07:00:25 #startmeeting Sailfish OS, open source, collaboration -- 5th August 2021 07:00:25 Meeting started Thu Aug 5 07:00:25 2021 UTC. The chair is sledges. Information about MeetBot at http://wiki.debian.org/MeetBot. 07:00:25 Useful Commands: #action #agreed #help #info #idea #link #topic. 07:00:42 #info Meeting information and agenda can be found here: 07:00:44 #link https://forum.sailfishos.org/t/community-meeting-on-irc-5th-august-2021/7159 07:00:51 I am the meeting's chairperson today, and will be doing my best to keep time and order. Please respect Olympic timings. 07:00:58 #topic Brief introduction (5 min). Please prefix your name/handle with #info 07:01:05 #info Simonas Leleiva -- privateer for Jolla 07:02:19 #info fridlmue - community 07:02:22 #info Joona Petrell - sailor @ Jolla 07:02:25 #info Otto Makela -- community 07:02:46 #info David Llewellyn-Jones - sailor @ jolla 07:03:15 #info Lukáš Karas - community developer 07:05:54 #info thigg - community dev 07:06:12 seems most community stil on holidays, and none of question askers came forth;) let's see 07:06:15 #topic Status of allowed APIs and restrictions to the store (12.5 min -- asked by ApBBB) 07:06:32 #info In the past there has been a discussion about all the things that are not allowed in the store and APIs needed. We also made a poll "API Priority poll, take II": 07:06:36 #link https://forum.sailfishos.org/t/api-priority-poll-take-ii/4159 07:06:47 #info to prioritize things. Is there any status update on that or parts of that. SFOS is held back by all these restrictions and the sooner we get rid of them the better for the end user. 07:07:08 #info The API priority poll was made by Jolla chief Ville Nummela to get feedback from community about the biggest omissions in the Sailfish OS API offering. We have lately put considerably more effort on APIs, which you should gradually start to see in the upcoming releases. Stay tuned. :) 07:07:32 any sneak peak? 07:07:42 morning, #info piggz, community proter 07:07:59 ie two apps that interest me are held back by Secretc not being allowed 07:08:05 secrets 07:08:21 sneak peaks: 07:08:23 #link https://github.com/sailfishos/sdk-harbour-rpmvalidator/pulls?q=is%3Apr+is%3Aclosed 07:08:46 so Sailjail was the last one that sneaked in before holiday period 07:09:16 piggz proter /ˈprəʊtə/ noun Biology In ciliate protozoa: the anterior of the two organisms formed by transverse fission. Origin 1950s. From French proter from ancient Greek πρότερος in front. 07:09:18 I heard some rummors about new sharing api, that works properly with sailjail. When it will be introduced and documentation exposed? 07:11:19 ExTechOp: its early 07:12:18 not that early if you woke up to watch karate at 2am:)) 07:12:36 sledges: are things finished for the next release or more things might end up in it? 07:13:12 ApBBB: we welcome all APIs in the chum: repo ;) 07:13:23 ApBBB: there is also another overview in 07:13:24 (just thought i'd sneak that in there) 07:13:24 #link https://sailfishos.org/#DX 07:14:40 piggz: as i said in the forum topic chum is great but the end goal should always be the official store/repo. 07:15:02 and GUIs 07:15:34 well, yeah, its different ... its a more general repo fr libs/apps etc, butt thats a different topic :) 07:15:38 i trully believe that a well design UI "democratizes" technology/features 07:15:53 #info Damien Caliste, community, sorry to be late 07:16:05 ApBBB, out of interest, what are the two apps? 07:16:20 sailtrix and the github one 07:16:35 Ah, Sailtrix, I agree that's interesting :) 07:16:37 at least secrets was one of the things holding them back 07:17:01 I could imagine a version of Sailtrix without Secrets available in the store. 07:17:09 sailtrix is huge. is the first time we have a matrix client that reach that level of features 07:17:19 and the dev seems to be on fire 07:17:33 That's very exciting, I agree. 07:18:37 time's up for this one, any last words? :) 07:18:47 sailtrix without secrets is something that feels really bad. All your matrix access tokens are just accessible for everyone... 07:18:51 and yes, the dev seems on fire :D 07:19:16 last words: piggz is chumming the devs - literally 07:19:24 we can move on if someone doesn't have anything to add 07:19:25 thilo[m], I take your point, but that's no different to how things are for other apps. I think it's not the disaster it sounds. But I also agree secrets is a good addition to it. 07:19:47 He did that really impressive x86_64 distro of sailfish too 07:20:09 thilo: but it is accessible to everyone via sfos secrets then, isn't it? 07:20:47 rinigus: shouldnt that be guarded per-app or something? how is that implemented? didnt take a look 07:20:57 rinigus, I didn't check the sailtrix implementation, but depending on how it's set up, in some cases it requires your lock code to access. 07:21:02 #info good mentions of the most promising native Matrix app: 07:21:04 #link https://forum.sailfishos.org/t/sailtrix-matrix-client-supporting-end-to-end-encryption/7199 07:21:34 movin' on up 07:21:39 #topic Different bug patch release strategy for 10 II? (10 min -- asked by EA3ICN) 07:21:42 cy8aer is also working heavily on the bluepill matrix app, it comes with offline storage as well 07:21:50 #info Would it be possible to release bug fixes in smaller but more frequent batches in order not to wait for the full next release of the OS? Hot fixes seem to be the norm in other projects, is there a reason why these are not used in SFOS? 07:22:25 I'd like to add Determinant to the matrix mix 07:23:30 EA3ICN isn't about, so i'll just answer without discussing with them.. 07:23:33 #info We are now consciously putting effort in making more regular releases to avoid bigger gaps between the updates. 07:23:59 #info We used hotfixes repo for the whole OS in the past (see `ssu lr`), however nowadays we instead choose well tested stable OS releases, after proper integration. Which unfortunately does not leave room for patching up individual devices. 07:24:46 Good morning all 07:24:56 Good morning. 07:25:25 do I understand correctly that your planning to have more often releases in the future? 07:26:14 idk, but things like the wrong swipe direction from the lockscreen (i think on all devices) in the current 4.1 are super annoying and cry for a hotfix. I think that was like one line of code? Fot things like that a hotfix would be such a great thing. 07:26:52 sledges, are there some areas where things go out, outside of the normal release schedule? I'm thinking about the MLS data for example. Or are they also synced? 07:27:14 they're store apps, so don't need to follow the release schedule 07:27:19 flypig: those are dealt by the store 07:27:21 what's this swipe direction thing? 07:27:36 MLS data is currently synced with releases, but would be good to decouple 07:28:09 What is MLS? 07:28:11 we did hot fixes when Jolla 1 came out and the overall stack was less stable 07:28:12 Once upon a time, the patch manager thing was excellent for this kind of quick fixes, but unfortunately it seems to have somehow fallen on the wayside, at least I haven't seen any patches getting ported to 4.1? 07:28:19 abr: https://forum.sailfishos.org/t/4-1-0-23-events-quick-access-swipe-is-from-the-wrong-side/6253 07:28:23 mozilla location service offline positioning data 07:28:31 MLS ^^ 07:28:58 abr: https://forum.sailfishos.org/t/4-1-0-23-events-quick-access-swipe-is-from-the-wrong-side/6253/20 07:29:08 ah, sorry, thanks dcaliste! 07:29:10 Hmm, not seen that one! 07:29:19 hot patch releases can be inconvenience for users who don't need the particular fix but then go through the additional upgrade process for little benefit 07:29:25 @jpetrell ty 07:30:18 abr, it has some specific preconditions in term of settings, but when you have it, you suffer from it every time you wake up the phone... 07:30:27 ah, no lock code 07:30:49 @jpetrell you arenot ddoing delta updates? So every update is huge? 07:31:10 delta rpms are generated for each release to minimize the downloading 07:31:40 abr: i dont think ive ever seen delta rps work :D 07:31:53 yeah you have. every time you update a sfos device :D 07:32:03 abr: i have a lock code, but still it happens. But its a little OT at the moment. Lets chat about that in the general discussion... 07:32:12 abr: no, i watch it, and it always says "failed, downloading full rpm" 07:32:29 probably due to brexit 07:32:36 Haha :D 07:32:57 i turned off delta in zypper settings 07:32:59 :) 07:33:33 time's up here too, thanks for resurfacing the swipe bug 07:33:42 #topic implement the fuctionality of ‘Personal Ringtones’ by coderus into Sailfish OS by default (30 min -- asked by tobset) 07:33:50 #info playing individual ringtones for specific contacts is a core functionality of a (smart)phone for at least 18 years. 07:34:13 #info Since coderus stopped maintaining 'Personal Ringtones', no aarch64 support is in sight. (So my X10 II matches my old 6310...and we’re back in 2002 ;) 07:34:18 #info The rest of the description: 07:34:21 #link https://forum.sailfishos.org/t/community-meeting-on-irc-5th-august-2021/7159/4 07:34:49 tobset said probably couldn't make it, so I promised to pick this up in that case. 07:35:03 awesome thanks flypig! (talkin' 'bout substitution:) 07:35:11 #info Contact-specific ringtones would indeed be a very cool feature. Getting Personal Ringtones to work on aarch64 probably only requires a recompilation, unfortunately looks like the playback implementation (voicecall plugin) is only stored in the personal-ringtones git repository in binary format. 07:35:28 #info Upstreaming would require integrating the functionality to platform's dedicated voicecall ringtone plugin, contact libraries, sound settings, audio policy definitions, which is a bit of work. 07:36:31 A kind of a reverse thing to ringtones is "night silence" which doesn't seem to have been ported to aarch64, so we're back to the early 2000's in this respect too? 07:36:53 Was "night silence" an app? 07:37:01 *Is 07:37:33 tobset was clear that while Personal Ringtones is really great, it would make more sense to have this integrated into the OS. 07:37:43 Yes, I have it on my Xperia XA2 but it isn't available for 10II 07:38:12 For Personal Ringtones, it's unfortunately not just recompilation (and the binary library) that needs fixing, but also access to contacts. 07:38:24 About ring tones, how much work can be done in the open in public repo, and how much requires changing proprietary parts ? 07:38:49 I mean, the UI to select ring tones per contact can be an app, but the internal machinery could be upstreamed already ? 07:38:51 dcaliste asks the right questions :D 07:39:45 Personal Ringtones appears to be almost exclusively QML. 07:40:02 Apart from this binary blob.... 07:40:19 where does that store its data though? should be in the contacts db really. 07:40:21 the whole "there is an app for that" thing is wrong if the fuction belongs in the settings or an excisting app (contacts) 07:40:35 binary implements the actual feature. QML is for setting the thing 07:41:25 ApBBB I concur, it would be nice to also have the "be quiet between hours XX/YY" as a part of the operating system. 07:41:46 jpetrell, that's a shame. I wonder what it does. 07:41:47 "Do not disturb" is still there, so a stepping stone for quiet hours 07:41:51 ApBBB, I agree, but having it as an app could be a first step to demonstrate the UI-design and importance of the functionality, before integration… 07:42:53 jpetrell, if the functionality of personal ringtones is done without source code through a blob, then no luck… 07:42:54 ApBBB, that's especially true if the app turns out to be brittle on OS updates. 07:42:57 dcaliste: if I had to guess there can be conflicts with the audio policies and platform's own ringtone voicecall plugin, e.g. platform one probably tries to play ringtone but is muted, better way would be to handle the states consistently in one plugin instead of two plugins fighting who gets to play 07:43:22 would it make sense to implement the API for personal ringtones first, thus a column in the contacts db and the usage of the column? that shouldnt be so much effort. Then people can step in and build a gui for it, which can then be ported in turn 07:43:33 jpetrell: indeed. Is the Jolla plugin open source ? 07:43:46 (or is this the main effort?) 07:44:11 Does Contacts have an "extensions" field already available? 07:44:13 dcaliste: yeah it is in voicecall github project 07:44:49 https://github.com/sailfishos/voicecall ?? 07:44:55 also looks like the personal ringtone is separately saved in dconf as remote uid (phone number) + ringtone path, more natural place would be part of the contact data 07:45:19 jpetrell, cool, then the work should be directed there to add the functionality in accordance to the current plugin. Well, I'm saying a lot, but I've no clue on how this is actually working internally. 07:46:20 I guess coderus saved the settings as dconf out of convenience to make it work. But a final implementationwould be to add it to the contact db, indeed. 07:47:33 flypig, we're missing chriadam for the relevant question of extension field in contact db… 07:47:38 dcaliste: seems the functionality could be implemented in open source side. the actual setup UI could be done as separate app as the contacts UI is not open source 07:47:50 flypig: yeah that is the one, I guess plugins/ngf/src/ngfringtoneplugin.cpp 07:48:10 dcaliste, yes, I'll ask him outside of the meeting and report back at the next one. 07:48:45 adding sound clip picker in contacts edit flow doesn't sound like a lot of work 07:48:57 jpetrell: at some point you should make it easier for the community to add stuff to the QML side of things -which are visible to all. (assuming the contributor doesn't care about licence) 07:49:21 yeah 07:49:58 ha, i put 30mins by mistake, which is good:) OP gave only 5mins for this topic 07:50:01 Could we add that link to the minutes for reference? https://github.com/sailfishos/voicecall/blob/master/plugins/ngf/src/ngfringtoneplugin.cpp 07:50:20 sure 07:50:24 #link https://github.com/sailfishos/voicecall/blob/master/plugins/ngf/src/ngfringtoneplugin.cpp 07:50:28 Thanks! 07:51:02 flypig: I think we're using QContacts or something similar and there is a setCustomField() function, so maybe like for calendar DB there is already a generic table to store custom properties… 07:51:18 dcaliste, that would be good indeed. 07:51:49 By the way, coderus is looking into the binary blob to see if any info can be made available. 07:51:56 movin' on up now 07:51:58 #topic General discussion (30 min) 07:52:11 Just to say, as I understand it tobset's hope was bring this to the attention of Jolla developers, so I think it achieved that. 07:52:13 flypig, good thanks coderus. 07:52:21 lqramen: your topic was announce too late to answer this time unfortunately, will move it to meeting after fortnight 07:52:29 I was here... nooo 07:52:44 cut off time is 3 days before the meeting 07:52:48 ah sorry 07:52:50 my fault 07:52:52 np 07:53:14 I wiill repost for the next time. I got disctracted on vacation.. wife would not allow me near my work latop 07:54:10 as I didn't get reply on sfos-porters channel - is the code for disk encryption UI and integration with cryptsetup closed source? 07:55:52 Just out of curiosity: the old app "IP Address" which I've occasionally used for connection debugging hasn't been ported to aarch64 -- is this just due to the original maintainer Jonni.Rainisto@jolla.com being gone? 07:57:26 IP Address is a neat app. That's a good question. 07:57:57 Could someone take it over and make it also work with the other architectures? 07:57:57 doesn't the developer mode settings applet show IP addresses? 07:58:13 It does, but it's not as convenient or thorough. 07:58:39 I see. I'll have to try that one out 07:58:44 I am using "Network Info" app to see my address... 07:58:51 Just my opinion! I tend to be using the Settings app for something else, and you can't show two pages simultaneously. 07:59:05 rinigus: rpm -qi sailfish-device-encryption-unlock-ui 07:59:09 is closed source 07:59:14 terminal -> ip a ;) 07:59:18 I take the general discussion also as chance to ask a somehow development related question: Can I somehow force the Sailfish-Keyboard to appear in an App when clicked on a TextField, also when a "USB-HW-Keyboard" is attached? Is there some Switch i can set? (in best case from the QML environment; no Harbour-Compliant solution needed) 07:59:44 sledges: don't have it installed, hence the question. any hope of it getting open sourced? 07:59:46 rinigus: but there should be some crypt mounting going on in the open (haven't checked closely): https://github.com/mer-hybris/hybris-initrd 08:00:18 sledges: thanks for pointer 08:00:35 also yamui is used for early boot UIs https://github.com/sailfishos/yamui 08:00:37 ooh ipv6 address. we should really show that too :) 08:01:09 and this for recovery iirc https://github.com/sailfishos/minui 08:01:55 any exciting news sailors can share regarding the next release? 08:01:55 rinigus: what sparked your curiosity about encryption? 08:02:14 Earlier Bluepill and Determinant were mentioned. What are the prospects of E2E for them? 08:02:23 @fridl did you try myTextField.forceActiveFocus(); ? 08:02:44 sledges: decided to encrypt it properly and feel bit safer of a data stored on device 08:03:19 flypig: in QtPIM, QContactDetail, there is already a field named TypeRingTone, see https://github.com/sailfishos/qtpim/blob/01d1f054785836fb7cd7c4c373eabd0d9ec29bba/src/contacts/qcontactdetail.h#L75 08:03:23 flypig: none was to the level of sailtrix 08:03:27 with the current key restrictions i do not consider my device as encrypted at all, really... 08:03:28 flypig: The actual branch works with e2e out of the box. But I need time to glue it to the GUI 08:03:28 afai know 08:03:48 (bluepill) 08:04:12 Are there fixes expected for the memory subsystem, since some Android apps (eg. Signal) are reacting unfavorably by being given isLowRamDevice() which I think is due to /dev/memnotify missing from current versions of SFOS? https://forum.sailfishos.org/t/android-subsystem-self-reports-islowramdevice/6957 and https://forum.sailfishos.org/t/4-1-0-24-xperia-10-ii-unknown-memory-level-reported-by-mce/6856/5 08:04:21 karry: No, but I think its not a focus problem. When a HW-Keyboard (or in this case a YubiKey which also registers as HW-Keyboard) is attached, then the Software Keyboard is not opened at all. But ill try this one as well! 08:04:28 sledges: not sure yamui and minui (that one is new to me) are used during normal boot to open /home. I would expect it is not in the initrd either but later in boot 08:04:31 sidetopic: podqast has an openrepos branch because the mpris integration is not allowed in the store. is this subject to change? 08:04:32 ApBBB, Cy8aer[m], thanks, that's interesting to hear. 08:04:46 sledges: ... but would have to check it out 08:04:50 rinigus: it's to ask for device code (hence encryption ui) 08:05:07 sledges: minui? 08:05:11 @fridl true, I don't tried it with hardware keyboard 08:05:13 But for the sfos harbour troup I would apreciate some look at the code if that would go into harbour without problems: https://gitlab.com/cy8aer/bluepill nio branch - because it is full of binary libs and python libs in the rpm itself... 08:05:25 dcaliste, that's promising. That's something to look into further for sure (I guess it's not clear whether it's used for anything currently, or what format it expects). 08:05:27 yamui (which also has (unrefactored?) minui as subfolder) *shrugs* 08:05:30 * But for the sfos harbour troup I would appreciate some look at the code if that would go into harbour without problems: https://gitlab.com/cy8aer/bluepill nio branch - because it is full of binary libs and python libs in the rpm itself... 08:05:46 rinigus: but it's generic framework, closed src component uses yamui to place numbers on screen and react to touch etc 08:06:22 Cy8aer[m], "full of binary libs and python libs in the rpm itself..". Does it pass validation? 08:06:45 flypig: indeed, I've no knowledge in this area, I was just looking here and there for API that may be suitable. Proper grepping should be done, in the SQlite plugin also to see if this field is ever stored. 08:06:52 rinigus: im curious if there is a performance hit once encryption is enabled (speed and battery), maybe that's why a less cryptic algorithm was chosen for default sfos? 08:07:09 it should because all stuff is inside `/usr/share/harbour-bluepill` with a binary coded to the lib path there as described in the manual. 08:07:17 sledges: ok, looks like I have to check it out on my dev device :) . but I wonder whether opening LUKS can be done later and not require messing with yamui 08:07:23 dcaliste, I'm impressed you found it so quickly :) 08:07:43 It seems like a good lead. 08:08:25 * it should because all stuff is inside `/usr/share/harbour-bluepill` with a binary coded to the lib path there as described in the development manual. 08:08:33 Cy8aer[m], the best thing to do would be to run the validator across it first yourself if you haven't already. It should be a straighforward step. 08:09:11 sledges: encryption should have the same performance. it is just real encryption key is not encrypted itself well enough with current implementation 08:09:32 flypig: yep probably - but I just have a two way building process - libs before the application and it is not done with one qmake. Let's see... 08:09:37 got it thx 08:10:08 Cy8aer[m], it doesn't have to be part of the build step, I'm sure you can just call the validator manually. 08:10:14 (on the rpm) 08:10:35 flypig: I'll see if I get the validator into my CI-Chaing. 08:10:46 Chain.. 08:10:51 ./rpmvalidation.sh .path/to/rpm-file 08:11:09 Ah, thanks. 08:11:42 Or, more likely: sfos engine exec ./rpmvalidation.sh path/to/rpm-file 08:11:55 Sorry: sfdk engine exec ./rpmvalidation.sh path/to/rpm-file 08:12:13 I will find it. 08:12:28 More info here: https://github.com/sailfishos/sdk-harbour-rpmvalidator 08:14:35 Any comments on the memory issue I asked above? 08:17:26 ExTechOp, I have no idea personally I'm afraid. If no-one else here does we can to try highlight your forum post to someone who might. 08:18:31 The repercussions are silly, you only get to send really small versions of photographs on Signal :-/ 08:19:14 ExTechOp: memnotify is used by mce and is to send signal to native browser to close tabs on low mem situations 08:19:24 ExTechOp: this has been filed internally and fixed for next release https://forum.sailfishos.org/t/4-1-0-24-xperia-10-ii-unknown-memory-level-reported-by-mce/6856/3 08:20:03 Would this also help the Android subsystem memory issue? 08:20:19 android subsystem uses its own lmkd 08:20:20 The connection between these two is just speculation on my part. 08:21:03 And also no more thoughts to my HW/SW-Keyboard-Issue, right ;-P 08:21:34 the Android image is purposely built with a low memory config 08:21:45 sledges Any ideas out of which hat lmkd pulls the isLowRamDevice() status? 08:22:33 ExTechOp: what does device reports as `free -m` in low mem periods in Signal? 08:23:27 it should kick in when whole device is in low mem, as expected 08:23:35 also are you experiencing this: https://forum.sailfishos.org/t/4-1-0-23-random-closing-of-all-apps/6457 ? 08:25:06 sledges As far as I can tell, this is not "a period of low memory", it's a permanent situation? I've seen that random close of all apps occasionally, but not on a regular basis. 08:25:36 What part of the output of "free -m" is relevant? 08:25:42 Mem and Swap 08:25:54 'free' column 08:26:17 Mem: free 73, Swap free 427 08:26:42 sounds plenty, it's strange this is not common occurrence with other Signal users 08:28:10 Well, are there any others here? 08:28:46 and/or keep eye on the forum, i'll ask around internally 08:28:47 time to wrap up 08:28:52 #topic Next meeting time and date (5 min) 08:28:53 sledges Thanks! 08:28:56 Proposing Thursday 19th August at 7am UTC 08:29:03 when compared to earlier SFOS versions (maybe due to not using whisperfish), memory pressure on aarch64 / tama port seems to be larger. I have seen some apps closing and RAM usage seems to be larger than 32 bit one 08:29:18 (sorry for late message) 08:30:14 rinigus This came up on Xperia XA2 Ultra under SFOS 4.1 08:30:36 19th looks good to me :) 08:30:51 rinigus: question is is it due to 64bit (which is expected), or general hunger from 4.0 to 4.1 08:31:26 but thanks for your data point! 08:31:32 #info Next meeting will be held on Thursday 19th August 2021 at 7:00am UTC: 2021-08-05T07Z 08:31:40 and thanks all for your presence! 08:31:41 #endmeeting