| T42 | <adampigg> @ankaos [Call fix. But call sound not fix], Whats wrong with call sound? | 04:47 |
|---|---|---|
| T42 | <ankaos> Send logcat log | 04:47 |
| T42 | 司马智款 %lastname% was added by: 司马智款 %lastname% | 09:13 |
| Mister_Magister | mal: will you be doing this reconfiguration thingy in gst-droid? | 11:43 |
| mal | maybe at some point | 11:44 |
| Mister_Magister | also mal r0kk3rz abranson https://github.com/r0kk3rz/gstdroid-player/blob/master/src/renderernemo.cpp#L127 this eglGetCurrentDisplay seems to not work. Interestingly enough on r0kk3rz's phone videos do play even with that error but here they dont | 11:45 |
| Mister_Magister | mal: so nowhere soon, ok | 11:45 |
| T42 | <adampigg> Erm whaaaaat? Obs is going? ??? | 13:05 |
| rinigus | @adampigg: good morning to you too | 13:08 |
| rinigus | :) | 13:08 |
| T42 | <adampigg> Rinigus: i was on holiday for few hours and this happens??? | 13:09 |
| T42 | <adampigg> (Photo, 2560x1920) https://irc.thaodan.de/.imgstore/nvY16I1ZyS.png | 13:09 |
| rinigus | @adampigg: pretty much all is collapsing over here | 13:10 |
| rinigus | ... now just imagine the news tomorrow, after your holiday has clocked 24h | 13:10 |
| T42 | <adampigg> I dont get it, obs does so much | 13:10 |
| rinigus | voice your concerns over here and on forum | 13:11 |
| T42 | <adampigg> Do they.want us to stop porting? | 13:11 |
| T42 | <adampigg> Mal, what gives? :( | 13:12 |
| Mister_Magister | @adampigg money | 13:12 |
| rinigus | hard to say what they want. cost maybe a factor, but I am waiting now for response to the corresponding questions in forum | 13:13 |
| Mister_Magister | it all comes to money | 13:13 |
| T42 | <kabouik> Please share the links when there is a thread about it so users can voice concerns too | 13:13 |
| rinigus | Kabouik: the discussion is in the announcement thread already | 13:13 |
| rinigus | iff it is about money (lack of personnel is money as well) then question is how much does it cost per month? | 13:15 |
| rinigus | if they start reducing costs for some small amounts, it all either tells that sfos boat is on a proper rocks or Jolla mngmt doesn't really care about community devs | 13:16 |
| vknecht | another question could be: can the whole obs config/setup be shared so community can setup its own instance ? | 13:16 |
| mal | running a service is not just the server cost, also people need the maintain it | 13:16 |
| rinigus | mal: maintaining is a part of a cost, probably the main one | 13:17 |
| rinigus | (I mean costs for admin salary) | 13:18 |
| T42 | <adampigg> Is there any alternative proposed? | 13:19 |
| rinigus | @adampigg: not in https://forum.sailfishos.org/t/changes-needed-to-merge-the-project-names-to-sailfish-os/1672 | 13:20 |
| rinigus | but I wonder if we can move all to suse obs | 13:20 |
| mal | there is PR to enable using self-hosted repos in ports | 13:21 |
| T42 | <adampigg> Yeah,.or something like gitlab ci? | 13:21 |
| rinigus | you need to have rpm repo for hosting. otherwise ota will not work | 13:22 |
| rinigus | so, doesn't look like gitlab ci is good for it. unless I am missing something | 13:22 |
| rinigus | mal: self-hosting is part of equation. building is another | 13:22 |
| mal | rpm repo is a simple thing, it's self-contained | 13:22 |
| mal | just a folder with proper content | 13:23 |
| mal | yes, building adaptations is a separate thing | 13:23 |
| T42 | <adampigg> Yeah...we'd have to fund it ourself i guess...i have a vps, but its the cheapest of cheap, build may take ages | 13:23 |
| rinigus | mal: hosting on a server somewhere. so, assuming it is going to happen - we are going now to be responsible for shipping those rpms as well | 13:24 |
| rinigus | mal: as for building, it is not just ports. my maps apps have 17 packages in corresponding projects. so, on every sfos update, I have to figure out how to rebuild it all | 13:25 |
| rinigus | so, apps are suffering as well | 13:26 |
| rinigus | btw, all is presented as "Changes needed to merge the project names to Sailfish OS" | 13:27 |
| mal | rinigus: full rebuild is actually not that complicated thing to do, just a new target to platform sdk, then something like a script to run needed commands and copy rpms to local repo like build_packages.sh does, then copying rpms from there | 13:27 |
| rinigus | (side comment regarding amazing PR skills, sorry for outburst) | 13:28 |
| rinigus | mal: sure, all these rebuilds and starting new projects are doable. we will just start wasting our time on something that was way better before | 13:28 |
| rinigus | (read now) | 13:29 |
| rinigus | I think some proper replacement would be needed. and question is what are our options? | 13:30 |
| rinigus | we are way better as a community if we have a build service and repo distribution scheme we can rely upon | 13:30 |
| T42 | <adampigg> I fear people just wont bother | 13:32 |
| T42 | <kabouik> ^ | 13:33 |
| rinigus | @adampigg: please expand/explain - 'wont bother'? | 13:35 |
| vknecht | switch to mainline linux, and native nemo ? :-) | 13:36 |
| T42 | <adampigg> Means too much effort,.so wont do it | 13:37 |
| rinigus | @adampigg: on local PCs? | 13:37 |
| rinigus | vknecht: it surely makes you consider the options, that's clear. as any significant change that hinders the progress | 13:39 |
| T42 | <adampigg> Rinigus: too much effort to have 1) all the build targets 2) host packages | 13:39 |
| T42 | <adampigg> Yes, its.much more effort on local pc's | 13:39 |
| rinigus | @adampigg: I agree | 13:40 |
| rinigus | so, let's see if we can come up with some proper alternative. assuming that they will pull the plug | 13:40 |
| T42 | <eugenio_g7> Real bummer 😥 | 13:40 |
| T42 | <eugenio_g7> I think the biggest issue is where to host repositories, rather than how to build (since that both gitlab and github provide enough resources for foss projects) | 13:42 |
| T42 | <adampigg> Will jolla still build and host the common and native-common repos? | 13:42 |
| T42 | <adampigg> Could we make use of any other obs providers? | 13:44 |
| T42 | <eugenio_g7> And this is also bad news for nemo, not only sfos ports :| | 13:45 |
| rinigus | @adampigg: maybe we can use openSUSE OBS. don't know what fedora is using | 13:51 |
| rinigus | but we need to make a list and start looking into it. again, assuming that it is happening and they don't decide to continue with obs | 13:52 |
| T42 | <adampigg> Maybe its as simple as adding the tatgets to suse obs, not sure | 13:52 |
| rinigus | maybe. | 13:54 |
| rinigus | question is how to pull sailfish proprietary bits into these targets | 13:54 |
| mal | rinigus: what do you mean? | 13:55 |
| rinigus | mal: I assume we cannot fully build apps and adaptations without sailfish bits that are distributed through Jolla's repos | 13:56 |
| rinigus | (stress on "assume") | 13:56 |
| mal | do you mean that system cannot download those automatically when needed? | 13:57 |
| T42 | <adampigg> The way i saw things, sailfishos ports increased the jolla ecosystem, in return for that, jolla faciltated the ports by providing services such as obs | 13:58 |
| mal | I know that OBS doesn't allow network access during build stage but does it still fetch those when setting up dependencies | 13:58 |
| rinigus | mal: I don't know OBS enough to state it with the confidence. from examples I have seen so far, OBS repos added as dependencies are internal to that OBS | 13:59 |
| rinigus | not sure I can add as a repo something like https://... | 13:59 |
| mal | rinigus: I wasn't talking about dependencies like that, I mean in the target | 14:00 |
| rinigus | mal: in the lines of config like <path project="sailfishos:latest" repository="latest_i486"/> | 14:00 |
| rinigus | @adampigg: that is probably true. also leading to initiative of users of the port to develop native apps | 14:01 |
| rinigus | mal: not sure if I understand "target" - getting rusty with the terms. but does my concern regarding project dependency make sense? | 14:02 |
| mal | rinigus: we are talking about different level of target | 14:03 |
| mal | rinigus: target as in the base target, like this and subprojects https://build.sailfishos.org/project/show/sailfishos:3.3.0.16 | 14:03 |
| mal | that is initialized other ways | 14:04 |
| mal | by the maintainer of OBS | 14:04 |
| T42 | <adampigg> Mal: so, what would your solution be for example.fp2 port packages? | 14:04 |
| mal | I can always host those on my server, I have 100/100 MB connection without any transfer quota and enough space | 14:05 |
| T42 | <adampigg> Yes, so we would need obs maintainer to add the sfos targets | 14:05 |
| T42 | <adampigg> And you would just build locally? | 14:06 |
| mal | I probably would setup some local build system for automated builds | 14:06 |
| rinigus | mal: if you say that we need maintainer to initialize OBS, that is not going to happen on openSUSE | 14:06 |
| mal | rinigus: I have no idea how openSUSE OBS handles setting up distro targets there | 14:07 |
| rinigus | mal: me neither. | 14:07 |
| mal | or would they even want to host a partially closed-source distro targets | 14:08 |
| T42 | <adampigg> For myself,.i have a laptop, and i do limited builds locally, so will.be quite a disruption | 14:08 |
| mal | @adampigg I have machine mostly for building, I updated the hardware to make local builds faster | 14:08 |
| rinigus | mal: but if we take NeoChapay nemo setup as an example. I don't think he is using any sfos targets, but compiles all in place | 14:09 |
| rinigus | with dependencies propagating nicely | 14:09 |
| rinigus | side note - would be major distraction for me as well to start building it locally | 14:09 |
| rinigus | mal: how sure is that obs shutdown? | 14:10 |
| mal | @adampigg my build machine is quad-core i5 with 16 GB of RAM and several terabytes of disk space | 14:10 |
| kalube | There's a lot of stuff that seems everyone needs to build themselves on OBS, that could maybe be more common | 14:11 |
| kalube | There's also the fact that op5 and 6 for example have a lot of common packages that we both need to build, like touch screen gesture daemon and tri-state switch stuff | 14:11 |
| T42 | <adampigg> Mal, contrast that with 8gb/1tb core i5 | 14:11 |
| kalube | You guys should check out how postmarketOS manage their stuff, perhaps a lot of it could translate over | 14:12 |
| kalube | like not having individual repos per device | 14:12 |
| mal | kalube: do they have some build system | 14:12 |
| kalube | mal: postmarketOS are alpine linux based so they have a single repo and a pmaports repo on gitlab, all pmaports packages get built and hosted by them | 14:13 |
| kalube | pmaports include device kernels and every part of their system | 14:13 |
| kalube | They used to host halium stuff too but they don't anymore | 14:14 |
| kalube | in sailfish the kernel and such are just chucked in with droid-hal most of the time where they could be a separate package. | 14:14 |
| kalube | http://build.postmarketos.org/ | 14:17 |
| kalube | imo gitlab CI + a central repo makes sense because anyone can fork the repo, add packages for their device and go through CI, style checks etc before getting accepted into the repo. | 14:18 |
| T42 | <adampigg> I see one outcome, inconsistency | 14:18 |
| kalube | @adampigg how do you mean? | 14:18 |
| T42 | <adampigg> Porters doing things differently | 14:20 |
| rinigus | I presume that is regarding local builds and distributing from garage | 14:20 |
| T42 | <adampigg> Yes....hopefully we can.come up with something.consistent | 14:22 |
| kalube | Surely consistency would come with style checking and manual review? | 14:22 |
| T42 | <adampigg> I mean consistency in build and.distibution method | 14:23 |
| T42 | <adampigg> We are already looking at options for CI, but mal content with building and hosting locally | 14:24 |
| kalube | Ah right | 14:25 |
| mal | @adampigg considering that I haven't even used the community image build system yet | 14:26 |
| mal | I have shared the images from my own server | 14:26 |
| Mister_Magister | about loosing OTA https://forum.sailfishos.org/t/changes-needed-to-merge-the-project-names-to-sailfish-os/1672/11?u=mister_magister | 14:28 |
| Mister_Magister | openrepos for rescue | 14:28 |
| T42 | <adampigg> Maybe it will be sorted by the time my holiday is finished,.so i wont have to worry about it :D | 14:36 |
| Mister_Magister | @adampigg hosting ota is resolved, building packages is separate thing | 14:38 |
| T42 | <adampigg> What is funding situation for openrepos? | 14:39 |
| T42 | <adampigg> Kabouik articulated it well | 14:41 |
| kalube | I'm pretty certain gitlab CI is free, at least for all the building sailfish would need | 14:43 |
| T42 | <adampigg> Yeah, maybe.we can build something around it with auto upload to obs... Rinigus tama port has auto upload of the artifacts iirc | 14:45 |
| kalube | That would be good, then through gitlab you get consistent packaging, packages get built and uploaded to OBS where any device can update to them | 14:47 |
| T42 | <adampigg> S/openrepos | 14:48 |
| kalube | I was looking into how the stock sailfish updates page works - it doesn't seem to hard to make it work for community devices | 14:49 |
| T42 | <kabouik> @adampigg [What is funding situation for openrepos?], Curious about that too | 14:49 |
| kalube | Would hosting on openrepos mean you lose the niceness of just using zypper and suse's existing package/dependency system? | 14:50 |
| T42 | <kabouik> Doesn't openrepos support installing dependencies already, as long as the corresponding reopsitories are enabled? This is from an single rpm point of view, it may be different for ports and lists of packages. | 14:54 |
| kalube | I don't think it makes sense to have repos per device | 14:55 |
| kalube | It should be possible to just update with zypper too right? Cuz issues will arise and not being able to resolve them in a terminal is no fun | 14:56 |
| rinigus | I still think we have to look for something as automatic as possible. let's look around and discuss. I don't see how CI + openrepos will automatically rebuild all packages that had one dependency updated, for example | 15:00 |
| Mister_Magister | r0kk3rz: any idea why playback woul be not working? i don't think the egl is reason anymore | 15:04 |
| Mister_Magister | or maybe it is idk | 15:04 |
| Mister_Magister | @kabouik hosting on openrepos doesn't have any drawbacks, you just need to upload packages | 15:05 |
| Mister_Magister | they resolve normally you have normal repos everything is same | 15:05 |
| T42 | <kabouik> ^ @kalube I presume | 15:06 |
| Mister_Magister | no, you | 15:06 |
| Mister_Magister | both of you tbh | 15:06 |
| T42 | * kabouik blushes | 15:06 |
| Mister_Magister | ye kalube | 15:07 |
| Mister_Magister | sorry | 15:07 |
| kalube | rinigus: might be worth asking the postmarketOS folks, if needed you could add a hook in CI to trigger packages that depend on the package being built to also rebuild but surely if a library gets rebuilt you wouldn't also need to rebuild the package using it, and surely there aren't that many cases of build deps where it could be an issue | 15:09 |
| Mister_Magister | on openrepos as user you have your own repo and everything you put in there will be working like obs repo | 15:09 |
| kalube | I really think it's worth moving away from a per device repo | 15:10 |
| kalube | it has no benefits that I can see | 15:10 |
| kalube | It's just a cheap way to have packages that need per-device changes work | 15:10 |
| Mister_Magister | i more and more thing it's worth moving away from sfos | 15:12 |
| Mister_Magister | because of what jolla is doing | 15:12 |
| Mister_Magister | let's switch to nemo :D | 15:12 |
| kalube | what's nemo? | 15:12 |
| Mister_Magister | but first lets do what jolla should do from beginning. switch nemo to opensuse base. if jolla just used upstream packages it would be way better | 15:12 |
| kalube | ¯\_(ツ)_/¯ I'm gonna wait for glacier UX to hit postmarketOS, get a hybris build going and go to alpine | 15:13 |
| kalube | If anything | 15:14 |
| kalube | don't see why you'd go to opensuse | 15:14 |
| Mister_Magister | glacier is ugly as shit but i could freely remake entire gui | 15:14 |
| Mister_Magister | kalube: obs | 15:14 |
| Mister_Magister | and opensuse ftw | 15:15 |
| kalube | OBS is not a very good system | 15:15 |
| kalube | Making an MR on gitlab and then just having your stuff be available is a nice system | 15:16 |
| T42 | <edp_17> Bummer, I just have completed the OBS stuff for my port and requested the "patternisation". 😳 And now I read the announcement. 😔 | 15:16 |
| Mister_Magister | kalube: imo OBS is great | 15:16 |
| kalube | @edp_17 me too, my goal was to not have to worry about sfos stuff again other than to update to new versions | 15:16 |
| kalube | Mister_Magister: your opinion isn't gonna build a good system for the community | 15:17 |
| Mister_Magister | ¯\_(ツ)_/¯ | 15:17 |
| Mister_Magister | obs worked fine so far | 15:17 |
| kalube | The porting/building process for postmarketOS compared to sailfish is night and day | 15:18 |
| kalube | OBS isn't really the problem, it's the whole process that needs improving | 15:19 |
| T42 | <edp_17> @kalube : Yeah, I wanted to be able to do OTAs and now this seems utopian anymore. I hope the guys who understand OBS better will come up with a viable option. | 15:19 |
| kalube | I will say that having to learn a whole new tool just to get obs working is not the way, if gitlab ci just pushed to obs that would be fine but having to locally build RPMs and then upload is a hack | 15:20 |
| T42 | <edp_17> Do we know a due date for this decommissioning the OBS action? | 15:20 |
| Mister_Magister | postmarket os isn't really an os | 15:21 |
| T42 | <edp_17> @kalube : I agree, if we need to upload all packages, that is like going back to the '80s. | 15:22 |
| Mister_Magister | you can have opensuse on postmarketos which is obs | 15:22 |
| kalube | Mister_Magister: I don't see your point? It's an alpine fork with a repo full of packages to make mobile devices run Linux in a usable fashion | 15:23 |
| Mister_Magister | ehh | 15:24 |
| kalube | It's a clean open system that anyone can expand upon and that's what makes it so great | 15:24 |
| Mister_Magister | on second hand android devs don't have place like obs either and they're fine. imo thanks to basil with openre\pos it will be fine just bit of pita | 15:27 |
| T42 | <kabouik> Android devs have much bigger incentive for staying on Android if they even like it, because the audience is huge | 15:28 |
| T42 | <kabouik> Also they don't have the OTA issue, do they? | 15:29 |
| kalube | For android 9+ at least, a ubports GSI style approach means a max of like 2 repos to have a device up and running - kernel repo and some device specific patches if needed | 15:30 |
| Mister_Magister | @kabouik cause they don't have ota xd | 15:30 |
| Mister_Magister | but with openrepos ota will be fine | 15:30 |
| kalube | e.g. to get ubuntu touch working with basically everything working a few things needed to be patches in the "GSI", then I just compile a halium friendly kernel and boot | 15:30 |
| T42 | <kabouik> Yeah that's my point, although I admit I don't know how it works for them | 15:31 |
| rinigus | kalube: (re libraries) on sfos, we frequently build static the last app. so, it is sometimes needed to rebuild it all | 15:59 |
| baloon | hi, my device is not booting, but i can connect via telnet, how can i take a log? | 16:03 |
| mal | which telnet do you use | 16:04 |
| mal | which port that is | 16:04 |
| mal | there are 23 and 2323 | 16:05 |
| baloon | 2323 | 16:05 |
| mal | baloon: so pastebin output of "journalctl -b --no-pager" | 16:06 |
| baloon | i don't know if you remeber me, but i'm trying to build for Z00L, i didn't read the part where i needed the same android base as stock, now i built using cm13 | 16:07 |
| mal | ok | 16:07 |
| mal | baloon: no, you don't have to have same android base as the device stock android, any working lineageos base is ok | 16:07 |
| kalube | rinigus: I see, in that particular case it should be pretty feasible to have a rebuild be triggered, like by having an extra file in the package that lists packages that should be rebuilt after it is done | 16:08 |
| mal | so if you have some working lineageos for the device you can use the same version as base | 16:08 |
| mal | baloon: so assuming your device would have for example lineageos-15.1 base then you can use hybris-15.1 branch in repo init | 16:09 |
| baloon | oh, k | 16:10 |
| baloon | baloon https://pastebin.ubuntu.com/p/bdNPBGGVKT/ | 16:10 |
| mal | baloon: usually newer bases are better, currently newest we support is lineageos 16.0 | 16:10 |
| baloon | can you please take a look? | 16:10 |
| mal | get output of command /usr/libexec/droid-hybris/system/bin/logcat | 16:13 |
| rinigus | looks like obs opensuse obs has armv7l workers. | 16:18 |
| baloon | mal https://pastebin.ubuntu.com/p/rcZ9rcm6RY/ | 16:18 |
| mal | baloon: try running "EGL_PLATFORM=hwcomposer test_hwcomposer" | 16:21 |
| baloon | EGL_PLATFORM=hwcomposer test_hwcomposercannot locate symbol "" referenced by "/usr/libexec/droid-hybris/system/lib/libdsyscalls.so"...Segmentation fault | 16:22 |
| mal | very strange | 16:27 |
| baloon | can it be related to selinux? when i manually restart the device it goes straight to twrp | 16:30 |
| T42 | <elros34> grep -i clang in $ANDROID_ROOT/device | 16:32 |
| mal | @elros34 ah, now I remember that | 16:37 |
| mal | baloon: https://github.com/mer-hybris/hadk-faq#libdsyscalls-is-cause-of-segfault-after-lipstick-or-minimer | 16:41 |
| baloon | Oh, so how can i do this? | 16:42 |
| mal | @elros34 can you help him, I need to go | 16:54 |
| baloon | ok, i'm doing grep clang on /device folder | 16:55 |
| T42 | <elros34> ok, baloon just run grep -ri clang $ANDROID_ROOT/device and disable it if it's enabled | 16:56 |
| T42 | <elros34> after you disable it I think you only need to "make libdsyscalls" in HABUILD. If it builds new *.so then you can copy it to device | 16:59 |
| baloon | is grep always this slow? Geeze | 17:10 |
| T42 | <elros34> it should be pretty fast, like few seconds | 17:12 |
| baloon | lol, i restarted it and now its done | 17:13 |
| baloon | asus/msm8916-common/gps/utils/Android.mk:LOCAL_CLANG := falseasus/msm8916-common/BoardConfigCommon.mk:# Clangasus/msm8916-common/BoardConfigCommon.mk:USE_CLANG_PLATFORM_BUILD := true | 17:13 |
| T42 | <elros34> USE_CLANG_PLATFORM_BUILD := false should help | 17:15 |
| baloon | ok, so i disable clang in boardconfig, make libdsyscall, and then i can run build packages, right? | 17:15 |
| T42 | <adampigg> Rinigus: regarding your last comment, its already available as build.sailfishos.org, so, its not really about merging | 17:16 |
| rinigus | @adampigg: :-) | 17:16 |
| T42 | <elros34> baloon: you only need to build droid-hal package so build_packages.sh --droid-hal but first confirm that lib builds then you can just copy it to device in /usr/libexec/droid-hybris/system/lib/ | 17:16 |
| T42 | <elros34> obviously after 'make' | 17:17 |
| baloon | T42 just wanted to say thanks, i managed to make it boot, the only things that doesn't work are wifi, bluetooth and rotation | 19:32 |
| baloon | just one more thing, the pdf is not really clear about wifi, can you tell me how should i make it work? | 19:37 |
| T42 | <elros34> read hadk-faq, it has information about wifi. If test_sensors works then you probably need hw-settings.ini ( also in faq) | 19:42 |
| *** Yardanico is now known as Yardanico_ | 21:01 | |
| *** Yardanico_ is now known as Yardanico | 21:01 | |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
