fury | but it's more like, people coming by to see AGL, they can't just play around with it at will because it'll be locked up, then if they reboot it, half the time weston doesn't start up...so my intros to it at work always start with "if you start it up and it sticks on the AGL loading screen..." | 00:00 |
---|---|---|
JEEB | LoveMHz: yea. also I recommend marcan's CCC talk on the PS4. gets interesting on how they've since gotten away from the "PC" while still being x86_64 | 00:00 |
fury | and then on the pi, if you start up more than a few apps, bam | 00:00 |
LoveMHz | Seen it. It's crazy how much a mess the PS4's arch is. | 00:01 |
fury | it's so awesome when it works tho | 00:01 |
JEEB | I know that feeling. I remember when audio policy management started working in late 2014 | 00:02 |
JEEB | when you'd actually get music and navigation properly mixed | 00:02 |
fury | :D | 00:03 |
fury | it's magical | 00:03 |
JEEB | (music goes down, navigation says words, music goes back up) | 00:03 |
fury | music is supposed to go down when navigation happens? | 00:03 |
JEEB | yes, so that you can hear the navigator's message | 00:04 |
fury | i just hear it kinda blend in, not that i can understand very well what it's trying to say anyway, but yeah, makes sense that it should duck | 00:04 |
JEEB | and this was back in the olden days with tizen ivi :P | 00:04 |
fury | oh :P | 00:04 |
JEEB | no idea what AGL does | 00:04 |
fury | billion things changed since then | 00:04 |
JEEB | I bet | 00:04 |
JEEB | but yea, I was one of the folk working on policy management stuff | 00:04 |
JEEB | it ended up working as a display manager as well :D (it would tell weston what to put up/down) | 00:05 |
fury | thatd be cool to be able to show a banner or something like that for navigation or notifications | 00:06 |
JEEB | yea, layer management | 00:06 |
fury | one step at a time tho :P i hear the home screen is getting worked into an app now | 00:06 |
JEEB | cool | 00:06 |
fury | i'd like to try to figure out how to modify everything to fit the pi's 7" in landscape | 00:07 |
* JEEB mostly worked with the "has functionality but is ugly as hell" ICO homescreen back in the day | 00:07 | |
fury | cause i find it hard to fit the scaled portrait version | 00:07 |
*** Mathieu_Du has joined #automotive | 00:07 | |
fury | i like both - little to no functionality but looks really pretty, or has functionality but is ugly as hell | 00:07 |
fury | i've done a little of both | 00:08 |
fury | mocked up an html5 ui that you could control using just arrow keys and space or with touch | 00:08 |
fury | no functionality whatsoever | 00:08 |
fury | looked pretty neat | 00:08 |
JEEB | yea, the neat one was HTML5 | 00:08 |
fury | in my case right now i'm leaning toward functionality but ugly as hell :P my QML did not translate well to 1080x1400something | 00:09 |
JEEB | the functional one was EFL+stuff | 00:09 |
*** treitter has joined #automotive | 00:09 | |
JEEB | hehe | 00:09 |
fury | fonts are all off, floating beyond the rectangles they're supposed to be contained in etc | 00:09 |
fury | but it opens up a network server that i can connect to from a client elsewhere on the network! | 00:09 |
fury | and i bet it would open up this serial port if i had a recent enough kernel with working usb to serial drivers | 00:09 |
fury | and it works the same as when it's compiled for windows, mac, or ios | 00:11 |
fury | bet it'd work on android too | 00:11 |
fury | (at least the client side) | 00:11 |
fury | which just opens the network connection to the server | 00:12 |
fury | not a serial port | 00:12 |
fury | qt's like, write once compile everywhere | 00:12 |
fury | isn't that what java was supposed to be? | 00:12 |
JEEB | java was supposed to be write and compile once, hope it runs well enough everywhere under the JVM | 00:13 |
fury | :P | 00:13 |
fury | i shoulda ran this dd through pv | 00:13 |
fury | i have no idea how far along it is in the process | 00:13 |
*** jlrmagnus has quit IRC | 00:17 | |
fury | whew, only 800 seconds | 00:18 |
fury | for 3.5 gigs | 00:18 |
fury | ogod | 00:18 |
fury | i forgot to sync before unplugging it | 00:18 |
fury | ._. | 00:18 |
JEEB | oh, found one pic of the olden homescreen, and this also shows that the layer management did actually work (the application gets put where it belongs on the screen) https://kuroko.fushizen.eu/screenshots/tivi/mpv_under_ico_homescreen.png | 00:19 |
JEEB | this was me poking around way after the fact at home | 00:19 |
fury | cool :D | 00:19 |
JEEB | I was most surprised by the fact that the playback worked nicely, and after I told the application framework that it was media the pulse stuff worked nicely, too | 00:20 |
JEEB | (also mpv's old UI worked surprisingly well with touchscreens) | 00:20 |
fury | that'd be a fun thing to do, too, get some video up on screen | 00:21 |
fury | wonder what it'd take to do that | 00:21 |
JEEB | libmpv would be my recommendation | 00:21 |
fury | sheesh | 00:21 |
JEEB | give it an opengl context | 00:21 |
JEEB | (ES or normal) | 00:21 |
fury | now that i've compiled one thing and gotten it up on screen my mind's going 100 mph | 00:21 |
JEEB | and it will render into it | 00:22 |
fury | it took me like, a couple years to get to this point, i have no chill :P | 00:22 |
JEEB | I think it was this one which I got working on Android as well https://github.com/mpv-player/mpv-examples/tree/master/libmpv/qml_direct | 00:23 |
JEEB | QML example with libmpv | 00:23 |
JEEB | although I patched it to pop up a file picker | 00:24 |
JEEB | instead of hardcoding a path | 00:24 |
JEEB | yup, it was the qml_direct example :) | 00:25 |
* JEEB is an OSS multimedia guy if it kind of didn't show from the whole "puts media players into R&D platforms" thing | 00:29 | |
fury | :D | 00:31 |
fury | is bitbake for raspberry pi not generating sd card images anymore? | 00:56 |
fury | this is an old sd card image O_o | 00:56 |
smurray | it should be afaik | 01:00 |
fury | i get a rootfs.ext4.xz and a rootfs.tar.xz | 01:05 |
fury | and the old ota images i made from when sota was supposed to work | 01:05 |
fury | no agl-demo-platform-raspberrypi3.rpi-sdimg anywhere | 01:05 |
smurray | hmm, there was a change about forcing specific image types floating by, might be an unwanted side-effect | 01:08 |
smurray | doesn't seem like that's been merged yet, though | 01:09 |
fury | it was merged in master | 01:12 |
fury | i just grabbed from master right about then | 01:12 |
fury | https://gerrit.automotivelinux.org/gerrit/#/c/9765/ | 01:13 |
smurray | ah, he didn't merge it into the dab branch yet | 01:15 |
smurray | I'm not sure what the plan is for rpi then | 01:15 |
fury | https://jira.automotivelinux.org/browse/SPEC-409' | 01:15 |
fury | i have no idea | 01:16 |
smurray | fury: that issue is about improvements to the sdcard image class, the merged change drops the sdcard image from the list of targets | 01:16 |
smurray | dl9pf: awake? | 01:17 |
smurray | fury: there may be some work afoot to replace the rpi image creation with the new wic stuff in OE, but I'm not aware of it | 01:18 |
fury | ok cool | 01:18 |
*** waltminer has quit IRC | 01:19 | |
fury | hmm. if i knew how to manually put things in place on the pi i'd prolly do that, but i think i'll just go home and sleep a bit :P | 01:19 |
smurray | fury: might be worth pinging the ml, it's possible that an exception might need to be made for rpi in the short term | 01:20 |
fury | cool | 01:29 |
fury | i'm super excited, i might not be able to sleep :P | 01:33 |
*** mdurnev has joined #automotive | 02:10 | |
*** mkelly has joined #automotive | 02:19 | |
*** mkelly has quit IRC | 02:28 | |
*** mkelly has joined #automotive | 02:29 | |
*** mkelly has quit IRC | 02:33 | |
*** D9 has quit IRC | 02:34 | |
mranostay | heh | 02:35 |
fury | I am an amazing predictor of the future | 02:36 |
*** D9 has joined #automotive | 02:38 | |
*** mkelly has joined #automotive | 02:39 | |
*** mkelly has quit IRC | 02:51 | |
*** mkelly has joined #automotive | 02:58 | |
*** mkelly has quit IRC | 03:02 | |
*** mkelly has joined #automotive | 03:03 | |
*** akilaw has joined #automotive | 03:10 | |
*** mkelly has quit IRC | 03:25 | |
*** AlisonChaiken has joined #automotive | 04:12 | |
*** akilaw has quit IRC | 04:36 | |
*** fredw has joined #automotive | 05:37 | |
*** kholdfuzion_ has quit IRC | 05:58 | |
*** msisov has joined #automotive | 06:09 | |
*** jobol has joined #automotive | 06:15 | |
*** leon has joined #automotive | 06:25 | |
*** leon is now known as leon-anavi | 06:26 | |
leon-anavi | morning | 06:26 |
mranostay | jobol: hey i'm noticing the service bindings in their own package are taking up blank icon spaces on the HomeScreen | 06:30 |
mranostay | any clue how easy it would be for that not to happen? :) | 06:30 |
*** mdurnev has quit IRC | 06:58 | |
*** mdurnev has joined #automotive | 06:58 | |
*** claneys has joined #automotive | 07:01 | |
*** mdurnev has quit IRC | 07:07 | |
*** Figure has joined #automotive | 07:18 | |
*** kooltux__ has joined #automotive | 07:20 | |
*** mfritzsch has joined #automotive | 07:23 | |
*** User3209 has quit IRC | 07:29 | |
*** kooltux__ has quit IRC | 07:30 | |
*** bruce has joined #automotive | 07:32 | |
*** bruce is now known as Guest76021 | 07:33 | |
jobol | mranostay, I'll look at that today | 07:33 |
*** [AD]Turbo has joined #automotive | 07:45 | |
[AD]Turbo | hi there | 07:45 |
mranostay | jobol: cool thanks.. i would but haven't touched that code yet | 07:46 |
*** kbroulik has joined #automotive | 07:49 | |
*** toscalix has joined #automotive | 07:57 | |
*** ZenEntropy has joined #automotive | 07:57 | |
*** mdurnev has joined #automotive | 07:59 | |
Vamsi | Hi I am student evaluating Automotive Grade Linux as an Automotive IoT Platform. And I am using Raspberry Pi as the Hardware. Can anyone please tell me where I could find documentation to set up an environment to generate a complete AGL build for raspberry pi? Also to set up an environment for app development for raspberry pi. | 08:13 |
msisov | Vamsi: Hi! You can refer to the following guide - http://docs.iot.bzh/docs/getting_started/en/dev/reference/machines/raspberrypi.html | 08:19 |
Vamsi | I did refer this page. And is it as simple as just a single command for generating the AGL image? | 08:28 |
Vamsi | msisov: Thank you :) | 08:30 |
msisov | yes, you can add other packages there as well | 08:30 |
msisov | http://docs.iot.bzh/docs/getting_started/en/dev/reference/setup-sdk-environment.html and this is for sdk | 08:30 |
msisov | even though, I've never used that | 08:30 |
msisov | but it should be helpful | 08:30 |
*** jobol has quit IRC | 08:36 | |
*** jobol has joined #automotive | 08:36 | |
Vamsi | msisov: Thanks | 08:42 |
Vamsi | has anyone already created some demo apps specifically meant for raspberrypi? I was wondering if there were available for testing. | 08:42 |
*** mfritzsch has left #automotive | 08:43 | |
jobol | mranostay, actually the permission urn:AGL:permission::public:hidden should solve I'll fill SPEC-702 | 08:44 |
*** jacobo has joined #automotive | 08:57 | |
*** fredw has quit IRC | 08:58 | |
*** fredw has joined #automotive | 09:14 | |
*** crankslider has joined #automotive | 09:24 | |
*** [AD]Turbo has quit IRC | 09:26 | |
*** [AD]Turbo has joined #automotive | 09:27 | |
*** fredw has quit IRC | 09:29 | |
*** praneeth has quit IRC | 09:41 | |
*** praneeth has joined #automotive | 09:42 | |
*** gattuso has joined #automotive | 10:07 | |
gattuso | Hi guys | 10:08 |
CTtpollard | hi gattuso | 10:10 |
*** chbae has joined #automotive | 10:20 | |
*** kbroulik has quit IRC | 10:36 | |
*** kbroulik has joined #automotive | 10:36 | |
*** chbae has quit IRC | 10:43 | |
*** fredw has joined #automotive | 10:46 | |
*** chbae has joined #automotive | 11:08 | |
*** mdurnev has quit IRC | 11:10 | |
*** mdurnev has joined #automotive | 11:10 | |
*** caiortp has joined #automotive | 12:19 | |
fury | Vamsi: there’s a few extra steps to getting the navigation and POI apps to work - the map data does not come on the image because it’s a couple extra gigs (and the SD card partition/FS itself needs to be expanded to fit it). Also, the POI app requires a Yelp developer account’s API credentials. As for the SDK, I just got up and running with the XDS yesterday and built my first app. I’ll be in the office in about an | 12:25 |
fury | hour, will see if I can dig anything up to help :) | 12:25 |
Vamsi | fury: Thank you very much. This is will be of great help :) | 12:27 |
*** crankslider has quit IRC | 12:28 | |
*** wesam has joined #automotive | 12:37 | |
*** waltminer has joined #automotive | 12:52 | |
Vamsi | Is it normal for the AGL build to take a very long time? | 12:57 |
CTtpollard | lots of variables in that question | 12:58 |
CTtpollard | but in general, yes | 12:58 |
CTtpollard | especially on first build | 12:58 |
Vamsi | CTtpollard: Well this is the first build and it is taking an enormous amount of time. by lots of variables, you mean there are other reasons also that lead to large build times? | 13:01 |
fury | I found that it took about 3 hours on the average laptop or desktop PC, and that was with a full download cache | 13:02 |
fury | It builds everything from source, even the cross compiler | 13:03 |
CTtpollard | Vamsi: build hardware, target image, internet connection | 13:03 |
Vamsi | CTtpollard: The build hardware and the internet connection are good. target image, the target is a raspberry pi 3. | 13:06 |
Vamsi | fury: right, I observed this too. | 13:06 |
CTtpollard | I wouldn't say it's any different to similar yocto based systems | 13:07 |
CTtpollard | (in terms of build time) | 13:07 |
CTtpollard | especially if building webkits & rust | 13:08 |
*** msisov has quit IRC | 13:30 | |
*** mdurnev has quit IRC | 13:40 | |
Vamsi | fury: You said that you were able to set up the app development environment, could you please share steps you followed? | 13:50 |
Vamsi | Also, has anyone here used the XDS for app development for AGL? | 13:50 |
*** wesam has left #automotive | 13:50 | |
fury | Vamsi: yeah, i've got the xds-server docker image running on a big build server in-house (same machine i use to build AGL images), then I run xds-agent on my Windows desktop or Linux laptop to set up a project, and finally xds-exec / xds-make to actually build the thing | 13:56 |
fury | i used it yesterday to get my first app up and running | 13:56 |
fury | i'm trying to get an SD card image built for pi at the moment (I customized my image a little bit) | 13:57 |
fury | are you running with master or with chinook (3.0.3)? | 13:57 |
Vamsi | fury: master has some problem because of a new check-in (or that's what I understood from the community). | 14:00 |
Vamsi | So it's not psooible to load the image onto the SD card. | 14:00 |
fury | yeah, i just ran into that myself last night. but i think (stop me if i'm wrong leon-anavi) if you configure it with agl-sota feature it'll generate the image? i'm testing that theory now, will check it in about 5 minutes when it's done with the build | 14:00 |
fury | the last few steps on mine take an abnormally long time despite it supposedly being a fast SSD O_o | 14:01 |
Vamsi | fury: Could you please share the link to documentation etc for setting up xds on my local linux host? | 14:02 |
fury | https://github.com/iotbzh/xds-server | 14:02 |
smurray | it's a one liner to enable the sd card image again | 14:02 |
fury | sweet | 14:03 |
fury | i like one liners | 14:03 |
fury | i thought sota was broke anyway because of some problem causing uboot to no longer work, so pi got switched over to image | 14:03 |
smurray | yeah, sota is still being debugged on rpi | 14:04 |
fury | cool | 14:04 |
smurray | likely the easiest fix would be to temporarily revert https://gerrit.automotivelinux.org/gerrit/#/c/9813/6 in your tree | 14:04 |
fury | easy enough | 14:05 |
*** fredw1 has joined #automotive | 14:05 | |
leon-anavi | fury, yes | 14:05 |
leon-anavi | the image will be successfully generated for agl-sota | 14:06 |
fury | but it won't actually work? | 14:06 |
smurray | fury: or add IMAGE_TYPES_append = " rpi-sdimg" in site.conf | 14:06 |
leon-anavi | there are some other known issues as described in SPEC-671 | 14:06 |
fury | that's even easier | 14:06 |
leon-anavi | furthermore the latest version of meta-updater is needed | 14:06 |
smurray | I'm a bit busy atm, so don't have the time to work out where the proper place would be to tweak IMAGE_TYPES in meta-agl | 14:07 |
fury | ok cool | 14:07 |
leon-anavi | I am working on updating in AGL-repo but before that my GitHub pull request for bumping the version of Aktualizr in meta-updater needs to be merged | 14:07 |
fury | thanks guys :D | 14:07 |
*** fredw has quit IRC | 14:07 | |
*** fredw1 is now known as fredw | 14:07 | |
fury | that'll let me test this serial port thing on the pi, my theory is the driver is buggy on the old kernel in the porter, so i'm hoping and crossing my fingers it's fixed by 4.4 | 14:08 |
CTtpollard | the pi has a serial port/ | 14:09 |
CTtpollard | ? | 14:09 |
fury | Vamsi: if you're following the steps for the docker image you can skip the "Manually start XDS server" | 14:09 |
fury | CTtpollard: no, usb to serial | 14:09 |
CTtpollard | ahh, ftdi uart | 14:09 |
fury | i think it has a serial console but i haven't tried to hook up to it...i'm trying to talk to an RS232 device | 14:09 |
fury | yeah something like that | 14:09 |
Vamsi | fury: okay. thanks. | 14:09 |
fury | hmm. IMAGE_TYPES_append = " rpi-sdimg" didn't seem to do it | 14:14 |
smurray | there's actually two serial ports on rpi, but one of them was repurposed for the bt module on rpi3 | 14:15 |
fury | both of the usb to serial devices i have (keyspan 1-port adapter and edgeport 4-port adapter) did not work on 3.10.31 kernel - it didn't even pick up the edgeport (it enumerated on the bus but the driver didn't grab it), and for the keyspan, it was sending/receiving garbage | 14:15 |
smurray | it's usable if you disable the bt | 14:15 |
fury | cool | 14:16 |
fury | prolly should disable the built in bluetooth chip anyway because of its performance issues | 14:16 |
fury | is that accessible on the GPIO? | 14:16 |
CTtpollard | I've always done serial to it over the tx/rx on the gpio | 14:17 |
smurray | fury: both serial ports are on the header, you need to play with pinmux to get the second one back from the bt on the rpi3 | 14:18 |
fury | ah ok | 14:18 |
fury | cool | 14:18 |
fury | that's another option then if this USB to serial flat out doesn't work | 14:18 |
fury | hallelujah, reversed 9813 and it built me an rpi-sdimg | 14:19 |
fury | is there anything on the demo HMI that shows the current IP, or do i have to guess with an nmap if i don't hook up to serial console? | 14:20 |
smurray | probably the latter | 14:21 |
fury | sounds good | 14:21 |
fury | i love talking to IT anyway :D | 14:21 |
CTtpollard | :P | 14:21 |
smurray | there might be something in the settings app if you're using wifi | 14:21 |
fury | ooh | 14:21 |
fury | that works too | 14:21 |
fury | i'll hop onto the wifi | 14:21 |
fury | i think this router in my office lets me see the IP addresses of wireless connected devices | 14:21 |
fury | but not wired | 14:22 |
*** khem has joined #automotive | 14:25 | |
*** jlrmagnus has joined #automotive | 14:26 | |
fury | Vamsi: once you've got the xds-server container running, you can access the dashboard at http://localhost:8000 - from there, you can download the agent, extract it somewhere memorable and open a new terminal window, cd to it and ./xds-agent | 14:32 |
fury | oh | 14:33 |
fury | the cross sdks | 14:33 |
fury | for the pi you'd want to ssh into the docker container, run this: /usr/local/bin/xds-utils/install-agl-sdks.sh --arch armv7vehf | 14:34 |
fury | that'll get the cross sdk downloaded and installed in the xds server | 14:34 |
fury | once THAT's all done, you should see blue rows on the XDS dashboard configuration | 14:35 |
fury | and the armv7vehf in the cross sdks configuration | 14:35 |
fury | so i guess from there it just depends on what app you want to build | 14:36 |
fury | i did a QML app since that's something i'm already pretty familiar with, but i hear you can do an HTML5 UI, or a completely native one | 14:37 |
*** halfline has quit IRC | 14:37 | |
*** halfline has joined #automotive | 14:37 | |
Vamsi | fury: Can I install the SDKs for both pi and inte | 14:39 |
fury | i think so | 14:40 |
Vamsi | oops! | 14:40 |
fury | if you want to build for qemu too, yeah that should work | 14:40 |
Vamsi | fury: Can I install the SDKs for both pi and intel in the same container right? | 14:40 |
fury | i'm a newbie at the SDK but that seems to be possible based on the way it looks in the video iot.bzh released | 14:40 |
fury | from what i understand, once you've got the cross SDKs installed, and you've got a project set up, you can go to the next tab in the dashboard and build it using either SDK | 14:41 |
fury | or you can use xds-exec and xds-make | 14:42 |
fury | on your command line | 14:42 |
fury | but probably easier to do from the dashboard | 14:42 |
claneys | dashboard is easier to begin | 14:43 |
fury | yeah. xds-exec and xds-make are for IDE integration, i would assume | 14:43 |
claneys | Vamsi, you can install both SDK in a XDS-server docker container then choose which one you want at build time | 14:43 |
claneys | fury, right | 14:43 |
claneys | or for cli aficionados =) | 14:44 |
fury | cool. i'm looking into how to put those into my qt creator project to make it as simple as building for desktop | 14:44 |
fury | i managed to build one yesterday from the command line after hacking up the qt creator project and adding the icon.svg and config.xml and package.pro | 14:46 |
fury | damn | 14:46 |
fury | i had an empty home screen on first boot of the pi, and now i see nothing on screen | 14:46 |
claneys | great | 14:46 |
fury | i've rebooted about 5 times now :( | 14:46 |
claneys | erf, good luck with that. I did not display anything from mine for a while because i don't care for the moment of the display :/ | 14:47 |
claneys | can't help on that sry | 14:47 |
fury | no worries | 14:47 |
fury | time to dust off my nmap googling skills | 14:48 |
claneys | mouarf | 14:48 |
claneys | sounds like a good quest ! | 14:48 |
*** jeremiah has joined #automotive | 14:58 | |
jeremiah | ohai | 14:58 |
fury | https://pastebin.com/54t88sPk | 14:58 |
fury | ever seen failures like that? failed to bind hdmi, alsactl failed to start, failed to reserve cma or something | 14:58 |
fury | it booted up one time and showed me an empty home screen, now every boot is just a blank screen | 14:58 |
fury | i can ssh in and get that journalctl tho | 14:59 |
CTtpollard | is weston alive? | 15:01 |
*** Vamsi has quit IRC | 15:03 | |
fury | seems to be running | 15:03 |
dl9pf | fixed the rpi-sdimg in gerrit. should land in master soon. | 15:03 |
fury | dl9pf: you're the best :D | 15:03 |
fury | i either have two broken pis, or something went wrong with my image on first boot | 15:04 |
*** rstreif has joined #automotive | 15:13 | |
smurray | fury: no good ideas, sorry | 15:14 |
fury | no worries. hopefully it was just a botched first boot or something, i'll try reflashing the image | 15:15 |
fury | does it save previous logs anywhere? | 15:15 |
fury | just in case it did turn out to be a botched first boot, i'd love to have had the log from that one | 15:15 |
khem | usually this means you have some sort of race in runtime services | 15:15 |
smurray | fury: no, logs don't persist | 15:16 |
fury | ok. if it happens again i'll grab the log and see if it tells any clues | 15:16 |
khem | smurray: you mean logs dont persist over reboots ? | 15:17 |
CTtpollard | you can tell journalctl to give you the log from previous boot no? | 15:17 |
smurray | fury: I did some tinkering locally yesterday to get them to, it's bit more involved on current OE than it used to be when I last tried | 15:17 |
khem | atleast on rpi you could do that easily you have SD card to write to | 15:17 |
smurray | /var/log points to /var/volatile/log, so no worky | 15:18 |
smurray | at least by default | 15:18 |
smurray | you can change it by hand, of course | 15:18 |
khem | do you have read-only-rootfs turned on ? | 15:18 |
smurray | that's the default in the base-files recipe and fs-perms.txt | 15:19 |
smurray | at least afaict | 15:19 |
khem | you can remove that from IMAGE_FEATURES | 15:19 |
smurray | I don't believe AGL enables it, though | 15:21 |
fury | dd status=progress is kinda useless on a machine with so much ram that it caches even the 4gb sdimg in ram. it just kinda jumps straight up to 3.8 gb and then freezes up for a few minutes | 15:21 |
fury | wish i could have it tell me the actual progress on getting it written to disk, rather than the progress to the cache | 15:22 |
smurray | use direct mode? | 15:23 |
smurray | fury: "status=progress oflag=direct" try | 15:24 |
fury | there's a direct mode? sweet | 15:24 |
smurray | s/try// | 15:24 |
fury | awwww yeaah | 15:24 |
smurray | I think it's mentioned in the AGL rpi docs, actually ;) | 15:24 |
fury | if i'd rtfm once in a while i'd know things :D | 15:25 |
fury | is there a way to shoehorn the nav data and other files into the image before i dd? | 15:25 |
fury | or would i need to cram those into the rootfs tar before it creates it | 15:25 |
khem | fury: you can use dcfldd | 15:25 |
khem | or use pv to view data transfer between two processes | 15:26 |
fury | ok yep happened again this boot | 15:28 |
fury | blank home screen | 15:28 |
fury | what logs do i need? just journalctl? | 15:28 |
fury | Jun 27 23:29:45 raspberrypi3 afm-system-daemon[476]: NOTICE: -- INSTALLING widget /usr/AGL/apps/hvac.wgt to /var/local/lib/afm/applications -- [/usr/src/debug/af-main/1.0-r0/git/src/wgtpkg-install.c:420] | 15:29 |
fury | Jun 27 23:29:45 raspberrypi3 afm-system-daemon[476]: ERROR: error in creation of workdir /var/local/lib/afm/applications/TMP0: Inappropriate ioctl for device [/usr/src/debug/af-main/1.0-r0/git/src/wgtpkg-workdir.c:141] | 15:29 |
fury | Jun 27 23:29:45 raspberrypi3 afm-system-daemon[476]: ERROR: failed to create a working directory [/usr/src/debug/af-main/1.0-r0/git/src/wgtpkg-install.c:425] | 15:29 |
fury | interesting... happening for every app | 15:30 |
dl9pf | fury: oh ... latest master ? | 15:31 |
fury | yep | 15:31 |
dl9pf | jobol: ^^^ | 15:31 |
fury | well, pretty recent anyway | 15:31 |
fury | last night at 8:30pm eastern | 15:31 |
*** jlrmagnus has quit IRC | 15:31 | |
jobol | hum | 15:31 |
*** jlrmagnus has joined #automotive | 15:32 | |
jobol | fury, I don't think that latest master changes how /var/local/lib/afm/applications/TMP0 is created or not. | 15:36 |
fury | no space left on device | 15:38 |
fury | sigh | 15:38 |
fury | time to edit how much extra space i give the image :D | 15:39 |
jobol | not lucky! | 15:39 |
fury | root@raspberrypi3:/var/local/lib/afm/applications# su afm | 15:39 |
fury | afm@raspberrypi3:~/applications$ mkdir TMP0 | 15:39 |
fury | mkdir: cannot create directory 'TMP0': No space left on device | 15:39 |
fury | why couldn't it have just said that in the first place :P | 15:39 |
jobol | I often get Inappropriate ioctl for device but look at [/usr/src/debug/af-main/1.0-r0/git/src/wgtpkg-workdir.c:141] | 15:40 |
jobol | if (!mkdirat(AT_FDCWD, workdir, dirmode)) | 15:40 |
jobol | break; | 15:40 |
jobol | if (errno != EEXIST) { | 15:40 |
jobol | ERROR("error in creation of workdir %s: %m", workdir); | 15:40 |
jobol | return -1; | 15:40 |
jobol | } | 15:40 |
jobol | how buggy is %m? Or do I change errno? ... | 15:41 |
fury | i guess it's mkdirat's fault then | 15:41 |
fury | for setting the wrong errno | 15:42 |
fury | mkdir's supposed to return ENOSPC, maybe %m is using the wrong thing? | 15:43 |
jobol | I think that it comes from %m | 15:43 |
fury | that error message is ENOTTY not ENOSPC | 15:52 |
fury | mkdir isn't supposed to return ENOTTY | 15:53 |
fury | O_o | 15:53 |
fury | at least according to https://linux.die.net/man/2/mkdir | 15:53 |
*** waltminer has quit IRC | 15:56 | |
*** jlrmagnus has quit IRC | 15:59 | |
jobol | fury, found | 15:59 |
jobol | hahaha I'm the madness | 16:00 |
fury | wahappen? | 16:01 |
jobol | int tty = isatty(fileno(stderr)); | 16:01 |
jobol | alters the errno | 16:02 |
fury | ooh | 16:02 |
fury | mkdirat doesn't overwrite it? | 16:02 |
*** [AD]Turbo has quit IRC | 16:04 | |
fury | YES. serial port works on 4.4 | 16:05 |
fury | hot damn | 16:05 |
fury | brb going to send a PSA to the mailing list | 16:05 |
*** fredw has quit IRC | 16:11 | |
*** ZenEntropy has quit IRC | 16:11 | |
*** waltminer has joined #automotive | 16:13 | |
*** akilaw has joined #automotive | 16:14 | |
*** kbroulik has quit IRC | 16:16 | |
*** Guest76021 has quit IRC | 16:21 | |
*** jlrmagnus has joined #automotive | 16:24 | |
*** jacobo has quit IRC | 16:26 | |
*** AlisonChaiken has quit IRC | 16:26 | |
*** jacobo has joined #automotive | 16:34 | |
*** martinkelly has quit IRC | 16:35 | |
*** martinkelly has joined #automotive | 16:36 | |
*** claneys has quit IRC | 16:39 | |
mranostay | dl9pf: ping | 16:42 |
dl9pf | pong | 16:42 |
mranostay | dl9pf: so we support mp3? isn't there some possible patent issues with that? | 16:43 |
mranostay | smurray: ah so all the patents have expired? | 16:44 |
smurray | mranostay: last patent expired earlier this year, but IANAL | 16:44 |
smurray | mranostay: Fedora has finally started shipping mp3 support, so I take that as a sign | 16:45 |
mranostay | ah ok. just wanted to be sure | 16:45 |
smurray | mranostay: for reasons unknown, the renesas BSPs include the gstreamer plugin for mp3. iirc, it's one of the reasons that pre-built images are problematic for them | 16:47 |
dl9pf | yes ... but IANAL | 16:47 |
martinkelly | sadly H.264/H.265 are still extremely patented i believe :( | 16:50 |
smurray | afaik they've been reasonably good about having the licensing only kick in only from commercial products, at least | 16:52 |
smurray | dominig's theory about udev somehow causing this tty issue might be right, I'm still seeing it with the vast majority of services pruned out | 16:53 |
*** fredw has joined #automotive | 16:55 | |
dl9pf | would explain the time delay .... udev trigger + settle takes some time | 16:57 |
smurray | I'd have thought there'd be something more visible in the way of errors | 16:57 |
jobol | dl9pf, Can we switch to branch=${BRANCH} or to nobranch=1 in recipes ? | 17:02 |
*** jlrmagnus has quit IRC | 17:02 | |
dl9pf | in which recipes ? | 17:02 |
jobol | in af-main should I put branch=dab for branch dab? | 17:03 |
dl9pf | yes, | 17:03 |
dl9pf | I ponder if we should do something like | 17:05 |
dl9pf | BRANCH ?= "${DISTRO_CODENAME}" | 17:05 |
kooltux_ | we should :) | 17:05 |
jobol | I ponder too | 17:05 |
dl9pf | or DISTRO_BRANCH (so we can still set it to master or master) | 17:05 |
dl9pf | or DISTRO_BRANCH (so we can still set it to master for master) | 17:05 |
jobol | AGLBRANH | 17:06 |
jobol | AGLBRANCH | 17:06 |
dl9pf | jobol: set it to dab in your recipe , ... I'll see if we can do the branch=${DISTRO_BRANCH} | 17:06 |
*** jacobo has quit IRC | 17:07 | |
jobol | i'll do you'll see the review | 17:08 |
khem | mp3 is patent-free right | 17:13 |
khem | since ~april | 17:15 |
mdp | mranostay: I have the issue assigned to me and specifically asked waltminer about this in the comments to confirm that LF is ok with hosting mp3 now given the change. | 17:15 |
*** AlisonChaiken has joined #automotive | 17:15 | |
mdp | mranostay: no response yet | 17:15 |
mranostay | ok | 17:15 |
khem | long live napster | 17:15 |
mdp | since IANAL and it's not my hosting/distribution legal burden, but rather LF's...LF should answer. | 17:16 |
mranostay | khem: kazaa forever man :) | 17:16 |
mdp | lol | 17:16 |
khem | yaeah but I think it was napster who made mp3 household name | 17:17 |
*** leon-anavi has quit IRC | 17:17 | |
*** jlrmagnus has joined #automotive | 17:18 | |
mdp | https://jira.automotivelinux.org/browse/SPEC-312, fwiw | 17:18 |
mdp | and gee, thanks, fury ;) | 17:18 |
fury | wahappen | 17:18 |
fury | oh | 17:18 |
fury | :P | 17:18 |
mdp | it was your issue! | 17:18 |
mdp | :P | 17:18 |
* fury hides | 17:19 | |
mdp | fury, btw, did you ever report that onboard codec audio issue on porter? I went looking and didn't notice an issue filed. | 17:22 |
fury | yeah | 17:22 |
waltminer | I forgot about that issue. Will look into it | 17:22 |
fury | https://jira.automotivelinux.org/browse/SPEC-602?jql=text%20~%20%22porter%20audio%22 | 17:23 |
mdp | fury, I was doing some testing this past week and hearing well-known tracks with that terrible slow rate was driving me nuts. | 17:23 |
fury | i thought it was an issue with bluetooth at first but it cleared up by plugging in a USB sound card and using that as the output instead | 17:23 |
smurray | waltminer: no joy with the Hilton, I guess? | 17:24 |
mdp | yup, it's specific to the onboard codec and independent of stream source. | 17:24 |
mdp | BSP issue | 17:24 |
waltminer | Amy should be contacting you directly. I gave her your email address. The hotel was having some kind of problem adding the extra day to the reservation. Not sure what. | 17:24 |
waltminer | Same for khilman | 17:25 |
mdp | fury, I don't think issues get from AGL-land to an appropriate renesas BSP maintainer though | 17:25 |
waltminer | It is Hyatt not Hilton | 17:25 |
fury | i switched my qml app to AGL style and all my vertical sliders now show up as really tiny horizontal ones :o | 17:25 |
smurray | waltminer: okay, thanks. Too many H-hotels ;) | 17:26 |
smurray | fury: there's some AGL specific Qt theming, not sure if you need to do anything to pick that up | 17:26 |
fury | yeah, i set QQuickStyle::setStyle("AGL") in my main.cpp and everything picked up the borderless, AGL-themed background, but my sliders are all horizontal | 17:27 |
fury | and i think they even reversed | 17:27 |
fury | weird | 17:27 |
smurray | fury: mdp might be you man there, his mixer app has many sliders | 17:28 |
fury | all horizontal tho :P | 17:28 |
*** jobol has quit IRC | 17:29 | |
*** araujo has quit IRC | 17:29 | |
mdp | fury, well, let's keep in mind that this is a _Demo HMI_. | 17:33 |
*** araujo has joined #automotive | 17:33 | |
*** araujo has joined #automotive | 17:33 | |
mdp | I'm just a user of it but I can tell you that the AGL qt quick controls 2 agl style slider template is hardcoded that way. | 17:34 |
fury | i know :D i think i see what file i'd edit to add that back in | 17:34 |
*** toscalix has quit IRC | 17:35 | |
mdp | so when they created the template to adjust the "groove" styling etc...they just fixed it to the horizontal need expected to be used in portrait mode. | 17:35 |
mdp | "patches welcome" | 17:35 |
fury | fo sho | 17:36 |
fury | https://gerrit.automotivelinux.org/gerrit/gitweb?p=src/qtquickcontrols2-agl-style.git;a=blob;f=imports/qtquickcontrols2aglstyle/Slider.qml;h=6c9863b3a0df5511f3e4983f8381a2d82ec282ec;hb=a78db8638590c797c78c36b2ef3de5ed9c012bba yep | 17:36 |
fury | ok so i clone that repo, make that change, and then do "git review"? | 17:37 |
mdp | fury, fwiw, some styling stuff is difficult on sliders period in qtquickcontrols2 as some of the stuff in the manuals will fail to work as documented if you override styles..I see that (and others too) in my own non-agl projects. | 17:37 |
fury | oh | 17:37 |
mdp | just some limitations as qml components continually evolve | 17:37 |
fury | yep, i'm pretty newbish at customizing qml | 17:38 |
mdp | like, if you use material style like I do on other projects...and then you want to override the style stuff..you break material style colors selection as it's an all or nothing thing. | 17:38 |
fury | thanks for the heads up tho :D | 17:38 |
*** fredw has quit IRC | 17:38 | |
fury | i love this stuff | 17:38 |
*** jlrmagnus has quit IRC | 17:38 | |
mdp | heh, you are full of energy..I could substitute you for my 3 espressos ;) | 17:39 |
fury | i'm changing lines of text and compiling an application and it's running on a little $35 computer. BLOWS MY MIND | 17:39 |
fury | and it comes up on a 15" touchscreen and is...well, not very smoothly animated | 17:39 |
fury | but $35 computer! | 17:39 |
mdp | waltminer: thanks, I forgot too until I saw the above conversation..thanks. didn't feel comfortable pushing something with an ack from LF. | 17:39 |
fury | and it's making it do things | 17:39 |
mdp | s/with/without | 17:39 |
mdp | fury, wrt the file..yup, that's the AGL slider component template. modify away. | 17:41 |
fury | hmm, wonder if i can overclock the pi enough to make it fast :P | 17:44 |
mdp | you should spend a little more on a better "tiny computer" :) | 17:45 |
fury | yeah, ordering a minnowboard turbot quad core as we speak :P | 17:45 |
mdp | best part is you don't have to worry about poor onboard audio | 17:50 |
fury | :D | 17:50 |
fury | ordered two, cause why not | 18:02 |
*** waltminer has quit IRC | 18:03 | |
*** jlrmagnus has joined #automotive | 18:09 | |
akilaw | Can you guys see the pr https://github.com/GENIVI/genivi-dev-platform/pull/97 | 18:13 |
akilaw | I have fixed the issues | 18:13 |
rZr | akilaw, sorry I wont have time today | 18:23 |
akilaw | ok rZr | 18:24 |
*** jonte_ has joined #automotive | 18:38 | |
*** jonte__ has quit IRC | 18:42 | |
*** _corrupt has quit IRC | 19:08 | |
*** _corrupt has joined #automotive | 19:09 | |
*** akilaw has quit IRC | 19:16 | |
*** waltminer has joined #automotive | 19:21 | |
*** caiortp has quit IRC | 20:52 | |
*** iSaul has quit IRC | 20:55 | |
*** iSaul has joined #automotive | 21:35 | |
*** iSaul has joined #automotive | 21:45 | |
waltminer | smurray thanks for the updates in SPEC-544 keep plugging | 21:46 |
fury | the app fw call, is that 8am eastern time or 9am? my calendar has it at both times O_o lol | 21:46 |
*** iSaul has joined #automotive | 21:47 | |
waltminer | 9 am EDT. 8 am Chicago time | 21:47 |
fury | ah ok | 21:47 |
waltminer | most calls are at 8 am Chicago time except CIAT | 21:48 |
fury | hmm... my outlook is telling me the last one i received is every 3 weeks from 7am to 8am central time | 21:50 |
fury | i'm trying to open the .ics and Outlook is telling me "The program used to create this object is Outlook. That program is not installed on your computer." | 21:50 |
fury | :D | 21:50 |
fury | oh well | 21:52 |
fury | deleted | 21:52 |
*** martinkelly has quit IRC | 21:57 | |
*** martinkelly has joined #automotive | 21:59 | |
*** iSaul has quit IRC | 22:22 | |
*** jlrmagnus has quit IRC | 22:59 | |
*** iSaul has joined #automotive | 23:07 | |
*** iSaul has joined #automotive | 23:07 | |
*** kholdfuzion has joined #automotive | 23:08 | |
*** jlrmagnus has joined #automotive | 23:22 | |
fury | is there a way to afm-util kill app_name@0.1? | 23:41 |
fury | i have hackjobbed almost all of the XDS build steps into my Qt project, the last one is killing the current process and reinstalling the widget | 23:42 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!