libera/#maemo-leste/ Thursday, 2021-12-30

Wizzupuvos: building -devel droid3 image atm, it won't be rsynced yet00:52
Wizzuphttps://phoenix.maemo.org/job/leste-image-droid3-dev/1/console00:52
Wizzupwill go to sleep and see if it succeed tomorrow, will add rsync and some -dev suffix then00:52
Wizzupfreemangordon: wrt subscribe, maybe use the right side bar and add 'notifications' or something https://gitlab.freedesktop.org/mesa/mesa/-/issues/580000:53
Wizzupuvos: also yes it is surprisingly thing for a slider00:56
uvosbtw often the clownboot kernel is not happy at all00:59
uvos[   40.878326] unexpected IRQ trap at vector 0 repeats at nausium00:59
uvosand the device effectivly hangs00:59
uvosdosent happen every boot00:59
uvosbut often00:59
uvosbut often01:01
uvoslooks like the irq issue happens when a cpu is not active on d311:08
uvoswhen kexec happens11:08
uvosseamspinning the cpus at full speed makes it go away11:08
Wizzupuvos: ah, I think tmlind also mentioned that12:03
WizzupI don't really see it tbh12:03
uvosWizzup: have solana firmwares?12:31
uvoshttps://firmware.center/ is gohne12:31
uvoshttps://uvos.xyz/maserati/stockinfo/dts/12:39
uvoshere is xt862 dts12:39
uvosdecompiled12:39
Wizzupuvos: sec, few mins12:54
wunderwHi!13:16
wunderwI have a problem with Maemo Leste on my N90013:16
Wizzuphi13:16
Wizzupwhat kind of problem?13:16
Wizzupuvos: ping13:17
wunderwI can't run Telegram Desktop13:17
uvosWizzup: hmm?13:17
Wizzupuvos: do you want VRZ_XT862_5.5.1_84_D3G-66_M2-10_1FF_01a.xml.zip ?13:17
Wizzupwunderw: what happens if you run it?13:17
uvosi dont need it anymore atm13:17
Wizzupuvos: ok13:17
uvosi just dded the device tree from the device13:17
wunderwWhen I try to run TG desktop (installed via apt-get) it automatically closes13:17
uvosWizzup: but a android 2.3.2 device tree for xt894 would be usefull13:18
Wizzupuvos: how did you decompile it?13:18
uvosi patched dtc13:18
Wizzupcare to share that patch please?13:18
Wizzupwunderw: not enough ram maybe? what if you run it from terminal?13:18
wunderwWhen I run TG in terminal, it stucks on warning related to GLib-GObject-CRITICAL13:18
wunderwI can upload img to imgur if you want13:19
Wizzupsure13:19
uvosWizzup: btw with leste-config-droid313:19
uvosand no accel matrix13:19
uvosin udev13:19
uvosi still have the wrong orientation13:19
uvosits wron 180 deg around y13:19
Wizzupuvos: right I think I still needed to fix that13:20
wunderwI don't have any smartphone in my hand, unfortunately :( Generally, the error looks like that: (telegram-desktop:6543): GLin-GObject-CRITICAL ** <time>: g_object_new_with_properties: assertion 'G_TYPE_IS_OBJECT (object_type' failed13:21
Wizzupwunderw: you can redirect the output13:21
Wizzuptelegram-desktop &> log.txt13:21
Wizzupor something13:21
wunderwAnd GLib-GObject-WARNING **: <time>:  cannot register existing type 'GdkDisplayManager'13:21
Wizzupwunderw: where did you get it from btw?13:21
wunderwApt-get, default devuan repo13:22
wunderwWhere I can find log.txt from TG?13:22
Wizzupwait, telegram-desktop is in the debian/devuan repo?13:22
uvosyes13:22
wunderwYeah13:22
Wizzupo13:22
wunderwBut older version13:22
uvosand it works fine on d413:22
uvosim not sure why the suprise13:23
uvosits foss13:23
Wizzupmaybe it also works better on the new n900 stuff we're working on13:23
wunderwI will try full upgrading via apt-get13:23
Wizzupuvos: yeah I assumed the protocol being proprietary/centralised that the sw would be too13:23
wunderwMaybe it will help13:23
uvosWizzup: nope13:23
Wizzupuvos: wrt the dts, this seems relevant:13:23
Wizzup            bus1devices = "lm3532";13:23
Wizzup            bus2devices = "qtouch-obp-ts,isl29030_als_ir";13:23
Wizzup            bus4devices = "akm8975,kxtf9";13:23
uvosthe protocoll isent proprietary either13:23
wunderwWow, this channel is really active13:23
uvosjust the central server is13:24
Wizzupwunderw: sometimes :)13:24
wunderwGreat work, BTW13:24
wunderwI only wait for calling and SMS-es with GUI on N900's Leste ;)13:24
Wizzupwunderw: I can try it on my n900 a bit later today, but I have seen reports of people having it working for them13:24
wunderwWhat working?13:24
wunderwLeste?13:24
Wizzupwunderw: yeah we're going to push a big update soon that will at least make 3d more stable and get newer kernel13:24
wunderwWow, very good!13:24
Wizzupwunderw: no, telegram-desktop13:24
uvosWizzup: right thats why i want old xt894 to compear13:24
wunderwCan't wait :)13:24
uvossince the newer dts is different13:25
Wizzupwunderw: to be clear, people have posted screenshots of telegram-desktop on a droid4 running leste is what I meant13:25
Wizzupuvos: mhm13:25
wunderwIt's not-believiable, the device from 2010 can run 5.x.x Linux kernel13:25
Wizzupuvos: I only have VRZ_XT894_9.8.2O-72_VZW-18-8_CFC.xml.zip13:25
wunderwAnd even some desktop apps13:25
Wizzupwunderw: :)13:25
uvosthats the version i have too :(13:25
uvosthats 4.0.413:25
wunderwWizzup: Really great work13:25
Wizzupit's the work of many folks over the years13:26
wunderwYes, the main reason that I don't use Leste as main OS (still Fremantle) (N900 is my daily driver :) ) is lack of calls and SMSs support. And also some other bugs13:27
uvosWizzup: http://uvos.xyz/maserati/patches13:27
uvossms should work13:28
uvoswith sphone13:28
uvoson n90013:28
uvos(calls dont)13:28
WizzupI am planning to switch to leste as daily driver before the end of feb13:28
Wizzupeither n900 or d413:28
Wizzupmaybe d313:28
uvosWizzup: better go write a new audio driver to fix the modem issue then :P13:28
Wizzupwunderw: yeah sms and calls work to some degree, but the UIs either aren't fully there or audio is not working well yet13:29
Wizzupuvos: d4 you mean? well, first tp13:29
uvosyeah13:29
Wizzupthat's part of the deadline end of january13:29
wunderwThe only things I wait for. Then, Leste as main OS ;)13:29
uvosthats the only thing that stops me using d4 atm13:29
Wizzupwunderw: well make sure to stick around13:29
uvosand flaky gsm, but i can use qmicli13:29
uvosgsm data that is / gprs13:29
Wizzupwell that's ofono or kernel fault13:29
uvosdid you catch the dtc patch?13:30
Wizzuphonestly I think if we fix the sim not being seen when it has a pincode (on boot), we'll fix most ofono problems since they're all similar13:30
Wizzupthere seems to be something registered for sim changes already but it doesn't fire13:30
Wizzupuvos: I saw the msg didn't look yet13:30
uvosmaybe, modem audio issue is kernel tho13:30
wunderwJust ran TuxPaint on N900 :)13:31
uvosreally thats the kernels direct fault cpacp tries to do the right thing even13:31
Wizzupwunderw: :p13:31
wunderwBut in tuxpaint-config I can't change any options unfortunately13:31
wunderwThey are grey13:31
Wizzupuvos: right, we'll try to fix that at least in our tree13:31
Wizzupwunderw: yeah not everything from the repos will work out of the box currently13:32
Wizzupuvos: saw the patch now, thx13:32
wunderwBeter to use apt-get or apt?13:40
Wizzupno difference13:40
wunderwOk, thanks13:41
wunderw79 packages can be upgraded :D13:41
wunderwIs it normal that when I upgrade, the device shutdowns?13:44
uvosnope13:47
uvosits a bug and design deficancy in leste13:47
uvosyour system may now be borked13:47
wunderwNo, it rebooted and still works13:48
uvosyou have to complete the upgrade13:48
wunderwHow?13:48
wunderwI ran sudo apt update && sudo apt upgrade13:49
Wizzupyeah just run that again13:49
wunderwOk13:49
wunderwThe following packages have been kept back: hildon-meta libicd-network-ipv413:50
uvosWizzup: current leste-config-droid3 is empty13:50
Wizzupwunderw: dist-upgrade?13:51
Wizzupuvos: hmmm let me look13:51
Wizzupuvos: I think there should be a mce file13:52
wunderwWizzup: Oh, I used normal upgrade13:52
uvosWizzup: its not there13:52
uvosWizzup: dpkg -L lists nothing13:52
Wizzupuvos: looks like the debian/ files are missing13:52
Wizzupwunderw: dist-upgrade is almost always better13:52
wunderwNormal apt upgrade upgraded resolvconf and libicd-network-ipv413:52
uvoshttp://uvos.xyz/maserati/50-iio-sensors.rules13:52
Wizzupuvos: this fix accelerometer?13:53
uvosWizzup: while at it add that file13:53
uvosfixes accel13:53
wunderwIn which phone?13:54
uvosxt86013:54
uvosaccel on n900 is still broken13:55
uvosis several ways13:55
Wizzupwunderw: long story short display rotation on n900 causes crashes but we have made many many changes that will fix it in the next big push of sw updates13:57
Wizzupso we didn't want to enable accelerometer to automatically rotate display since that would crash the display server13:57
Wizzupuvos: btw droid3 -devel image is being built here https://phoenix.maemo.org/job/leste-image-droid3-dev/3/console13:57
uvosalso we need to add the i2c address of the n900 accel to the st driver13:57
uvosand fix n900 dts13:57
Wizzupyes but we will do that only after the display is fixed :)14:01
Wizzupbtw -devel based images will go here https://maedevu.maemo.org/images-devel/14:01
Wizzupin ~1.5hrs there should be a droid3 one14:01
wunderwDist-upgrade helped14:11
wunderwWizzup: thanks for this link! Will try :)14:12
wunderwWow, Leste is even for PinePhone :P14:12
wunderwNow I saw the new build! Will dd now to SD ;)14:13
Wizzupwunderw: do you mean new n900 build? it doesn't contain the new stuff yet14:20
dreamerWizzup: batteries came in. you wanted 2 right?14:20
wunderwYes14:20
Wizzupdreamer: yes pls14:20
wunderwFrom dec 2614:20
wunderwWizzup: I want to test14:20
dreamerWizzup: even included bunch of tools and stickers. one for set each battery o.O14:20
dreamerWizzup: let me know how/where/when you want them :)14:21
wunderwWant to ask, who is the main creator, developer etc. of Leste?14:21
wunderw_Can you repeat answer? XChat disconnected me :-)14:22
dreamerthey are at nurdspace now. when are you back in .nl? can do a little leste-workshop here once everyone has their 3rd shot ;)14:22
Wizzupdreamer: I am in the netherlands for 3 weeks already14:23
Wizzupdreamer: for ~2 more weeks14:24
Wizzupwunderw: there is no one creator, but freemangordon parazyd and me were there at the start, but leste builds on work of many more people before it14:24
wunderw_Thanks14:27
uvoswhat kind of battery are we horse trading?14:31
Wizzupuvos: I think some aliexpress d4 battery14:44
Wizzupdreamer: regarding when/where, not sure what makes sense14:45
Wizzupwunderw: regarding testing n900 image, there is a known problem that we don't need to test and freemangordon is working on it, once that is fixed, then it makes sense to test14:45
WizzupI mean testing the (non existing) -devel image14:46
Wizzupuvos: I think https://nl.aliexpress.com/item/32673985375.html14:46
dreamerWizzup: ah! I won't be in 020 until 10th14:48
uvosWizzup: those are oem motorola eb41 with a fake sticker14:48
Wizzupuvos: we'll find out14:49
Wizzupthis also looks interesting https://nl.aliexpress.com/item/32865155759.html?spm=a2g0o.productlist.0.0.5ba52dd5FDvTgm&algo_pvid=218dc63b-a298-444d-b825-a0e63b98cd37&algo_exp_id=218dc63b-a298-444d-b825-a0e63b98cd37-1&pdp_ext_f=%7B%22sku_id%22%3A%2265425049439%22%7D&pdp_pi=-1%3B12.24%3B-1%3B-1%40salePrice%3BUSD%3Bsearch-mainSearch14:50
Wizzupuvos: yeah rotation works14:59
Wizzup(d3)14:59
uvosgreat15:18
uvoslooks like you fixed the package15:18
uvosWizzup: wrt images15:23
uvosi think it would be better if we moved the images to devices we dont currently build/support to images-archive or something like that15:24
mighty17[m]`[    1.238769] omap-dmtimer-pwm dmtimer-pwm@10: period 31250 ns too short for clock rate 32768 Hz` any clues what the period can be?15:24
uvosso that /images/ contains only xt875 xt894 n900 rpi* and vms15:25
mighty17[m]https://github.com/Unlegacy-Android/android_kernel_ti_omap4/blob/3.4/common/arch/arm/mach-omap2/board-espresso-display.c#L149 downstream says 1200, but what?15:25
freemangordonuvos: why is that?15:25
uvosfreemangordon: to make it clear that those images are really out of date15:25
uvosand the underlying devices lack support15:25
uvosor dont work even15:25
uvoslike the n9/50 images dont even boot reportably15:26
freemangordonbut, aren't those rebuild anyways?15:26
uvosno15:26
freemangordonah15:26
uvoswe only build d4/bionic/n900/pp/rpi/vm and now d315:26
freemangordonok15:26
Wizzupuvos: right regardng images15:31
wunderwuvos: Yes, I tried to run Maemo Leste for N9, wasn't able to15:35
Wizzupwunderw: yeah ok, good point15:37
Wizzupuvos: so something like images-unsupported?15:37
uvosWizzup: that would be fine yeah15:38
uvosWizzup: or archive15:38
wunderwWizzup: the build for N900 from Dec (both) are broken. All of the screen is black, only few interface elements are visible (pink). There are not icons at all. Really strange15:41
WizzupI am not too surprised, we haven't tested that one yet15:42
Wizzupsounds like hildon-desktop not starting15:42
wunderwPossible. Flashing again the older build15:44
Wizzupuvos: untested https://maedevu.maemo.org/images-devel/droid3/20211230/16:05
Wizzupwunderw: I see the same on the images, looks like some theme symlink is messed up17:15
Wizzupuvos: the image is not working, at least theme seems broken, maybe more17:20
WizzupI wonder what broke this, no icons load, no images either17:31
Wizzup** (/usr/bin/hildon-desktop:4374): WARNING **: 16:30:59.350: hd_launcher_tile_set_icon_name: couldn't create texture for /usr/share/pixmaps/htop.png17:32
Wizzup** (/usr/bin/hildon-desktop:4374): WARNING **: 16:30:59.411: hd_launcher_tile_set_icon_name: couldn't create texture for /usr/share/pixmaps/debian-logo.png17:32
Wizzupetc17:32
Wizzupafk17:32
Wizzupfreemangordon: could this be the glib patch somehow?18:08
Wizzupparazyd: maybe the resolv.conf change somehow affects arm-sdk ?18:10
Wizzupparazyd: it's not clear to me what exact step is failing18:11
Wizzupin say https://phoenix.maemo.org/view/Images/job/leste-image-droid3-dev/4/consoleText18:11
WizzupI think it'18:12
Wizzups probably not related looking at the blend but it's suspicious it fails18:12
parazydI'll check. There should be a hardcoded resolv.conf in that stage18:15
parazydI dunno what's the problem.18:16
freemangordonWizzup: how's that?18:17
freemangordonWizzup: I don;t see how it could be related18:18
freemangordonit is about .desktop files, nothing in common with ions18:18
freemangordon*icons18:19
Wizzupgdk_pixbuf_new_from_file_at_size seems to fail18:19
Wizzupmore specifically:18:20
Wizzuppixbuf = gdk_pixbuf_new_from_file_at_size(fname, HD_LAUNCHER_TILE_ICON_REAL_SIZE, HD_LAUNCHER_TILE_ICON_REAL_SIZE, 0);18:20
Wizzupcould be something else is up, but the file path it prints definitely loads18:20
Wizzuphmm:18:22
Wizzup(osso-xterm:5923): sapwood-engine-WARNING **: 17:22:29.650: sapwood-theme: Failed to load pixmap file /usr/share/themes/alpha/images/toolbar_button_normal.png: read 1, expected 80 bytes18:22
* Wizzup reinstalls theme18:24
WizzupI don't know what is up18:25
Wizzup(sapwood-server:6040): sapwood-server-WARNING **: 17:25:41.264: /usr/share/themes/alpha/images/toolbar_button_normal.png: Couldn?t recognize the image file format for file ?/usr/share/themes/alpha/images/toolbar_button_normal.png?18:25
freemangordonbroken fs?18:26
Wizzupfreemangordon: reinstalled, same file sizes18:29
Wizzupand my d4 has the same18:30
Wizzupit looks very much like sapwood not working ok18:31
Wizzupit also seems to happen to stable images afaict18:31
Wizzupparazyd: 15:41 < wunderw> Wizzup: the build for N900 from Dec (both) are broken. All of the screen is black, only few interface elements are visible (pink). There are not icons at all. Really strange18:32
WizzupI also confirm this on -devel images18:32
Wizzupfreemangordon: seems unlikely to be broken fs since I just flashed and it also happened for this guy using stable image18:32
Wizzup/etc/hildon/theme seems ok18:32
Wizzupas is /usr/share/themes/default18:32
Wizzupit very much looks like glib-based file reading not working, since it affects h-d and sapwood18:34
Wizzupalso:18:36
Wizzup** (controlpanel:6561): WARNING **: 17:36:38.789: Couldn't load icon "general_locked": Unrecognized image file format18:36
Wizzupmaybe it's mime type related?18:37
Wizzupthat's the only thing I can think of that changed18:37
freemangordonmakes sense18:40
freemangordonbut I don;t see how it would be glib related18:40
freemangordonI made some changes to libhilodnmime18:40
freemangordonbut it just adds one more binary that creates a cache file18:40
freemangordonIt should not break mime handling, in theory18:40
freemangordonWizzup: do we have install log?18:42
freemangordondo we have shell?18:43
Wizzupfreemangordon: dinner first sry18:44
WizzupI have ssh yes18:44
freemangordonok, ping me when you're back18:44
freemangordonno issue booting maemo-leste-1.0-amd64-20211226.vdi19:00
Wizzupok19:20
Wizzupfreemangordon: ping19:20
freemangordonpong19:21
freemangordonwhat "gio mime text/plain" says?19:21
Wizzup$ gio mime text/plain19:22
Wizzup-bash: gio: command not found19:22
freemangordonhmm19:22
WizzupI also found mimetype is not installed19:22
Wizzupwhich belongs to libfile-mimeinfo-perl19:22
Wizzupshall I install that and see what happens19:22
freemangordonsec19:22
freemangordonhmm19:23
freemangordonlibglib2.0-bin should be installed, unless I am missing something19:23
freemangordonWizzup: better install libglib2.0-bin19:23
Wizzupok, sec19:23
Wizzup$ gio mime text/plain19:24
WizzupDefault application for “text/plain”: vim.desktop19:24
WizzupRegistered applications: vim.desktop debian-vim.desktop19:24
WizzupRecommended applications: vim.desktop debian-vim.desktop19:24
freemangordonso mime is ok19:24
Wizzupat least for text/plain19:24
Wizzupit might be messed up badly if this tool is needed during any preinstall19:24
freemangordoncheck /usr/share/mime19:25
Wizzupdon't know what to check but seems ok19:26
freemangordonbut mime should not be used by gdk_pixbuf_new_from_file_at_size19:26
Wizzupin any case I attached gdb to hildon-desktop and I found that that function (you just mentioned) returns NULL19:26
Wizzupfor correct path19:26
freemangordonwhat is the filesize of that file?19:26
Wizzupno icon is loaded19:26
Wizzupany19:26
Wizzupat any point in any hildon part19:27
Wizzuphildon-desktop, hildon-home, osso-xterm, controlpanel19:27
freemangordonyeah, got that19:27
Wizzupthe file syzes were the same as on the d419:27
Wizzup(which works fine)19:27
Wizzupsizes*19:27
freemangordonok, d4 works, vm works19:27
Wizzupno, hang on19:27
Wizzup*my* droid4 works19:27
WizzupI am not saying the droid4 image works19:27
freemangordonso somehting must have failed during install on n90019:27
WizzupI am testing this on a droid3, with an image I just built on the CI19:27
freemangordondo we have image build log?19:27
Wizzupand it doesn't work19:27
freemangordonah19:28
Wizzupit has the same problems as the n900 image from a few days ago, which was not -devel19:28
Wizzupthat is, status area and top left are purple19:28
Wizzupeverything else is black19:28
Wizzupwhich reminded me of the times the theme was not set up ok19:28
freemangordonwait, is this the same issue as on n900 or not?19:28
Wizzupbut the theme looks to be ok19:28
Wizzupyes19:28
Wizzupbuild log is here: https://phoenix.maemo.org/view/Images/job/leste-image-droid3-dev/4/consoleText19:28
freemangordonsymlink is ther?19:28
Wizzupyes19:28
Wizzupif you mean $ ls /etc/hildon/theme -lsh19:29
Wizzup0 lrwxrwxrwx 1 root root 25 Dec 30 17:35 /etc/hildon/theme -> /usr/share/themes/default19:29
Wizzup$ ls -lsh /usr/share/themes/default19:29
Wizzup0 lrwxrwxrwx 1 root root 23 Dec 30 17:35 /usr/share/themes/default -> /usr/share/themes/alpha19:29
freemangordonmhm19:29
Wizzupparazyd: looks like we still have debugfs not loaded btw19:29
Wizzups/loaded/mounted/19:29
freemangordonWizzup: shall I write some small test app for you to run strace on?19:31
Wizzupfreemangordon: I suppose, but I can also strace say osso-xterm19:31
freemangordonhmm, ok19:31
freemangordonplease do19:31
freemangordonyou will need ti run it by using maemo-summoner I suppose19:32
WizzupI have been doing that19:33
Wizzupd3 stuck on shutdown because of the kernel oops I think19:33
* Wizzup removes battery and restarts19:33
Wizzupfreemangordon: doesn't show anything out of the ordinary19:38
Wizzupin fact it is indistinguishable from broken and non-broken19:38
freemangordonhmm19:40
Wizzupat least anything that refers to those icon names I think19:41
freemangordonFailed to load pixmap file /usr/share/themes/alpha/images/toolbar_button_normal.png: read 1, expected 80 bytes19:41
freemangordonread 1, expected 8019:41
WizzupI think this is an error in sapwood19:41
Wizzupnot osso-xterm19:42
freemangordonyes19:42
freemangordonbut 'read 1'19:42
Wizzupwant me to strace that instead?19:42
Wizzupyeah it is weird but could be anything19:42
freemangordonsome 32 bits issue?19:42
Wizzupsome *new* 32 bits issue maybe19:42
freemangordonmhm19:42
freemangordonyes, please strace sapwood-server19:42
Wizzuphttps://dpaste.com/BCTEYFQ8P19:43
Wizzuplooks fine to me19:43
Wizzup4.0K -rw-r--r-- 1 root root 154 Jun 28  2021 /usr/share/themes/devel/images/toolbar_button_normal.png19:43
freemangordonright19:44
Wizzupsomething else that is odd, for osso-xterm, it only prints this for this png19:44
Wizzupnot for others19:44
Wizzupmaybe it has just one icon :)19:44
freemangordonnot really19:44
Wizzupbut again, h-d also cannot open pngs19:44
freemangordonbut anyways, it seems to be on the other side19:45
WizzupI see this in sapwood:19:45
Wizzup2964  write(2, "\n(sapwood-server:2964): sapwood-server-\33[1;33mWARNING\33[0m **: \33[34m18:45:22.247\33[0m: /usr/share/themes/devel/images/toolbar_button_normal.png: Couldn?t recognize the image file format for file ?/usr/share/themes/devel/images/toolbar_button_normal.png?\n", 252) = 25219:45
Wizzupwhich I also mentioned above19:46
freemangordonah, so sapwood-server complains?19:47
Wizzupyes19:47
freemangordonsec19:47
Wizzupbut it *must* be some underlying library19:47
Wizzupsince it happens everywhere19:47
freemangordonmhm19:47
freemangordonlemme check the code19:47
freemangordonhmm, this error message comes from elsewhere, not from sapwood itself, iiuc19:49
freemangordonWizzup: do you have XDG_DATA_DIRS set?19:50
Wizzupnot on faulty droid3 and not on working droid419:51
freemangordonok19:52
freemangordonchecking one more thing19:52
freemangordonWizzup: /usr/lib/x86_64-linux-gnu/gdk-pixbuf-2.0/gdk-pixbuf-query-loaders, modified for 32 bits19:53
freemangordonI bet this one differs on 2 devices19:55
WizzupI was looking at that just now19:55
Wizzupdoesn't look like it19:56
WizzupI'll paste19:56
Wizzuphttps://dpaste.com/4ZWM2YG99 droid319:57
Wizzupwait let me run it as user to be sure19:57
freemangordonhmm, yeah,looks fine19:57
Wizzupdroid3: https://dpaste.com/EMBPTGNYH    droid4: https://dpaste.com/9LR83AVF519:57
freemangordonbut we're getting close19:57
Wizzupyup19:57
Wizzupsomething is truly messed up19:58
freemangordonldd /usr/lib/arm-linux-gnueabihf/gdk-pixbuf-2.0/2.10.0/loaders/libpixbufloader-png.so ?19:58
Wizzupdo you want to compare?19:59
Wizzupthey look the same on d3 and d419:59
freemangordonok19:59
freemangordonhmm19:59
freemangordonwait, you said you don;t have XDG_DATA_DIRS set for which user?19:59
Wizzupall devices do not have that set for usersyes19:59
Wizzupuser yes19:59
Wizzupuser@maindroid:~$ env|grep XDG20:00
WizzupXDG_RUNTIME_DIR=/run/user/100020:00
freemangordoncould you run-standalone.sh bach20:00
freemangordon*bash20:00
freemangordonand then check XDG stuff20:00
freemangordonI have XDG_SESSION_COOKIE=devuan-1640512024.348006-1839873804 here20:00
Wizzupno difference20:01
Wizzupon both devices20:01
freemangordonno XDG_SESSION_COOKIE?20:01
Wizzupno20:01
freemangordonhmm, why do I have that here20:01
freemangordonin the VM that is20:01
Wizzupdon't know but unlikely to relate to png file parsing20:01
freemangordonyeah20:01
Wizzuptoo bad that h-d doesn't even log the error of gdk_pixbuf_new_from_file_at_size20:03
Wizzupalthough I suppose it will be the same: cannot recognize image format20:03
freemangordongdk-pixbuf-query-loaders > /usr/lib/arm-linux-gnueabihf/gdk-pixbuf-2.0/2.10.0/loaders.cache20:04
freemangordonupdate-mime-database /usr/share/mime20:04
freemangordonI guess as root20:05
Wizzupsee https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=234784#c420:05
Wizzupfreemangordon: ok let me try20:05
Wizzupfwiw /usr/lib/arm-linux-gnueabihf/gdk-pixbuf-2.0/2.10.0/loaders.cache are the same on both devices before updating20:05
freemangordonok20:06
Wizzupfreemangordon: yes, update-mime-database solved it20:06
freemangordonhmm20:06
freemangordonwhy it is not called?20:06
Wizzupparazyd: ^^^^^20:06
parazydThat should be part of some postinst20:07
freemangordonI see no update-mime-database call in build log20:07
Wizzupfreemangordon: I do see this:20:07
Wizzup* Updating MIME database in /usr/share/mime...20:07
freemangordonshared-mime-info20:07
freemangordonWizzup: this is from hildon-update-=category-database20:08
Wizzupright20:08
freemangordonProcessing triggers for shared-mime-info (1.10-1)20:08
freemangordonand nothing20:09
Wizzupyeah and then nothing happens20:09
Wizzupyou're sure it's in that postinst?20:09
Wizzuptriggers are not the same as postinst fwiw20:09
freemangordonnot sure20:09
freemangordonthis is not our package20:09
WizzupI would expect it to happen here: Setting up shared-mime-info (1.10-1) ...20:09
freemangordonmhm20:09
freemangordongetting source, lemme check20:09
WizzupI suppose I can reflash the image and apt --reinstall shared-mime-info20:10
Wizzupfreemangordon: fwiw here is the vm build log https://phoenix.maemo.org/view/Images/job/leste-image-virtual/83/consoleText20:10
freemangordonlooks good to me https://pastebin.com/VWW4ewmn20:11
freemangordonWizzup: no idea :(20:12
Wizzupsame.20:12
Wizzupthis is truly SNAFU20:12
freemangordonbut yeah, please reflash and reinstall20:12
Wizzupk20:12
freemangordonparazyd: is it possible /me adding a trigger in another package to break this?20:12
freemangordonthis https://github.com/maemo-leste/libhildonmime/blob/master/debian/hildon-update-category-database.triggers20:13
parazydWhat do you mean with "trigger"?20:13
freemangordonparazyd: ^^^20:13
parazydhmm perhaps. I don't know what those do20:13
Wizzupwell I think the point is that we seemingly didn't change anything but the mime files of (most?) images are broken to the point it cannot even load a single png20:14
Wizzupand the fix is to run update-mime-database20:14
Wizzupbut that must already get run20:14
freemangordonmhm20:14
freemangordonbut I don;t see it being run20:14
freemangordonwhich means postinst didn;t get called, IIUC20:14
Wizzupso perhaps we set some env var now that causes the postinst not to run or something, or it runs too soon, or whatever...20:14
WizzupI suppose we could force-add the call to arm-sdk or something20:19
Wizzupanyway I'll reflash20:20
freemangordonWizzup: I think we shall find the reason for postinst not being called20:20
freemangordonwe could have similar breaks elsewhere20:20
Wizzupit might get called as well, but the file just not kept20:21
Wizzupfwiw there is also /var/lib/dpkg/info/shared-mime-info.postinst20:22
freemangordonhmm, is it possible that hildon-update-category-database breaks mime info somehow?20:22
Wizzupand that definitely contains a check for triggered20:23
Wizzupfreemangordon: maybe, can check20:23
Wizzupdd'ing now20:23
freemangordonWizzup: could you save /usr/share/mime before reinstalling?20:23
Wizzuptoo late20:24
Wizzupbut I can re-run after dd20:24
freemangordonno, I mean the broken one20:24
freemangordonso we can compare after fixing20:24
freemangordonwell, why VM is not broken tren?!?20:26
Wizzupfreemangordon: I understand, I will cp before fixing20:27
Wizzupbut the dd already started when you asked20:27
freemangordonthat's fine as you can;t get the broken one without dd :)20:28
Wizzupfreemangordon: regarding the triggers, that seems likely the cause somehow20:33
Wizzupdoes the shared mime info package also have those triggers20:34
freemangordonI guess yes, but I would expect more that one trigger per directory to be allowed20:34
* freemangordon checks20:34
Wizzupyeah me too20:34
Wizzuphttps://manpages.debian.org/unstable/dpkg-dev/deb-triggers.5.en.html20:35
Wizzuphttps://wiki.debian.org/DpkgTriggers20:35
Wizzuphttps://sources.debian.org/src/dpkg/1.21.1/doc/triggers.txt/20:35
freemangordonwell, I see no issue with what I did20:38
freemangordonunless i am missing something20:38
Wizzupfwiw wl12xx is also not probed on d320:38
Wizzupunless I probe it manually20:38
WizzupI also feel like we need to provide some default nameserver to dnsmasq20:40
freemangordonhmm, hmm20:40
Wizzupin case of ethernet or usbnet20:40
Wizzuphttps://dpaste.com/G6ZCFNC5Z20:40
Wizzupfreemangordon: is it possible your trigger exits with non zero exit code20:40
freemangordonI doubt20:41
freemangordonhttps://github.com/maemo-leste/libhildonmime/blob/master/debian/hildon-update-category-database.postinst#L820:41
Wizzuphmm20:41
Wizzupin any case the --reinstall works20:42
Wizzuplet me share old and new mime dirs20:42
Wizzuphttps://wizzup.org/mimes.tar.bz220:43
Wizzupfreemangordon: # cat /var/lib/dpkg/info/hildon-update-category-database.postinst20:43
Wizzup#! /bin/sh20:43
Wizzupis that valid, with the space?20:43
freemangordonhmm, is it possible there is no /usr/share/mime by the time hildon-update-category-database is run?20:43
freemangordonyes, it is20:43
Wizzupok20:43
Wizzupweird20:43
freemangordonsee my question ^^^20:44
WizzupI have no idea what installs into /usr/share/mime20:44
Wizzupor what creates it20:44
freemangordonshared-mime-info, hildon-application-manager, osso-pdf-viewer, gcr, meld: /usr/share/mime20:44
Wizzup# dpkg -S /usr/share/mime20:44
Wizzupshared-mime-info, hildon-application-manager, osso-pdf-viewer: /usr/share/mime20:44
Wizzupright20:44
freemangordonmhm20:44
Wizzupmaybe you can have it depend on shared-mime-info in any case20:44
freemangordonright20:44
freemangordonlemme see what is the $? if non-existent dir is passed20:45
Wizzupok20:45
Wizzupplease also fix the space in the shebang20:45
Wizzupit's confusing20:45
WizzupI've never seen it20:46
freemangordonyep. sudo hildon-update-category-database /usr/share/mimes gives exit code 120:46
WizzupI think this is probably the problem20:46
freemangordonmhm20:46
freemangordonor, I can just create that dir20:46
Wizzupbetter to depend on shared-mime-info20:46
Wizzupimho20:46
Wizzupbut in any case20:47
Wizzupit still should not cause the postinst to return non-zero20:47
Wizzupif I read it correctly20:47
Wizzupso I don't see how fixing that would help20:47
Wizzupit is set -e no?20:47
freemangordonyes20:47
Wizzupmaybe see what the postinst returns20:47
Wizzupif there is no dir20:47
Wizzupmaybe it is the shebang somehow :)20:48
freemangordon"-e  Exit immediately if a command exits with a non-zero status."20:48
freemangordonnothing wrong with the shebang20:48
Wizzupoh20:48
WizzupI always confuse -e and +e20:48
Wizzup:)20:48
Wizzup-e to me sounds no "minus errors"20:49
freemangordonI don;t because I know neither :)20:49
Wizzupsounds like20:49
Wizzupyeah I know it controls if shell exits right away if any process erors20:49
freemangordonanyway, I think this is it, will make it depend on20:49
Wizzupbut I always forget what is - and what is +20:49
Wizzupok20:49
freemangordonmaybe pre-depends?20:49
freemangordonWizzup: how we're going to test that?20:50
Wizzupfreemangordon: test what?20:53
freemangordonimage building with fixed package20:53
WizzupI will trigger another build20:53
Wizzupyou can push to -devel20:53
freemangordonok20:53
freemangordonok20:53
Wizzupmy droid3 image uses -devel20:53
Wizzup:)20:53
Wizzupuvos: btw I still see hangs sometimes on droid3, do you see them as well?22:10
Wizzupfreemangordon: any idea how telepathy-logger works?23:02
Wizzupit seems useful but I don't see any users of it really23:02
freemangordonI think there are plugins to it23:04
freemangordonwhich expose dbus interfaces23:04
freemangordonbut that's all I know about it23:04
freemangordonimage seems to be almost ready. lets hope it is fixed23:05
WizzupI mean it seems to suggest it logs somewhere23:11
Wizzupand you can read the logs23:11
Wizzupbut maybe I am misunderstanding it23:11
Wizzupfremantle doesn't use it23:11
Wizzupso I guess we will not use it23:14
Wizzupah they have log store plugins23:15
Wizzupthat is interesting23:15
Wizzuphttps://github.com/TelepathyIM/telepathy-logger/blob/master/telepathy-logger/log-store-empathy.c23:16
Wizzupwe could make a log-store-rtcom23:16
Wizzuphm I think it only reads and doesn't write23:17
Wizzuphttps://github.com/TelepathyIM/telepathy-logger/blob/master/telepathy-logger/log-store-sqlite.c23:18
Wizzuplooks like there is some add_event23:19
Wizzuphmm https://github.com/TelepathyIM/telepathy-logger/blob/master/telepathy-logger/log-store-sqlite.c#L9023:20
freemangordonWizzup: image is ready23:27
Wizzupyes23:28
WizzupI am downloading already23:28
freemangordonok, please LMK if it is ok23:29
Wizzupworking on it23:29
Wizzupso I think telepathy-logger could be interesting potentially23:32
Wizzupwe could just use it's binary and write a rtcom plugin23:32
Wizzupthen we don't need a logger per program23:32
Wizzupi.e. sphone could drop it's tp interface23:32
Wizzupbut it's also relatively limited in what it does, so maybe it's not that useful23:32
freemangordondoes it store enough info?23:33
Wizzupwell we will have to fork it anyway to add our plugin23:33
Wizzup(most of the code is untouched for years)23:33
Wizzupbut I am not sure if it makes sense to have it23:33
freemangordonwhy fork?23:33
Wizzupthe plugins are built in it seems23:33
freemangordonI am almost sure it can be extended, lemme check23:34
Wizzupthe htmldoc is pretty poor too23:34
Wizzupand my fav. telepathy guide doesn't even log it23:34
Wizzupen the tp homepage 404's for the telepathy-logger23:35
Wizzup(not saying we should not use it, just some things that discouraged me a bit from using it)23:35
Wizzup(dd almost done)23:35
Wizzupfreemangordon: still broken it looks like :(23:39
freemangordon:(23:39
freemangordonok, going to have some sleep, will see what we can do tomorrow23:40
Wizzupmaybe focus on the n900 stuff, I can get this fixed surely23:40
freemangordonyeah23:41
freemangordonnight!23:41
Wizzupfreemangordon: we can also run it at the end of the arm-sdk for now23:42

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!