**** BEGIN LOGGING AT Tue Dec 22 02:59:56 2009 Dec 22 07:02:38 g'morning Dec 22 07:04:42 prahal: probably not Dec 22 07:05:16 yay, openttd again on my phone :D Dec 22 07:35:04 Maybe not.. Dec 22 07:45:09 blindcoder: what's the interface for that? Dec 22 07:46:26 moin Dec 22 07:46:34 tmzt: interface? Dec 22 07:47:11 the hardware interface of openttd Dec 22 07:47:19 tmzt: mouse Dec 22 07:47:27 or what do you mean? Dec 22 07:47:30 * blindcoder is confused Dec 22 07:47:46 me too Dec 22 07:47:51 what is this program? Dec 22 07:47:58 open transport tycoon deluxe Dec 22 07:48:36 ah, ok Dec 22 07:48:53 very cool game Dec 22 07:49:01 rr2d Dec 22 07:49:13 on the suse discs years ago Dec 22 07:51:50 yeah, too bad that 0.7.4 segfaults with a stack corruption and 0.6.3 doesn't compile (const char* to char* conversion) Dec 22 07:51:54 and I'm too lazy to fix ;) Dec 22 07:53:07 what compiler? Dec 22 07:53:18 shr-u build system Dec 22 07:53:19 just use an older one if you can Dec 22 07:57:07 ste: tell harold his ts0710 driver might be used on motorola sholes/milestone Dec 22 08:25:12 blindcoder: when does openttd segfault? Dec 22 08:25:58 blindcoder: do you have installed the graphics and sounds too? Dec 22 08:40:53 Heinervdm: yeah, all installed, segfaults during startup, gdb says stack is corrupted Dec 22 08:42:08 hmm Dec 22 08:42:42 Swap is enabled Dec 22 09:24:07 freesmartphone.org: 03morphis 07msmcomm * r5e0204e1285e 10/msmcommd/src/msmcommd.c: msmcommd: adjust schedule delay for read/write timer Dec 22 09:24:08 freesmartphone.org: 03morphis 07msmcomm * re2bc3241f3de 10/msmcommd/src/msmcommd.c: msmcommd: change serial configuration and add different colors for log output Dec 22 09:24:09 freesmartphone.org: 03morphis 07msmcomm * rf2a58d78b1dd 10/msmcommd/src/msmcommd.c: msmcommd: Change datatype of data field from char to unsigned char Dec 22 09:24:10 freesmartphone.org: 03morphis 07msmcomm * re5d34405bbc1 10/msmcommd/src/msmcommd.c: Merge branch 'master' into remote-master Dec 22 09:38:05 Hi all... what is the best way to run a simple command on SHR boot? I tried /etc/rc.local or /etc/boot.local but it doesn't seem to exist :( Is there an eqivalent file in SHR/Openembedd? Dec 22 09:40:17 ~logs Dec 22 09:40:20 All conversations are logged to http://ibot.rikers.org/channel, where "channel" is replaced by the URL-encoded channel name, such as %23freenode for #freenode. Lines starting with spaces are not logged. Dec 22 10:47:07 freesmartphone.org: 03morphis 07msmcomm * ra06262f8c6a7 10/msmterm/ (10 files in 2 dirs): msmterm: add basic skeleton Dec 22 10:47:08 freesmartphone.org: 03morphis 07msmcomm * r8697eafcf817 10/msmcommd/src/msmcommd.c: msmcommd: add initial work for network control interface Dec 22 10:47:09 freesmartphone.org: 03morphis 07msmcomm * r6b697ed4f5ba 10/msmcommd/.gitignore: msmcommd: add gitignore file Dec 22 11:01:55 hi someojne know how to #define a variable with the value return by a system call Dec 22 11:02:20 like# define APP_REVISION system("svn info |grep Revision") (but this one does not work like this) Dec 22 11:08:32 GarthPS: i think you have to do that with some autotools magic Dec 22 11:09:05 eu.. for the moment I use qmake for this project Dec 22 11:09:55 Heinervdm: and for now it does not work if i do something like this in my .pro file REVISION = $$system(git rev-parse HEAD) Dec 22 11:09:55 DEFINES += APP_REVISION=$$REVISION Dec 22 11:10:15 replaceed by a svn Dec 22 11:10:49 another way to do it, is to let svn do it Dec 22 11:15:14 GarthPS: http://www.compuphase.com/svnrev.htm Dec 22 11:16:08 Heinervdm: ouhh sweet will look at this Dec 22 12:10:46 mrmoku, suggestion for contactlist (and prolly callscreen): i did not try it and i am not athome so i can't: use the pictuer widget to embed the photos, because it is squared and doesnt use transparency (at least not in gry*) so scrolling might be smoother Dec 22 12:11:59 JesusMcCloud: will try, thanks (makes sense) Dec 22 12:13:33 mrmoku, or if you want to make sure that no transparency is used, simply pust some solid frame with solid backgroudn underneath it (that i think i can try on fr, as ssh client is broke on the box i am currently using) Dec 22 12:13:52 where is the edj located for phoneui? Dec 22 12:14:01 /usr/share/libphone-ui-shr/default.edj Dec 22 12:14:20 k will try Dec 22 12:16:16 JesusMcCloud: I'm already trying the image instead of icon thing Dec 22 12:16:32 as it is _very_ easy to change Dec 22 12:16:55 ah ok... let me know if it wokred as i hoped Dec 22 12:17:03 sure Dec 22 12:17:21 (and send me edj if it worked) Dec 22 12:18:15 (hmm did i customize image widget in gry... i cant recall) Dec 22 12:18:26 JesusMcCloud: change is not in edj... as the phonto get's added by the genlist icon callback Dec 22 12:18:35 ah ok Dec 22 12:19:25 hmm... difficult to tell the effect Dec 22 12:19:31 might be a bit smoother... dunno Dec 22 12:19:41 in any way it looks sane to me Dec 22 12:19:45 so I will just commit it Dec 22 12:20:10 but if i understand correctly only contacts without customized photo will have the default-without-transparency icon? Dec 22 12:22:36 (waaaahgr ssh is segfaulting on this stupid box) Dec 22 12:22:43 SHR: 03mok 07libphone-ui-shr * r63d5d35b5375 10/src/view/dialer-view.c: dialer: fix a typo in an error message Dec 22 12:22:54 SHR: 03mok 07libphone-ui-shr * r7ecb1d09e439 10/src/view/contact-list-common.c: contact-list: use image instead of icon for the photo Dec 22 12:22:54 SHR: 03mok 07libphone-ui-shr * r8640b51a7e62 10/src/util/ (common-utils.c common-utils.h): common-utils: add new_gvalue_pointer helper Dec 22 12:26:03 mrmoku, but if i understand correctly only contacts without customized photo will have the default-without-transparency icon? Dec 22 12:48:37 JesusMcCloud: no, it is the same for both Dec 22 12:48:54 both get added in the icon callback... just different files get set Dec 22 12:48:56 mrmoku: xorg-7.5 pushed to oe.dev.. lets see what happens Dec 22 12:49:03 ohh :) Dec 22 12:49:40 JaMa: rebuilding an image after your 'for in clean eglibc' trick Dec 22 12:49:54 looks good so far Dec 22 13:20:57 freesmartphone.org: 03morphis 07msmcomm * r471820b26bde 10/msmcommd/ (15 files in 3 dirs): msmcommd: refactor the code a little bit Dec 22 13:33:11 hi guys Dec 22 13:33:31 guess what ? http://blog.hackable1.org/2009/12/hackable1-rev5-is-out.html Dec 22 13:51:43 Weiss: ping Dec 22 13:52:15 pong Dec 22 13:52:30 Weiss: any update for gdrm-2.6.31 branch? Dec 22 13:52:56 Weiss: your patches apply cleanly to 2.6.32 larsc pushed last night.. but compilation failed later Dec 22 13:53:19 Weiss: so I've added linux-openmoko-2.6.32 without gdrm patches now.. Dec 22 13:53:27 ah.. I haven't tried that yet.. I'll give it a go (probably tomorrow) Dec 22 13:53:35 do you have the compilation failure message handy?> Dec 22 13:53:51 Weiss: mmt I can rebuild in sec.. Dec 22 13:54:26 with 2.6.31 I think it basically works, but with a few gremlins lurking and still the same WSoD for people (I'm trying to rework the regulator stuff in line with DocScrutinizer's suggestions) Dec 22 13:54:48 I've had Work Doom and now Christmas Doom preventing me from doing much exciting (see my blog :) Dec 22 13:56:28 Weiss: yes, I've read blog already, but it looked like US holidays, not Doom :) Dec 22 13:57:03 hehe Dec 22 13:57:05 Weiss: and yes.. too much gremlins here in Europe :) Dec 22 13:57:07 "holiday" Dec 22 13:57:20 I didn't see daylight for a whole week.. Dec 22 13:57:27 apart from to take that photo Dec 22 13:57:54 Weiss: I guess you worked there.. but you wrote about photo oportunities.. so it was more like about holidays.. (a bit geeky but still.. :)) Dec 22 13:58:31 :) Dec 22 13:58:58 it was certainly an experience Dec 22 14:00:39 Weiss: do I need "S3C2410 LCD framebuffer support"? I didn't notice it in 2.6.31.. Dec 22 14:01:01 not at all... there's no LCD plugged into those pins (or however it works) Dec 22 14:01:05 JaMa: no Dec 22 14:01:28 larsc's answer is more concise as ever :) Dec 22 14:02:08 and whole FB support? Dec 22 14:02:32 larsc: btw gta* configs were dropped intentionaly or have I overlooked them? Dec 22 14:06:16 JaMa: you'll needfor gta01 Dec 22 14:09:45 Weiss: http://pastebin.ca/1723708 Dec 22 14:09:50 DocScrutinizer51: ah, thanks Dec 22 14:32:09 hi all, it's been a while, is shr-unstable usable these days? and is testing being updated regularly? Dec 22 14:34:02 daniele_nothome: have you found sth? Dec 22 14:34:15 JaMa, about freetype-native? Dec 22 14:38:25 Zorkman, updated regularly = about every 4 weeks Dec 22 14:38:50 otherwise i could just call it -unstable as well :-) Dec 22 14:39:08 JaMa: they were droped unintentionally Dec 22 15:16:06 Weiss: with this simple patch http://pastebin.ca/1723773 I can compile it but then linking failed with http://pastebin.ca/1723774 Dec 22 15:18:10 interesting.. what's that about hid-quirks.c? Dec 22 15:20:03 does drm_crtc_helper_set_config still exist if you grep the source? Dec 22 15:20:16 Weiss: seems like bug in finding source for built-in.o to me Dec 22 15:20:57 ./drivers/gpu/drm/drm_crtc_helper.c:int drm_crtc_helper_set_config(struct drm_mode_set *set) Dec 22 15:21:46 there isn't some new config option for enabling the DRM helpers? Dec 22 15:21:59 mmt I'll check Dec 22 15:23:44 might also be a problem with the makefile in drivers/mfd/glamo.. would be good to compare against the one in drivers/gpu/drm/i915 Dec 22 15:24:47 in fact, that sounds likely Dec 22 15:26:07 Weiss: DRM_KMS_HELPER is missing, I'll check why Dec 22 15:26:18 a new option? Dec 22 15:27:56 seems so.. but not shown in menuconfig (even when its depending only on selected DRM Dec 22 15:29:06 interesting Dec 22 15:29:15 sounds like Glamo DRM+KMS should depend on that Dec 22 15:29:40 http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=13a8195b148615b15a4f4385f695f2a232095414 Dec 22 15:30:46 ah there is no "config name" so it cannot be selected by *config and drivers need to select it itself right? Dec 22 15:31:01 ah, maybe Dec 22 15:31:05 I didn't know you could do that Dec 22 15:44:07 ? Dec 22 15:44:09 ? Dec 22 15:46:11 Guest23181: yes Dec 22 15:46:48 Guest23181: maybe Dec 22 15:47:02 Weiss: with those 2 patches compiled fine.. I'll test it at home (it boots, but I'm not sure even if display is on) Dec 22 15:47:07 dunno. Can you repeat the question Dec 22 15:47:29 Weiss: or display on wsod :) Dec 22 15:48:25 could anyone teach me how to use this shit?! Dec 22 15:48:27 http://jama.homelinux.org/org.openembedded.shr.images/om-gta02/ 2.6.32*-r2 is with gdrm Dec 22 15:48:39 quite new to this program Dec 22 15:48:41 LOL Dec 22 15:48:42 Guest23181: ah then no Dec 22 15:48:43 is there some library for talking the UBX protocol used by antaris gps chip? Dec 22 15:49:08 lindi-: afaik not really Dec 22 15:49:25 mickey wanted one though, I think :) Dec 22 15:49:28 ELOW Dec 22 15:49:59 spaetz: ok. gpsd seems to have basic support. ogpsd has a different set of supported stuff but is not a library Dec 22 15:50:11 spaetz: alphaone wrote some libubx stuff but it's very incomplete Dec 22 15:50:38 spaetz: then there's ubxgen.py that just does minimal checksum generation and notihng more Dec 22 15:52:52 alphaone|gone: seems you wrote also the ogpsd stuff Dec 22 15:54:20 alphaone|gone: i'd like to get FixNow(tm) work as a normal user (ogpsd wants to poke /sys and wants to register org.freesmartphone.frameworkd dbus name) Dec 22 15:55:56 lindi-: where's the problem? Dec 22 15:56:39 DocScrutinizer51: org.freesmartphone.frameworkd is already registered by ogsmd Dec 22 15:57:59 errr Dec 22 15:58:09 Maybe I'm just being paranoid, but it almost looks like the hackable:1 guys are reinventing many wheels :P Dec 22 15:58:31 TAsn: what have they reinvented now? Dec 22 15:58:38 h1settings Dec 22 15:58:41 don't they use fso? Dec 22 15:58:49 TAsn: that shouldn't gice you paranoia Dec 22 15:58:56 TAsn: not sure Dec 22 15:58:59 I mean, if so, it's practically almost everything dos did. Dec 22 15:59:10 and they also chose not to use OE Dec 22 15:59:27 but they use some other building env Dec 22 15:59:30 cool Dec 22 15:59:34 and that's just what I saw at first glance. Dec 22 16:00:03 OE is for toasters anyway Dec 22 16:00:08 TAsn: they cross compile some packages and pull some packages from debian afaik Dec 22 16:00:23 lindi-, yes. Dec 22 16:00:28 TAsn: (which is very unnice for the user that does not know how to build a package) Dec 22 16:00:31 by definition of "E" Dec 22 16:00:55 TAsn: looking at the TODO for rev6, no... they want to switch to FSO with next release :P Dec 22 16:01:06 Again, there is no longer a need to spend 24+ hours until your cross platform build system is up. Dec 22 16:01:07 JaMa|Away: cool :) Dec 22 16:01:19 it almost looks like they are talking about OE :P Dec 22 16:01:37 mrmoku, link please. Dec 22 16:01:43 Ainulindale, !~!! here? :P Dec 22 16:02:44 TAsn: http://blog.hackable1.org/2009/12/hackable1-rev5-is-out.html Dec 22 16:02:51 ok Dec 22 16:02:54 found it in the roadmap Dec 22 16:03:03 rev6: "First version based on the FSO telephony software stack " Dec 22 16:04:05 TAsn: what is the new ui-utils way for the frame thing? Dec 22 16:04:13 frame? Dec 22 16:04:33 with the old crap we had views which could consist of different framew Dec 22 16:04:36 +s Dec 22 16:04:50 like for contacts right now its show and edit Dec 22 16:05:05 oh, frames are just windows Dec 22 16:05:08 different screens within one window Dec 22 16:05:09 no need for special caring Dec 22 16:05:17 they are not within one window Dec 22 16:05:29 I removed the frame stuff as it looked just redundant Dec 22 16:05:37 different windows are different windows. Dec 22 16:05:51 frames of one view shared the window... that was the sense of frames Dec 22 16:06:10 sec, let me take a look. Dec 22 16:08:04 mrmoku, just took another look Dec 22 16:08:06 and Dec 22 16:08:13 frame_show and view_show are identical... Dec 22 16:08:34 (almost 100% but both provide the same functionality) Dec 22 16:08:44 so either we define what we want frames to do Dec 22 16:08:48 or we just use views Dec 22 16:10:07 well... reuse an existing window would be my definition Dec 22 16:10:23 so I could set the show_cb of the view to something different Dec 22 16:10:39 show_cb changed definition btw Dec 22 16:10:48 iirc in window.c Dec 22 16:10:55 it was used Dec 22 16:11:17 in a weird way Dec 22 16:11:29 now it just gets called when a window is shown Dec 22 16:12:15 it did the creation of the widgets Dec 22 16:12:32 mrmoku, that show, and is being done in init now Dec 22 16:12:36 show = should Dec 22 16:12:42 I just don't get what you mean by reusing an existing window Dec 22 16:13:12 don't add a new one with elm_window_add... and use the existing one instead... show something different in there Dec 22 16:13:17 could do that with pager Dec 22 16:14:04 TAsn: in the contacts show view you tap on a field and the 'edit the field' screen is shown in the same window Dec 22 16:14:15 mrmoku, just us Dec 22 16:14:19 ui_utils_view_show Dec 22 16:14:27 and set whatever you want Dec 22 16:14:33 it's already initialized Dec 22 16:14:47 therefore already exists. Dec 22 16:14:54 ? Dec 22 16:15:02 I want to set a different layout... different widgets... Dec 22 16:15:09 which is init Dec 22 16:16:00 so what are you trying to "save" Dec 22 16:16:04 just elm_window_add? Dec 22 16:16:15 more the additional window in the task list Dec 22 16:16:26 mrmoku, just destroy the previous then Dec 22 16:16:32 creating the window is not such a big deal Dec 22 16:16:41 or use a pager Dec 22 16:16:43 especially not when you already clear *everything* Dec 22 16:16:52 probably the correct elm way to do it Dec 22 16:16:52 or use a pager, yeah. Dec 22 16:17:03 seems alphaone|gone's code has been forked at least https://open.grcnz.com/svn/albatross/branches/groundstation/groundstation/gs/gps/ Dec 22 16:17:07 mrmoku, depends Dec 22 16:17:10 what are you trying to do? Dec 22 16:17:21 in contacts, showing a new window, might make sense Dec 22 16:17:32 but mostly, pager is the way to go. Dec 22 16:17:37 TAsn: take new message view then... Dec 22 16:17:53 explain Dec 22 16:18:00 first screen enter the message Dec 22 16:18:06 second screen select recipients Dec 22 16:18:09 that's classic pager Dec 22 16:18:11 yup Dec 22 16:18:42 the way it's done atm is crazy :P Dec 22 16:18:53 TAsn: thing is I'm trying to debug the 'add new contact from phonelog' bug Dec 22 16:18:57 oh Dec 22 16:19:00 saw my comments? Dec 22 16:19:02 and considdering to rewrite the crap Dec 22 16:19:03 already debugged most of it. Dec 22 16:19:14 the bug is obvious Dec 22 16:19:17 I wrote it in the bug report Dec 22 16:19:23 it's because you add a number Dec 22 16:19:24 number? Dec 22 16:19:28 listen Dec 22 16:19:30 you add a number Dec 22 16:19:30 (of bug) Dec 22 16:19:34 oh Dec 22 16:19:35 819 Dec 22 16:19:35 iirc Dec 22 16:19:38 read there Dec 22 16:19:49 anyhow, the issue is, that you add a field Dec 22 16:20:00 (phone number from phonelog) Dec 22 16:20:11 and then when you want to add another field Dec 22 16:20:49 you see there's already one field (phone number) Dec 22 16:20:56 so you think you are updating an existing contact Dec 22 16:21:00 instead of creating a new one Dec 22 16:21:02 and trying to pass Dec 22 16:21:05 no Dec 22 16:21:06 NULL path Dec 22 16:21:07 :P Dec 22 16:21:19 it happens after you add the Name Dec 22 16:21:19 field Dec 22 16:21:21 it determines via path if it should do an update or add a new one Dec 22 16:21:27 if path is null... add a new one Dec 22 16:21:30 because of exactly this. Dec 22 16:21:38 mrmoku, I'm talking about before Dec 22 16:21:43 I can't remember exactly where it was, sec. Dec 22 16:21:45 TAsn: click add field for a contact... and click back Dec 22 16:21:48 it will segfault :| Dec 22 16:22:11 #1 0xb7fcfa1e in frame_show_show (_data=0x81b20d8) Dec 22 16:22:11 at view/contact-show-view.c:262 Dec 22 16:22:17 exactly Dec 22 16:22:20 what I did Dec 22 16:22:22 here's the bt Dec 22 16:22:24 ^ Dec 22 16:23:13 btw, you are right, I first thought it's what I just said Dec 22 16:23:16 and then I found something else Dec 22 16:23:18 sec. Dec 22 16:23:23 I already gdbed this crap :P Dec 22 16:23:52 well... have to run and pickup my son Dec 22 16:23:54 mrmoku, anyhow, don't rewrite yet Dec 22 16:24:07 I think you should just let this bug be until we rewrite it Dec 22 16:24:10 (hopefully this weekend) Dec 22 16:24:20 this weekend I'm gone ;) Dec 22 16:24:37 will be back on 2nd then Dec 22 16:24:38 going to itally Dec 22 16:24:42 anyway... bbl Dec 22 16:25:04 hmmm Dec 22 16:25:30 going to italy. away... bbl looks strange Dec 22 16:57:20 I just pushed shr-testing upgrade. Please read announcement before upgrading Dec 22 16:57:44 will be back later... Dec 22 17:23:17 hehe, shit, I broke e :P (upstream) Dec 22 17:24:16 go fix it, then? Dec 22 17:24:19 ;) Dec 22 17:25:01 someone already discussed about it in ML, that's how I know. Dec 22 17:25:10 I can fix it, though we were discussing possible solutions Dec 22 17:25:20 I think my suggestion is probably the most suitable... Dec 22 17:25:48 I'll wait and see. Dec 22 17:27:07 And it's actually a *very* annoying bug, it's funny no one noticed it up until now Dec 22 17:27:10 it affects all of e Dec 22 17:27:15 and it's always visible :P Dec 22 17:32:22 TAsn: what does it do? Dec 22 17:32:30 hm.. Dec 22 17:32:51 you can insert characters at the end Dec 22 17:32:58 of text fields Dec 22 17:33:17 the function that finds the correct position Dec 22 17:33:54 returns the position of the *last char* up to coordinates, and not the last position in string (string index) up to coordinates. Dec 22 17:34:20 spaetzN800: wouldn't be safer to use "opkg remove shr-today; killall -9 shr-today;" before update? Dec 22 17:34:31 :P Dec 22 17:34:40 JaMa|Away, no Dec 22 17:34:40 spaetzN800: at least before upgrading over ssh Dec 22 17:34:45 killall shr-today Dec 22 17:34:47 won't work Dec 22 17:34:48 TAsn: no such process? Dec 22 17:34:51 as it's a python app :P Dec 22 17:35:02 ahh.. :) Dec 22 17:35:05 yeah. Dec 22 17:35:11 kill all snakes! Dec 22 17:35:51 hehe Dec 22 17:36:25 then just kill right pid as its still running even after opkg remove right? Dec 22 17:42:26 yes. Dec 22 17:45:28 jej, calibration node in sysfs moved again in 2.6.32.. Dec 22 17:49:19 AAARRG Dec 22 17:49:36 * mwester shakes his head sadly over the ongoing saga Dec 22 17:51:07 * mwester invents the HAPI layer - the Heuristic API, which dynamically searches and guesses at the kernel API on behalf of user-space apps :D Dec 22 17:51:29 kewl Dec 22 17:52:01 * DocScrutinizer51 invents the kernel devels whip Dec 22 17:52:50 :P Dec 22 17:53:07 I think it's nicer this way. Dec 22 17:53:25 it helps us find out what apps are are not developed anymore Dec 22 17:53:29 *which Dec 22 17:53:34 and which are still active. Dec 22 17:54:21 especially with the calib node Dec 22 17:54:26 just need to wait a week (for the next API change) Dec 22 17:54:54 and anything that doesn't work is proabably not as active as it should. Dec 22 17:54:58 probably* Dec 22 17:55:45 anyhow, ciao. Dec 22 18:04:35 /sys/devices/platform/s3c2440-i2c/i2c-0/0-0073/s3c2410-ts/calibration/ instead of Dec 22 18:04:44 /sys/devices/platform/s3c2440-i2c/i2c-adapter/i2c-0/0-0073/s3c2410-ts/calibration Dec 22 18:09:38 oh well. At least it's shorter Dec 22 18:10:06 hmm, what should use that? Dec 22 18:10:12 i thought tslib did calibration in userland Dec 22 18:11:44 I thought calibration in userland is identified as nonsense Dec 22 18:14:35 DocScrutinizer: huh, why ? Dec 22 18:16:22 DocScrutinizer51: who identifies it as nonsense? Dec 22 18:17:17 wpwrak: imagine you need to temporarily disable LCD-BL converter to avoid interference with r-ts. Or any other arbitrary low level hw foo. maybe you calibrate via PWM for the VDD over the r-planes. whatever Dec 22 18:18:33 what would you do if ts is so badly decalibrated it exceeds range of the AD converters Dec 22 18:19:18 userland has simply no way to cope with calibration in a sane way Dec 22 18:19:55 lindi-: its used in shr in /etc/X11/Xsession.d/89xTs_Calibrate, then evdev driver is calibrated Dec 22 18:20:31 DocScrutinizer: if you exceed the range, you've lost anyway Dec 22 18:21:13 and please don't mix up hw calibration with conversation of metrics or user convenience things like acceleration Dec 22 18:21:29 wpwrak: says who? Dec 22 18:21:54 DocScrutinizer: if your ts gives you lots of noise unless you do some special tricks, then that same noise will exist during normal operation, unless you integrate these tricks into the measurement process -> calibration is in the same position as normal measurements Dec 22 18:22:42 err. you *totally* lost me Dec 22 18:22:56 DocScrutinizer: if your taps yield results outside of the measurement range, a part of your screen is effectively inaccessible. that wouldn't be good news, would it ? Dec 22 18:23:18 DocScrutinizer: noise = supposed need to turn off BL Dec 22 18:23:34 that's exactly why userland can't fix that Dec 22 18:24:19 (noise) doesn't help me to understand your point Dec 22 18:24:48 DocScrutinizer: (out of range) yes, but in what way would the kernel be in a better position ? Dec 22 18:25:07 the kernel knows the hw Dec 22 18:25:12 DocScrutinizer: (noise) you claim that you have to turn off the BL in order to get accurate readings from the ts, right ? Dec 22 18:25:44 I say it might be a scenario we may find Dec 22 18:27:07 and as tslib also is intended to do jitter filtering that's another point where you will fail in userland Dec 22 18:27:49 DocScrutinizer: well, whether the kernel knows the hardware or not, the adc won't magically increase its range :) _if_ the hardware is able of range adjustments (e.g., though a preamp or by switching the reference voltage), and you need dynamic adjustments for calibration, you'll need the same feature for normal operation too, so again calibration is not a special case Dec 22 18:28:11 or do you think userland should be able to power down BL, trigger a DA conversion' and then retsart BL' in less than one millisecond? Dec 22 18:29:16 what *is* normal operation??? o.O Dec 22 18:29:31 DocScrutinizer: one ms is pretty long :) but my argument is a different one: if you need to create a special setup in order to obtain a good acquisition, that does not only apply to calibration but to any ts sampling Dec 22 18:29:57 so what? Dec 22 18:29:59 DocScrutinizer: thus, whatever you do to read the ADC, this function needs to implement that setup. Dec 22 18:30:13 DocScrutinizer: therefore, calibration is not a special case Dec 22 18:30:18 err. so what? Dec 22 18:30:51 the "calibration" is a *process* in kernel Dec 22 18:31:06 that calibrates each single DA Dec 22 18:32:14 DocScrutinizer: hmm, now i'm confused. what exactly do you calibrate ? the adc ? Dec 22 18:32:21 it's basically a multiplier applied to the raw DA reading Dec 22 18:32:59 DocScrutinizer: "ts calibration" normally refers to matching ts "coordinates" with screen coordinates Dec 22 18:33:07 or make that a function N-th order Dec 22 18:33:08 DocScrutinizer: are you talking about something else then ? Dec 22 18:33:35 no, thats conversation of metrics Dec 22 18:33:52 and should be fixed Dec 22 18:33:55 DocScrutinizer: (n-th order) so you mean things like adc nonlinearity ? Dec 22 18:34:03 for all devices of same type Dec 22 18:34:29 yes. things like nonlinearity etc Dec 22 18:35:16 it could depend on temperature. on batvoltage. on dunno what Dec 22 18:35:26 you cant do that in userland Dec 22 18:35:56 metrics conversation otoh is a nobrainer Dec 22 18:36:22 hmm, but even if your adc needs frequent calibration. why would you not perform this as part of the regular acquisition ? Dec 22 18:36:48 see above Dec 22 18:37:14 in the adc driver, i mean. if it's adc inaccuracies, it has nothing to do with the ts Dec 22 18:38:19 sorry. I'm somewhat getting tired of trying to explain the issue Dec 22 18:38:35 I made all my points Dec 22 18:38:59