Tuesday, 2021-03-02

*** zbenjamin_ is now known as zbenjamin02:41
*** frinring_ is now known as frinring03:54
noobnothing is switching off the phone05:52
noobeven power+ low volume05:52
dcalisteHello @chriadam, how are you ?07:55
chriadamhi dcaliste, I'm well thank you.  How are you?07:55
dcalisteI'm great thank you.07:56
dcalisteI'm looking into the failing test in QMF you spoke about last week.07:56
chriadamah!  thanks, any luck so far?07:56
dcalisteI've just succeeded into reproducing it so far ...07:56
chriadamyeah there's something broken somewhere inside but it's not immediately obvious to me what ;-)07:57
chriadamand I haven't had a chance to revisit it since our last meeting, either, unfortunately.07:57
chriadamso I can't offer any more advice07:57
dcalisteWell, I've been side tracked looking for a non existing bug because the storage plugin was not found but no warning appeared and the code segfault at one point.07:58
dcalisteSo I spend part of yesterday investigating that !07:58
dcalisteUp to this morning discovering that it was just storage plugins that were not found.07:59
dcalisteNow I think it properly reproduces the bug. The add is failing and later on in the restore one, an assert is raised.07:59
chriadamthat is the behaviour I was seeing08:07
chriadamhowever the storage plugin thing is interesting - I didn't see that / run into that problem08:07
dcalisteOk, well, I'll investigate later today now.08:07
chriadamthanks very much08:07
chriadamI've poked flypig_ to see if he can look at the buteo-syncfw MR now that he's back from vacation ;-)08:07
dcalisteAbout the plugin problem, it was my fault (of course), I forgot to export QT_PLUGIN_PATH properly.08:08
chriadamah, fair enough :-)08:08
dcalisteSpeaking about flypig (thank you by the way), I've pushed upstream to kcalendarcore a minor change that allow not to pay attention to lastModified field being change at each change, see https://git.sailfishos.org/mer-core/kf5-calendarcore/merge_requests/308:10
dcalisteAt the moment, the calendar is changing lastModified field for every custom property change when we add the etag, or flag the event for failure.08:11
dcalisteIt's boring because we need to save and restore the lastModified field by hand.08:11
dcalisteWith the latest upstream, this is not necessary anymore when switching a flag in the calendar.08:12
chriadamah, nice.08:12
chriadamone thing worth noting:08:12
chriadamcurrently I think we cannot accept code which is LGPLv3 (hopefully that changes soon) so we will have to be careful with the kf5-calendarcore stuff to make sure we don't accidentally start depending on code which has that license.08:12
chriadamit is not a problem so far, as far as I'm aware, but something that I thought I should mention.08:13
dcalisteOk, the v3 license seems to exist in the LICENSES directory of kcalendarcore, but not any of the source or header files is using it, only v2 at the moment.08:15
chriadamyeah.  I think there might have been one unit test with that license, but we don't ship that one afaik.08:15
chriadamanyway, we just a note :-)08:16
chriadamregarding the calendar color being overwritten by the server value: blam followed your example fix in caldav and did the same for google calendar plugin08:16
chriadamnot reviewed/merged yet but should go in this week, I think08:16
chriadamthanks very much for your work on the caldav plugin for that one, and identifying the issue08:17
dcalisteI've just seen the MR from blam this morning, indeed.08:17
dcalisteAbout the v3 license, indeed, some tests are using this version actually.08:17
chriadamoh, more than one?  we should make sure we don't package those ones in our package.08:18
chriadamI will make a note of that internally08:18
dcalisteWell, if the tests are not installed by default, it will do no harm, won't it ?08:19
dcalisteThe problem is with installed packages, I guess.08:19
chriadamoh, I guess that's true08:19
dcalisteAnyway tests are not packaged from the spec...08:21
chriadamin caldav side, have you ever seen a case where the server redirects requests for the .well-known path, to the root path?08:22
chriadamwas seeing this in some case in carddav: a community member had this issue a long time ago and apparently I completely forgot to fix it / investigate it properly08:22
chriadambasically, we try to perform path discovery, but get redirected back to the root path, and so we abort discovery08:23
chriadamthis occurs even if the user provides an addressbook path via the account creation UI (which is clearly a bug, I need to fix.  i.e. skip discovery if the path is provided manually)08:23
dcalisteActually no, but I'm using only one provider. But I didn't remember any community question like that in TJC or the forum.08:23
chriadambut just wondering if this might be an issue for caldav also08:23
chriadamit's probably not an issue for caldav, since there is some custom handling of calendar paths (stored into the account setting etc)08:24
chriadamI will look into the carddav plugin later this week and try to figure out how to resolve.  just thought i'd ask, in case :-)08:24
dcalisteSure, but sorry, I've no advice for this.08:24
chriadamdid you have anything else to discuss / raise?  is there anything else waiting for input from either flypig_ or myself?  I think pvuorela_ said he hopes to spend some time this week or next week looking into both of our QMF upstream changes / reviewing, also.08:25
dcalisteCompiling your branch of QMF, maybe we can discuss a bit about the build system for the qmflist_p.h file.08:25
chriadamah, yes08:25
chriadamI saw a comment on gerrit but didn't yet digest it08:25
chriadambasically, currently that file just gets manually included everywhere via a .pri hacky thing08:25
dcalisteMaybe you've seen my message there, but I can develop here.08:26
chriadamplease do08:26
dcalisteIndeed, if you don't add the .pri include you get screwed, and since the header is leaking into public place (it's in qmailmessage.h), one should not have to include the .pri I think.08:26
dcalisteThe qmflist_p.h should be installed like longstring_p.h is or qcop*_p.h things.08:27
dcalisteLike that, just putting QT += qmfclient will do the job, without needing to add the .pri include.08:27
chriadamhow do we handle the qmailmessage.h case currently?  I mean, is it part of the public API (e.g. a signature of a public method)?08:28
dcalisteI mean that it's a public header so a client code can write #include <qmailmessage.h>08:28
chriadamyeah, but is the qmflist_p.h used by qmailmessage.h as an implementation detail, or is it part of its API?08:29
chriadamI don't recall08:29
dcalisteBut the qmflist_p.h is not installed, so one cannot do the include anymore from outside QMF.08:29
chriadamah! yes, that's clearly a problem also haha.08:29
dcalisteGCC complains about the include not finding the file.08:29
dcalisteThe easier solution in my opinion is to move the file somewhere in qmfclient/ and add it to the HEADERS variable.08:30
chriadammakes sense08:31
chriadamI will update the appropriate commit in the chain to do that instead of the util thing.08:31
chriadamthank you for pointing that out08:31
dcalisteTo test the possibility to compile properly, in the tests, you can just remove the .pri include, it should just work with the QT += qmfclient qmfclient-privqe things I think.08:32
chriadamI wonder if there's any case which doesn't include qmfclient but still needs qmflist_p.h08:38
chriadamI guess we can special case that specific case, if it does exist08:38
chriadamseems unlikely anyway08:39
dcalisteYes, that's a good point. I think that every piece is using the qmfclient part, but I cannot guarantee it.08:39
chriadamdid you have anything else to discuss today?08:41
dcalisteNo, that's all. Thank you for the discussion.08:42
chriadamthank you for the discussion, and your time and effort as always!08:43
chriadamI hope that you have a great week :-)08:43
* chriadam -> away, gnight!08:43
eagle74what is the best way to reflash the phone08:59
eagle74looks like something not right with it08:59
eagle74and nothing is helping fixing it09:00
ggabrieleagle74: what device? :D09:01
eagle74xperia 10 plus09:02
ggabrieleagle74: follow the flash instructions on the sfos guide, as you did originally?09:03
ggabrielobligatory note: you'll lose all your data, so get a backup09:05
eagle74i don't mind as long as i can fix it09:06
eagle74the steps are for installing from android09:34
eagle74not reflash09:34
eagle74not sure how to do it09:34
eagle74and it requires the phone to be switched off09:35
eagle74something i can't do ( reason for reflash)09:35
eagle74even with power and low volune09:35
BlumenkraftWhat do you mean "from Android"?09:36
eagle74i have already sailfish os09:37
eagle74installed on the phone09:37
BlumenkraftThen just skip to executing flash.sh.09:37
eagle748.1  Make sure your phone is powered off.  The LED must not show any light.09:46
eagle748.2  Hold down the Volume Up button and connect the other end of the USB cable to the phone. The LED next to the speaker on the phone should light up blue.09:46
eagle748.3  On terminal,  use “cd” to proceed to your Sailfish OS folder the same as in 4.209:46
eagle748.4  Launch the flashing script by entering the command:09:46
eagle74i can't turn the phone off09:46
*** Malinux is now known as Maltias09:49
*** Maltias is now known as Malitas09:50
ggabrieleagle74: try vol-up + power key for a long time09:56
ggabrielor vol-down + power key09:56
ggabrielI had the same with my Xx yesterday, rebooted, waiting for the crypto pass screen and used the power key there to switch it off09:56
eagle74it reopens ggabriel10:11
ggabrielcan you get a CLI in sailfish od?10:13
eagle74what is that '=D10:14
ggabrielthe terminal, you have to enable developer mode10:14
ggabrielotherwise I don't have many ideas tbh... normally vol up/down + long press power key shuts the device down10:15
ggabrielbut from the CLI I was going to suggest that you run `shutdown -h now` as root10:15
eagle74ok thanks i will also now try to reflash10:15
eagle74after it close10:15
albertuxquestion....Im license-owned sailfishos installed on xperiax. I'ts possible install the same licence on pinephone?10:44
ggabrielalbertux: not that I'm aware of10:48
albertuxevreryone not tryed it?10:51
ggabrielalbertux: tried what? you want to flash sailfix x on a non xperia device?10:51
ggabrielif there's any contacts/calendar experts in here, I would really appreciate if you could take a look at: https://forum.sailfishos.org/t/4-0-1-45-birthdays-dont-show-up-in-calender-and-feed-events/4684/11?u=gabrielg :)10:51
albertuxggabriel: almost aliendalvik for android compatibility... there are just SFOS image for pinephone. https://gitlab.com/as400l/pp-sailfishos-images10:56
albertuxthe  last image work very vell and Adam Pigg is working well for last imminet release on pinephone https://twitter.com/i/status/136532854211709337810:59
ggabrielalbertux: well, you aren't licensed to run android anywhere else than in your xperia device, fwiw. Anything else you do, it's up to you11:03
albertuxggabriel: thanks for clarifications, but overall I payed the licence...it will be a nice thing if it possible...11:07
ggabrielalbertux: I understand but if your angle is legal, then the licence is just for the devices you paid it for11:17
ggabrielyour question is quite old anyway (can we have android for community ported devices), and sadly it isn't as straightforward as you would think11:17
eagle74how to unlock boat loader in sailfish os 411:57
malwhat do you mean?11:58
Blumenkrafteagle74, You have to unlock bootloader before installing SFOS.12:00
eagle74ok so if it's unlocked from 1st flash no need12:02
eagle74ok how i can lock it after flash12:03
eagle74i just reflashed sailfish12:29
eagle74i can't turn the phone off12:29
eagle74can it be hardware issue12:29
ggabrielyou can't lock the bootloader after flashing sfos (easily, if at all possible)12:34
eagle74i'm just frustrated about the inability to switch off the phone12:48
eagle74the only way is volume up and power for 10 seconds12:48
eagle74i don't think this normal12:48
ggabrieleagle74: try to go back to android (using the emma tool), see if you can reproduce there13:12
eagle74it happened after the 4 update13:13
ggabrielI don't have one of those devices, so I can't tell. Reproducing or not in android could tell us if it's got anyting to do with sfos13:16
eagle74will try that13:16
*** _inte_ is now known as inte14:50
attahdcaliste: hello :)18:25
dcalisteHello attah !20:27
attahdid you see my last pm(s) on the forum?20:28
dcalisteI'm not yet falling asleep ! Do you still have questions about Calligra converter ?20:28
dcalisteYes, I've seen your messages.20:29
dcalisteCan you open in the document application with stock Calligra the files that are failing ?20:29
attahYes, both of them open just fine20:29
dcalisteI admit I tested only one ODF file on device.20:29
attahbut at least it worked on device :)20:30
dcalisteBut a simple call to calligraconverter foo.odt foo.pdf actually produced the PDF.20:30
dcalisteAh, you mean it's generating the PDF on device ?20:30
attahNo, it is segfaulting for me still with calligraconverter20:31
attahbut the test documents open fine in the documents app20:32
dcalisteOk, I see.20:32
attahand then there was the matter of that it seems some patches are missing, does that sound reasonable, or should i suspect i messed something up?20:33
dcalisteWell, that missing patches ring a bell. This morning I wanted to submit the MR in SailfishOS and tried a latest rebase on upstream master then compiled.20:34
dcalisteI noticed then that many patches were missing and I still have issues to compile.20:35
dcalisteSince this morning, I'm trying to fix my mistakes.20:35
attahI couldn't obviously see that these were things that had been patched before and got lost, so i suspect they are new20:35
attahoh... i have dragged you down the rabbit hole20:36
dcalisteHopefully I'll submit the MR tonight or tomorrow morning.20:36
attahno stress!20:36
dcalisteNo problem, if it could help you, the better.20:37
attahYeah.. i don't think i could have done this myself, so i'm really grateful20:37
attahAnd it helps me help others even more20:38
dcalisteWell, I've already upgraded Calligra from 2.9 era to 3.1 some years ago, so it's not totally alien to me.20:38
attahgood :)20:39
attahcould i email you my test .odt docment and we can see if it hits something new?20:39
attahit is trivial really, but would be nice to rule out20:40
dcalisteSure you can mail it to dcaliste A free.fr20:41
attahone odt coming up20:42
dcalisteReceived and sent back as PDF !20:47
dcalisteIt seems to work on device with the packages I compiled yesterday from my fork of Calligra. I'm now recompiling calligra master since the patches have been accepted in SDK.20:48
dcalisteIf conclusive, I'll submit a MR from my converter branch.20:48
attahhope it was something did to get it to compile that made me hit the segfault20:49
attahalthough i thought it was just things close to the GUI that i "fixed" poorly20:49
dcalisteYeah, no idea why it's segfaulting for you. Maybe some missing patchs , there were several things to be patched here and there for it to work, even on desktop.20:51
attahi'll race you to compilation ;)20:52
attahdcaliste: done, and it went completely without errors21:11
dcalisteMine is not finished yet !21:11
attahAMD 3950X ¯\_(ツ)_/¯21:12
dcalisteGood is it working ?21:12
attahtesting now21:13
attahno :( Segmentation fault (core dumped)21:13
attahmaybe it is because i didn't install it properly, but i wouldn't have expected segfault for that21:14
dcalisteWell, maybe, I had a segfault on desktop at first because the ODT plugin was not found and the main part was not assuming it could fail...21:15
dcalisteI'm sorry it's getting late here. We may try to continue investigating tomorrow.21:16
dcalisteKeep me informed if you go further in any direction.21:16
attahsleep well, you certainly deserve it21:17
attahwill do, but i i'll be off to bed quite soon too21:17
dcalisteThanks, see you, and thank you for your work also.21:17

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!