| kalube | Does OBS pick which spec file to use based on the package name? | 00:30 |
|---|---|---|
| r0kk3rz | yeah | 00:31 |
| kalube | Sweet | 00:33 |
| kalube | hmmm I seem to have broken something here https://build.sailfishos.org/project/monitor/nemo:testing:hw:oneplus:enchilada | 02:06 |
| kalube | I think droid-config-enchilada is the actual source of the issue? I'm not sure what provides pkgconfig(android-headers) though | 02:06 |
| r0kk3rz | hybris does | 02:26 |
| kalube | libhybris/? | 02:33 |
| kalube | It wants droid-hal-devel | 02:33 |
| r0kk3rz | yeah so libhybris is complaining about your hal package | 02:33 |
| r0kk3rz | so something is not right there | 02:33 |
| rinigus | kalube: check out Tama obs and ci, it has 6 devices behind it. At obs, also note advanced settings as they were required | 05:08 |
| kalube | rinigus thank you I'll take a look. What's in advanced settings that might be important? Sorry I'm a little new to OBS... I did have everything build before but it broke after I reorganised my spec files | 05:09 |
| rinigus | kalube: there are not many over there. But mainly preference for packages when multiple provide the same. Although, note that Sony devices in a family are quite close, don't know about your case | 05:32 |
| kalube | in my case the only differences that affect sailfish are the model names, and potentially fingerprint HAL but | 05:33 |
| rinigus | kalube: then it should work nicely. You may want to use the same ci scripts as I have. Those download ks from obs for each device during build | 05:59 |
| rinigus | piggz: do you have Android system and vendor also packaged as rpm and installed into your root? | 06:09 |
| rinigus | As for files, the one going to system_b is for restoring device. You don't need it for boot and can use system partitions to extend lvm | 06:10 |
| piggz | rinigus: no, thats what i was wondering about | 06:11 |
| T42 | Santosh %lastname% was added by: Santosh %lastname% | 06:11 |
| T42 | <Santosh %lastname%> 0949 | 06:11 |
| piggz | rinigus: as your image is a lot larger | 06:12 |
| rinigus | piggz: yeah, don't know if you can do it that way for license reasons. Sony aosp allows it, others not sure. | 06:13 |
| rinigus | I wonder how to get around it... How do you usually boot? Does it use system partitions as root? | 06:14 |
| piggz | rinigus: im happy with the system_b install, its overwriting/vendor im not sure about! | 06:14 |
| rinigus | piggz: we have also OEM on Sony, vendor can be reflashed | 06:16 |
| piggz | no, just the old fashioned way, /data/.stowaways/sailfish get mounted as / | 06:16 |
| piggz | i think it should probably work so long as i dont flash vendor | 06:17 |
| piggz | wouldnt mind mal's input too tho | 06:17 |
| rinigus | Let's see what wisdom mal can share this time | 06:17 |
| piggz | ultimately, i could probably incude system and vendor as the oem is wanting to ship sailfish anyway | 06:18 |
| piggz | how to you package those on your build? | 06:18 |
| rinigus | piggz: Sony hadk has separate section regarding it. See one for Xperia 10 or short version on Tama main repo | 06:19 |
| piggz | k, will browse later | 06:19 |
| rinigus | piggz: after compile on pc, I uploaded rpms to obs into separate repo under testing:tama | 06:20 |
| rinigus | But it would depend on whether all in system and vendor are open source. Sony has all binary bits separately under OEM. But if they want to ship sfos, they can also host small server providing repo with those rpms. That way you could combine obs and that extra repo to build images | 06:23 |
| rinigus | piggz ^ | 06:23 |
| rinigus | piggz: all this fp support becomes real mess... Issues seem to come daily now :) | 09:37 |
| r0kk3rz | sounds fun | 10:24 |
| piggz | rinigus: that last one he sorted it himsef didnt he? | 10:26 |
| piggz | flashing persist | 10:26 |
| T42 | <adampigg> oh there is some more too | 10:29 |
| T42 | <adampigg> :D | 10:29 |
| T42 | Simon %lastname% was added by: Simon %lastname% | 12:10 |
| T42 | <Simon %lastname%> 3608 | 12:11 |
| kalube | @piggz do you have any idea what might be on persist that FP uses? In my port as part of the installation process I disable persist mounting and copy the persist partition in the folder in rootfs instead. That's because sailfish breaks selinux contexts on files which prevents android from being able to access them properly in a dual boot | 13:03 |
| kalube | rinigus: I see you specified rpm_device too https://build.sailfishos.org/package/view_file/nemo:devel:hw:sony:tama/droid-config-h8216/_service:tar_git:droid-config-h8216.spec?expand=1 was there a reason for that? | 13:07 |
| T42 | Анна %lastname% was added by: Анна %lastname% | 13:15 |
| kalube | hmm replacing droid-hal-enchilada rpm and forcing droid-hal-device to rebuild seems to have fixed it | 13:15 |
| T42 | <Анна %lastname%> 4747 | 13:15 |
| piggz | kalube: i used rpm_device on pro1 because the early dev units had an android devive name of t5, and the release model was QX1000, by specifying rpm_device as QX1000 and device as t5 we could keep the same packages for both | 14:25 |
| piggz | kalube: no idea, each device has stores fingerprint data differently | 14:26 |
| kalube | Thanks | 14:26 |
| piggz | s/hal | 14:27 |
| kalube | droid-hal-version-enchilada is failing with some odd error running zic? I've seen the same error on builds that have succeeded before too... https://build.sailfishos.org/package/live_build_log/nemo:testing:hw:oneplus:enchilada/droid-hal-version-enchilada/sailfishos_3.3.0.16/armv8el | 14:56 |
| kalube | I don't think that's the cause... | 14:57 |
| T42 | <elros34> real error is at the beginning: remove your custom changes from droid-hal-device inc | 14:59 |
| kalube | @elros34 do you mean the breaking dependency cycle errors? the build has only broken since adding fajita packages, https://github.com/sailfish-oneplus6/droid-hal-device/commits/8316b26fb9bfcc0b6f0e54b528198dc3afb4f8da/droid-hal-device.inc do you think one of these caused the issue? | 15:06 |
| T42 | <elros34> I think it's because of: "file /boot from install of filesystem-3.1+git3-1.4.1.jolla.noarch conflicts with file from package droid-hal-enchilada". filesystem package in common repo has already fix for /boot permission issue | 15:09 |
| kalube | Ah ok, so my latest dhd commit isn't needed, thanks | 15:09 |
| rinigus | piggz: somehow, flashing persist does not sound right. but it seems to me that with fp issues, porters would have to work around the quirks | 15:10 |
| T42 | <elros34> ahh fajita is your second device. I didn't get it. So maybe it's something else | 15:10 |
| kalube | oh? | 15:10 |
| piggz | rinigus: yes, i think if a port requires quirks, we need to porters to contribute the code for those quirks .... all we have to go on is the names of the official quirk tho | 15:11 |
| kalube | It may well be, that fix for /boot was needed to build locally | 15:11 |
| piggz | i mean, we could do some ourself, but we couldnt test | 15:11 |
| kalube | piggz: I'm more than happy to test stuff! | 15:12 |
| rinigus | kalube: on sony, we havce code names and device ids. I took that part from the official ports and its notation used there. so, in your example, we have codename akari and model h8216 (other device with the same codename would be dual sim with the other model) | 15:12 |
| rinigus | kalube: we are happy to accept PRs with working fp fixes :) | 15:13 |
| rinigus | piggz: I agree, I think it is sensible way | 15:14 |
| kalube | rinigus: I see, it seems like there are a few ways to do it then... I think this build failure is probably the last hurdle... | 15:15 |
| kalube | I'd love to make PRs for FP but it might be a little too much for me to get my head around... | 15:16 |
| rinigus | kalube: in practice, you probably have to see how it is done on some android port. at least that is a start | 15:18 |
| kalube | Once I get CI and OTAs working I'll start properly looking into it, gotta get this port DONE :D | 15:20 |
| rinigus | kalube: good priorities! | 15:22 |
| kalube | Will android HAL be affected by SELinux context in hybris? As selinux is disabled it won't be right? | 15:25 |
| kalube | rinigus: "have choice for pkgconfig(android-headers): droid-hal-enchilada-devel droid-hal-fajita-devel", in droid-config-X, checking tama shows you just require droid-hal, how does it pick which to use for android-headers? | 15:34 |
| rinigus | kalube: preference is as in https://build.merproject.org/project/prjconf/nemo:devel:hw:sony:tama | 15:35 |
| kalube | Ah I see, and because the devices are so similar it doesn't matter to prioritise one? | 15:36 |
| kalube | Oh right it's devel so it really doesn't matter | 15:37 |
| rinigus | kalube: same is for testing. devices are very similar and you can compile other packages with any of them, I presume | 15:39 |
| kalube | Should be ok then :D | 15:42 |
| kalube | All building now, finally! | 15:42 |
| kalube | r0kk3rz: sorry to bother, can you make me the maintainer of https://gitlab.com/sailfishos-porters-ci/enchilada-ci/-/branches so I can edit repo settings? | 15:52 |
| kalube | rinigus: thanks for all your help, finally made it :D https://gitlab.com/sailfishos-porters-ci/enchilada-ci/-/pipelines/178429243 | 16:16 |
| rinigus | kalube , you're welcome | 16:24 |
| T42 | <elros34> @kabouik I was able to enable cursor in qxcompositor but as you said mouse coords are correct only in portrait. I don't know where this should be fixed, maybe in lipstick | 21:41 |
| T42 | <kabouik> And was the cursor position updated at all times or just when calling an event @elros34? | 21:43 |
| T42 | <elros34> cursor is visible as soon as you move mouse and hide when you touch screen | 21:43 |
| T42 | <kabouik> I believe it would be acceptable to hardcode a rotation of coordinates if doable, since I don't think any user would use an external mouse in portrait. But it's ugly, and not sure how to do it anyway | 21:43 |
| T42 | <kabouik> Nice! | 21:44 |
| T42 | <elros34> maybe rotation option in evdevtouch could be easily added | 22:08 |
| T42 | <kabouik> I tried to add the option but it didn't work, but yeah it probably just is not implemented | 22:09 |
| T42 | <elros34> yes, such a option is not parsed in code | 22:17 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!