:11:50 Vajb: first I'd disable the longpress-symbol nonsense Jan 16 11:11:58 ~jrtools Jan 16 11:11:58 extra, extra, read all about it, jrtools is http://wiki.maemo.org/User:Joerg_rw/tools Jan 16 11:12:14 i rarely use vkb.. Jan 16 11:12:25 don't have that problem in terminal Jan 16 11:12:30 gconftool-2 -s /apps/osso/inputmethod/ext_kb_repeat_enabled --type boolean true Jan 16 11:12:44 symbols i need sometimes Jan 16 11:13:14 then I'd probbaly try to renice yappari and see if that _maybe_ helps Jan 16 11:13:36 if i disable long press symbols how do i reach them then? fn+button with wanted symbol? Jan 16 11:13:46 exactly Jan 16 11:14:32 will fn lock still work? (double click locks it on) Jan 16 11:14:32 in the end I guess it's a bug in yappari which prolly has some busyloops Jan 16 11:14:41 sure Jan 16 11:14:48 i see. Jan 16 11:15:17 you *might* want to run yappari under strace Jan 16 11:15:27 actually that's why i asked if someone else is having this "feature" too Jan 16 11:15:33 and if you disable HIM altogether, sym (fn+ctrl) works as the normal Xorg compose key. Jan 16 11:15:52 im afraid i lack the skill of know-how there :) Jan 16 11:17:16 then try (re)nicing yappari. Pretty simple from terminal: nice yappari Jan 16 11:17:20 anyway, thx for all of ur suggestions. Will investigate it more when i get back home. Jan 16 11:18:39 says "no such folder" Jan 16 11:19:21 nice yappari says no such file or directory* Jan 16 11:20:13 then find the yappari binary. Prolly another fsckdup package that has no proper symlink in /usr/bin, so I'd search /opt for it Jan 16 11:20:52 then: nice /opt/yapparidir/yappari-starter or whatever the right path and name Jan 16 11:21:39 find /opt -iname '*yappari*' Jan 16 11:21:46 yes just looked opt it's not there Jan 16 11:22:30 start yappari, then use htop or ps aux to find the full pathname of the binary Jan 16 11:23:16 or look into the .desktop file, iirc it's to be found under /user/share/apps/hildon-desktop/ or something like that Jan 16 11:23:49 /usr/bin/Yappari Jan 16 11:24:04 maybe it was me typing y instead of Y Jan 16 11:24:36 nice'ing started yappari again Jan 16 11:24:56 Unsupported date format character at 1 (%-d.%-m.) Jan 16 11:25:14 and that was the output in terminal Jan 16 11:26:10 no output expected really Jan 16 11:26:22 just for the record: ls -l /usr/share/applications/hildon/*appar* Jan 16 11:27:42 anything specific to look from that output? Jan 16 11:28:52 sh: /usr/share/applications/hildon/yappari.desktop: Permission denied Jan 16 11:29:00 http://paste.opensuse.org/9388254 :-D Jan 16 11:29:01 last line Jan 16 11:29:03 unrelated# Jan 16 11:29:15 ah :D Jan 16 11:29:49 Vajb: /usr/share/applications/hildon/yappari.desktop has the commandline to yappari, I guess Jan 16 11:29:56 cat /usr/share/applications/hildon/yappari.desktop Jan 16 11:31:16 it's the file that defines your yappari desktop icon which you click to start the app Jan 16 11:31:53 anyway not needed anymore since you already found /usr/bin/Yappari Jan 16 11:31:54 yes i played with them once to get certain pdf file to start from icon Jan 16 11:32:23 you could edit the .desktop file if it turns out `nice /usr/bin/Yappari` fixes your problem Jan 16 11:33:37 ok. So basically Exec=/usr/bin/Yappari Jan 16 11:33:42 yes Jan 16 11:34:08 turns to Exec=nice /usr/bin/Yappari Jan 16 11:34:22 iirc tthat's absolutely correct Jan 16 11:34:38 hooray Jan 16 11:34:55 i'll see how yappari works now for few days Jan 16 11:35:01 :-) Jan 16 11:35:15 and maybe even inform ppl here :) Jan 16 11:35:48 im guessing nice'ing is valid till i restart program from app menu? Jan 16 11:35:54 yes Jan 16 11:36:07 it's no setting but rather a runtime option Jan 16 11:36:14 cool, thx for ur suggestions DocScrutinizer05 Jan 16 11:36:20 Vajb: new problem, or always had it? Jan 16 11:36:31 care to explalain what it does? Jan 16 11:36:39 -la* Jan 16 11:37:00 Vajb: suggestion: nice --help; man nice Jan 16 11:37:11 Sicelo: i think i've had it most of the time Jan 16 11:37:15 yappari doesnt do any busy loop that i know of Jan 16 11:37:26 input textbox is just a qt widget Jan 16 11:37:32 Vajb: xyz --help is always a good idea under linux Jan 16 11:37:38 my Yappari behaves. (just got myself banned throught stupidity recently :P ... moved to another number) Jan 16 11:37:58 although i also suffer that bug, Vajb Jan 16 11:38:06 oh. Jan 16 11:38:09 i think it got samething to do with heavy cpu use at same time Jan 16 11:38:21 something* Jan 16 11:38:21 i believe that as well Jan 16 11:38:31 which is a busy loop usually Jan 16 11:38:50 dunno if it's yappari hogging resources or some other process Jan 16 11:38:56 * Maxdamantus remembers some discussion here about yappari and busy loops. Jan 16 11:38:59 but i don't know if the loop is of Yappari's or if it's a coincidence or if it is of QT Jan 16 11:39:31 maybe qt then.. just guessing here :) Jan 16 11:39:35 if it wasn't yappari, then the same problem would show up in every kbd input Jan 16 11:39:41 i've never profiled a qt app, but i guess i can do it Jan 16 11:40:03 how many qt apps are out there? Jan 16 11:40:06 i tried that facebrick or what ever it was, but it was written in qt and also felt sluggish Jan 16 11:40:12 quite a few Jan 16 11:40:38 * ceene has just acquired his second N900 to tinker with Jan 16 11:40:40 ceene, how hard would be rewriting yappari in different toolkit? (gtk2 etc) Jan 16 11:40:44 and even qt is part of yappari, just a linked lib Jan 16 11:40:50 ceene: \m/ Jan 16 11:40:50 maybe doing libyappari ? Jan 16 11:40:52 very hard Jan 16 11:41:01 there's libwa Jan 16 11:41:08 that is a work of coderus mostly Jan 16 11:41:30 but i won't be porting yappari to anything... Jan 16 11:41:36 too much work Jan 16 11:42:21 it's not like yappari is perfect, but it isn't unusable either Jan 16 11:42:29 to throw in my two cents it seems that nice'ing helped Jan 16 11:42:33 nvm, it was about the loop inherent in event-based systems. Jan 16 11:42:33 my very own guess is: yappari has several threads and the internet communication thread blocks the whole process while waiting for or processing some network activity Jan 16 11:42:48 so... if it's (more or less) working, don't touch it Jan 16 11:42:53 i had browser opening and high cpu use and yappari got my input perfectly Jan 16 11:42:57 or searching trough the database Jan 16 11:43:24 though that would be IO bound, not CPU Jan 16 11:43:33 unless it is encrypted Jan 16 11:43:38 freemangordon: yeah, db is another candidate. Does yappari use any db? Jan 16 11:43:49 yes. may db :) Jan 16 11:43:51 there's still no crypto on yappari Jan 16 11:43:53 *many Jan 16 11:44:00 each conversation is a sqlite db Jan 16 11:44:10 freemangordon: IO wait blocks CPU too Jan 16 11:44:28 but nice wouldn;t help Jan 16 11:44:34 particularly when it's maybe swap Jan 16 11:44:35 if it is IO Jan 16 11:44:47 yes, on IO nice doesn't help Jan 16 11:44:57 but ionice would Jan 16 11:45:09 that's why I said **maybe** nice helps Jan 16 11:45:24 ceene: does yappari do some search on key entry? Jan 16 11:45:29 ceene: yes, but only on storage-IO, right? Jan 16 11:45:41 also, how many threads are there? Jan 16 11:46:00 nope, it doesn't act on textchange Jan 16 11:46:07 DocScrutinizer05: actually no, swap should be affected as well Jan 16 11:46:15 freemangordon: number of threads is easy to diagnose, unless you don't have the binry to run Jan 16 11:46:32 freemangordon: by what? Jan 16 11:46:35 DocScrutinizer05: how do you think, do I have watsapp?!? Jan 16 11:46:41 DocScrutinizer05: ischeduler Jan 16 11:46:46 *ioscheduler Jan 16 11:46:53 Isn't yappari written in Python? I'd expect basically one thread to do computations at once. Python code itself can only be running in one thread at a time. Jan 16 11:46:55 yeah, stroage/swap/whatever Jan 16 11:47:07 Maxdamantus: no, QT Jan 16 11:47:14 i can't answer about threads... none explicitly created by yappari, but i guess network connection is run by qt on a separate thread Jan 16 11:47:23 I doubt Jan 16 11:47:28 it's c++/qt Jan 16 11:47:29 Ah. Jan 16 11:47:37 ceene: htop is your friend ;-) Jan 16 11:47:49 that is why you have signals/slots in Qt Jan 16 11:48:01 freemangordon: hehe, right Jan 16 11:48:12 DocScrutinizer05: i think what you said about network could be it Jan 16 11:48:23 I'd almost bet on it Jan 16 11:48:27 ceene: my bet would be that there is sync wait somewhere Jan 16 11:48:39 if you're writing something and you receive loads of data, they will be processed Jan 16 11:48:42 wich blocks Qt from executung the event loop Jan 16 11:49:00 and cpu would go to that slot and get stolen from the UI Jan 16 11:49:09 :nod:: Jan 16 11:49:26 strace will tell Jan 16 11:49:39 ceene: but that can be fixed, just run local Qt event loop on every 1k of data (for example) Jan 16 11:49:57 you just need to somehow set a marker on strace output as soon as you notice the issue occuring Jan 16 11:50:01 that way processing won;t block the UI Jan 16 11:50:26 ceene: is source available on github or some other place besides repos? Jan 16 11:50:33 sure Jan 16 11:50:46 github.com/agamez/yappari Jan 16 11:51:17 I'd be very happy if I got one or two contributors :) Jan 16 11:51:55 ceene: not me, I don't use watsup,viber and similar shit Jan 16 11:52:16 oh, but you enjoy coding! :P Jan 16 11:52:31 i don't use yappari much myself, to be honest Jan 16 11:52:49 i don't have friends lol Jan 16 11:53:52 i've got family who aren't receptive to persuasion Jan 16 11:53:54 :( Jan 16 11:54:14 *maybe* you could run `strace yappari` to log into a file, and have a concurrent session via ssh+wlan with a `echo '#######MARKER########' >>logfile.txt` Jan 16 11:55:11 anyway use the strace options for profiling Jan 16 11:55:23 there's also gprof and others Jan 16 11:56:14 and actually you should be able to tell each single keypress from strace log as well Jan 16 11:56:41 to locate the context where the issue happened Jan 16 11:56:49 yep, and also check if something is runnig without any user event Jan 16 11:57:23 of course there will run quite a lot of stuff without _user_ event Jan 16 11:57:53 hmm, looks scary https://github.com/agamez/yappari/blob/master/src/yappari-application/Whatsapp/connection.cpp Jan 16 11:57:58 but yeah, you will see where the CPU spikes are Jan 16 11:58:04 if there's no network input nothing should happen? Jan 16 11:58:20 yes, that's quite scary if you ask me Jan 16 11:58:23 yep, except of a few keep-alive timers etc Jan 16 11:58:26 there's duped code Jan 16 11:58:36 that i've yet to refactor Jan 16 11:59:05 i even added some debug because i thought some function was never called, but it seems it is Jan 16 12:00:22 https://github.com/agamez/yappari/blob/master/src/yappari-application/Whatsapp/connection.cpp#L116 any questions, Watson? ;-P Jan 16 12:00:29 DocScrutinizer05: IIUC keep-alive, as every othe message, passes through 700 LOC function Jan 16 12:00:36 exactly Jan 16 12:01:21 ceene: you'd better try to log entryfunction time of that monster Jan 16 12:01:32 :) Jan 16 12:01:35 s/entryfunction/function Jan 16 12:02:11 * DocScrutinizer05 headdesks Jan 16 12:02:37 else if (child.getTag() == "sync") else if (child.getTag() == "participant") else if (child.getTag() == "add") Jan 16 12:02:40 etc pp Jan 16 12:03:02 child.getTag() looks like a candidate for a local var, eh? Jan 16 12:03:18 the whole thing looks like a candidate for a case Jan 16 12:03:57 at least it's now else Jan 16 12:04:09 i think when i got it, it were only if's... Jan 16 12:04:16 the compiler prolly isn't that smart to optimize child.getTag() for a single call Jan 16 12:04:52 no way IMO Jan 16 12:04:56 ceene: see http://doc.qt.io/qt-4.8/qcoreapplication.html#processEvents Jan 16 12:05:00 after all this method could deliver a different result on each new invokation Jan 16 12:05:54 ceene: this function is your friend once you found where is the UI thread blocked Jan 16 12:05:59 and iirc child is a tree of sorts, this could get really nasty Jan 16 12:06:54 yep Jan 16 12:06:59 thanks, freemangordon Jan 16 12:07:25 usually those lists have some kind of "isSubset()" stuff Jan 16 12:07:48 change that whole monster to a case construct, with a single child.getTag() Jan 16 12:08:09 dunno the english mathematical terms for operations on rows of values :) Jan 16 12:09:16 c++ supports case on qstring? Jan 16 12:10:07 c++ or Qt? Jan 16 12:10:16 aah Jan 16 12:10:36 no, but you can make a QStringList and get indexOf Jan 16 12:10:54 and then switch on that index Jan 16 12:11:17 ceene: also, while at it, please covert https://github.com/agamez/yappari/blob/master/src/yappari-application/Whatsapp/connection.cpp#L124 to: Jan 16 12:11:27 if (!in->nextTree(node)) Jan 16 12:11:33 return false; Jan 16 12:12:03 that way you'll ident the whole case left, making the code way more readable Jan 16 12:12:45 18 times Child.getTag. And a real master instance of WAAAH: else if (child.getTag() == "media" || child.getTag() == "duplicate") Jan 16 12:13:00 lol Jan 16 12:13:37 well, at least a QString childTag seems like something useful Jan 16 12:13:52 hehe, code slaughter day? i mean review Jan 16 12:14:16 DocScrutinizer05: relax, it is not thaaaat bad, you should see what happens in adeclock :p Jan 16 12:14:25 passing integers through QString ;) Jan 16 12:14:28 hehe Jan 16 12:14:30 http://commadot.com/wp-content/uploads/2009/02/wtf.png Jan 16 12:14:41 the best review is the one that removes the most code and keeps functionality Jan 16 12:15:52 like that " Showing with 0 additions and 427 deletions." Jan 16 12:15:55 ? Jan 16 12:16:03 https://github.com/community-ssu/clock-ui/commit/f7a59c2ea09255b09c742796a2690f902dca8395 Jan 16 12:16:23 unfortunately I still can't pull myself together and find time to finish that Jan 16 12:16:46 rewrite from scratch? Jan 16 12:17:16 no, just rewrite it the way it should've been written from the start Jan 16 12:17:22 ooh, isn't adeclock what had been pushed to replace the stock maemo clock without any real need? Jan 16 12:17:30 * freemangordon hides Jan 16 12:17:36 yes, the same Jan 16 12:18:51 DocScrutinizer05: if you need technical reasons why it should not have been included, look at the last ~10 commits there :) Jan 16 12:18:56 cya folks, have fun! Jan 16 12:19:02 bye Jan 16 12:28:31 you say no real need, but it has extra functionality. Jan 16 12:28:53 my battery is funny Jan 16 12:29:07 mine is uncalibrated Jan 16 12:29:27 Last Measured Discharge: 1410 mAh Jan 16 12:29:32 Charge:0 NOACT:0 IMIN:0 CI:0 CALIP:0 VDQ:1 EDV1:0 EDVF: 0 Jan 16 12:29:35 ILMD=1869 EDVF=3000 EDV1=3248 ISLC=18 Jan 16 12:30:14 i can't get mine calibrated Jan 16 12:30:23 sooner or later vdq turns itself to 0 Jan 16 12:30:36 i don't know why Jan 16 12:31:26 maybe pali knows what to do Jan 16 12:33:18 Cycle Count since Learning: 14 Total Cycle Count since last full reset: 24 Jan 16 12:33:44 ceene, try the maxd suggestion, start some app, music player for example Jan 16 12:34:20 sixwheeledbeast: I was not talking about usability, but for quality Jan 16 12:34:23 fully charge and fully discharce with a heavy load? Jan 16 12:34:49 ceene, medium load Jan 16 12:35:02 maybe just enable wifi Jan 16 12:35:17 but that's basically what i'm doing now Jan 16 12:35:19 normal use Jan 16 12:35:44 freemangordon: I don't disagree, but the extra bits are nice Jan 16 12:36:40 "without any really need" Jan 16 12:38:43 ~flashing Jan 16 12:38:43 from memory, maemo-flashing is http://wiki.maemo.org/Updating_the_tablet_firmware, or - on linux PC - download&extract http://maemo.cloud-7.de/maemo5/patches_n_tools/maemo-my-private-workdir.tgz, cd into it, do sudo ./flash-it-all.sh Jan 16 12:40:23 also it was a closed package before anyway IIRC Jan 16 12:45:40 sixwheeledbeast: sure, all this is correct, it is just the quality of the code is... Jan 16 12:46:00 ...pants Jan 16 12:48:09 ceene: you seen what I suggested? Jan 16 12:48:21 re bq27200 calib Jan 16 12:48:36 i think i missed that Jan 16 12:49:29 when was that? Jan 16 12:49:32 i can serch it on the log Jan 16 12:49:56 ceene: just leave the flash LEDs on for a few hours. Jan 16 12:50:09 ceene: that will discharge it at a reasonably stable rate. Jan 16 12:50:19 on airplane mode? Jan 16 12:50:34 Wouldn't bother with airplane mode. Jan 16 12:52:09 airplane mode? hmm? Jan 16 12:52:25 ~bq-calibrate Jan 16 12:52:25 [bq-calibrate] http://maemo.cloud-7.de/maemo5/patches_n_tools/calibrate-bq27k.sh, or http://talk.maemo.org/showthread.php?p=1247595 Jan 16 12:53:04 I doubt those scripts are going to help. Jan 16 12:53:48 I'm pretty sure it's not going to calibrate if there's to great a jump between the calculated discharges. Jan 16 12:53:55 i'm reading it... what's "fastcharger"? Jan 16 12:55:05 as evidenced by my log where NAC sits on the magic number, LMD/16 (mentioned in the cryptic text about learning cycle disqualification) for two and a half hours before VDQ is set to 0 at the same time as NAC continues decreasing. Jan 16 12:55:20 s/to great/too great/ Jan 16 12:55:50 the thing is, mine is a new polarcell, so it's gonna be almost twice as big as the old one Jan 16 12:56:03 Yes. That's the situation I was in. Jan 16 12:56:32 The LMD was 1088 or 1089 mAh, while the Polarcell ended up being over 1590 mAh. Jan 16 12:56:36 and you did only that, charge fully and let it discharce with the flash on? Jan 16 12:57:22 I didn't discharge it all the way with the flash on. I left it on overnight. Jan 16 12:57:33 but i've done that with no success Jan 16 12:57:34 (so it would've been on for the last few hours) Jan 16 12:58:00 i mean, i've charged it, used it for a while, left it during the night, although with leds off Jan 16 12:58:05 shouldn't that work? Jan 16 12:58:06 You should probably also log the output of something like bq27200.sh to see where it resets. Jan 16 12:58:11 or it needs a heavier load? Jan 16 12:58:20 No. It needs a load. Jan 16 12:58:24 ok Jan 16 12:58:28 so i can use it for today Jan 16 12:58:32 check if vdq=1 Jan 16 12:58:37 and light on leds or whatever Jan 16 12:58:40 and go to sleep Jan 16 12:58:45 Yes. Jan 16 12:58:58 a loop of bq27200.sh to check the log tomorrow won't hurt Jan 16 12:59:09 Preferably while it still reports 10% or so CSOC. Jan 16 12:59:12 (at least) Jan 16 12:59:57 ok Jan 16 13:00:04 i'll try that today, thanks! Jan 16 13:01:22 The point of that is to cause it to remember *something* closer to the actual value, so it doesn't sit at 6% (100/16)% for a few hours before giving up. Jan 16 13:01:58 (batteries usually become less efficient when you're drawing more current from them—I guess bq27200 just doesn't know how to compensate for that) Jan 16 13:04:10 ceene: fastcharger is a nokia wallwart Jan 16 13:05:37 ((almost twice as big)) then you need to run several calib learning cycles, until LMD doesn't adjust significantly anymore Jan 16 13:05:58 LMD adjusts only 1/8 per calib, iirc Jan 16 13:06:21 or was it 1/6. can't recall Jan 16 13:07:11 ~100*1.125*1.125*1.125*1.125 Jan 16 13:07:11 160.1806640625 Jan 16 13:07:24 ~100*1.125*1.125*1.125*1.125*1.125*1.125 Jan 16 13:07:24 202.728652954102 Jan 16 13:07:52 seems you'll need 6 calib cycles to adjust from a relative 100% LMD to a 200% Jan 16 13:09:14 ((charge fully and let it discharce with the flash on?)) please read ~bq-calibrate Jan 16 13:09:34 ceene: my solution was to discharge as fast as possible to get it calibrated Jan 16 13:09:48 Vajb: too fast is also wrong Jan 16 13:09:55 i was streaming radio from internet till it shutdown Jan 16 13:09:57 DocScrutinizer05: can you explain how ~bq-calibrate works around the behaviour I observed? Jan 16 13:10:00 looks like there's no right way Jan 16 13:10:02 lol Jan 16 13:10:07 it works but you get massively incorrect calibration Jan 16 13:10:19 my problem was also that too early shutdown Jan 16 13:10:28 but i changed from stock to mugen Jan 16 13:10:52 for first calibration jump it needed fast discharge Jan 16 13:10:53 Vajb: that's why my script shuts down bme prior to calibration Jan 16 13:11:10 others were more gentle discharges Jan 16 13:11:11 in 'useless' news .. am more & more happy with fennec. definitely not fast, but useable Jan 16 13:11:24 Sicelo: did you try midori on debian's chroot? Jan 16 13:11:50 Vajb: aah yes, that works Jan 16 13:12:12 yup can confirm it worked :) Jan 16 13:12:23 i still will try. had problems with my chroot. old squeeze on works, but the wheezy gets corrupted when i install openoffice. meant to ask in here what i could do about that Jan 16 13:12:35 i just don't install openoffice Jan 16 13:12:39