Tuesday, 2019-02-19

*** frinring_ is now known as frinring01:40
*** zbenjamin is now known as Guest2025302:46
*** zbenjamin_ is now known as zbenjamin02:46
r0kk3rzoh wow, LG decided to open source some of their webos stuff02:49
*** feodoran is now known as Guest9681603:15
*** feodoran_ is now known as feodoran03:15
*** SpeedEvil is now known as Guest7867903:26
dcalisteHello chriadam_ and jpetrell: I hope you're fine. I'm afraid I'll be 15 minutes late today. I'm still at home looking over the children. I'll leave for office in 10 minutes.07:52
chriadam_dcaliste: hi.  no worries :-)07:54
dcalistechriadam_, thanks. Children are ready, I'm leaving. Should be online again in 15 to 20 minutes.07:59
dcalisteHello again chriadam_ and jpetrell, back online.08:21
chriadam_dcaliste: hi :-)  I hope you had a nice holiday08:21
dcalisteYes, weather was particularly great for the winter season, which was pleasant. I've seen you've pushed (or Andrew) the modification in silica, thanks.08:22
dcalisteAs discussed on the MR in Gitlab, for the part in QtDeclarative, I'm going to propose it upstream this week and continue the discussion there.08:23
dcalisteI haven't time yet to further test if it's correcting the issue with lipstick-security-ui :(08:23
chriadam_sounds good, and no worries08:24
dcalisteMy main concern is that I'm not able to reproduce the bug with a simple QML test opening and closing a window that has constant colouring refresh…08:24
chriadam_nor have I spent any more effort on that one, yet, unfortunately have been sidetracked onto other tasks so far08:24
chriadam_dcaliste: application quit could be involved, depending on if deferred delete events are promoted in the event queue for handling before other events, etc08:25
dcalisteWhich is strange, since the gdb trace are clear and reading the code show clearly a possible fault corrected by the patch. But well, it's a race condition anyway, so very sensitive to context…08:25
chriadam_not a Qt event loop expert, but there are some complexities when dealing with application exit at least08:26
chriadam_I saw that pvuorela_ provided some comments on the outstanding PRs for the email signing feature08:27
chriadam_thanks for that ^08:27
dcalisteYes, I've begun to address them yesterday and will continue today. Thanks to him for the review ;)08:28
dcalisteHe pointed out some missing dependency issues in the packaging making the experience no so nice when some packages are missing. Need to further investigate.08:29
dcalistehello jpetrell.08:29
chriadam_no rush - hopefully once those review comments have been addressed, we can merge and tag everything08:31
dcalistejpetrell, pvuorela, as said in a comment in Bitbucket, I've slightly changed the Loader layout in account-setting, now the crypto settings are packaged separately in a sub-package.08:31
dcalisteInstalling this subpackage should pull the required dependencies from Sailfish Secrets and GnuPG.08:31
jpetrelldcaliste: cool. yeah pvuorela was talking about it, that would be cool to have one package that pulled in dependencies08:31
jpetrelldcaliste: yeah08:31
dcalisteIf this sub-package is absent, there is no crypto section at all in the setting page.08:32
dcalisteThe loader for it should fail silently not finding the source file.08:32
dcalisteAs pointed out in the comment in Bitbucket, this sub-package is for settings only and pull in only dependencies for the account setting, which means it's not pull in the QMF crypto plugins.08:34
dcalisteMaybe I should do something equivalent in the jolla-email part, having a sub-package for crypto stuff that would pull in the account setting crypto sub-package and the QMF ones.08:35
jpetrelldcaliste: pvuorela mentioned he had issues getting the email signing to work. signing is fairly technical domain, and we still lack some enablers to make managing keys (sync, maintain, import, etc.) as easy and transparent as it would need to be for ordinary people to succeed in setting it up08:36
jpetrelldon't know the details, but saw pekka commenting on pull requests08:36
jpetrelldcaliste: btw I have continued working on Documents app to get rid of the split view pattern. the design wants it to work more like Email Viewer, where document is scrolled vertically, and you back-step with simple swipe to right like with normal pages. actions (share, delete, details,  etc.) are shown on top of as overlayed controls like with newer Gallery viewers08:38
jpetrellI've done the split view removal on other apps so can do it quite quickly also for documents08:38
dcalistejpetrell: yes, managing keys is still not possible at a production quality, you have to have a working key for the email signing to work comfortably.08:38
jpetrellI remember you mentioned you also started similar work08:38
chriadam_jpetrell: initially, I think it's fine for some manual step to be required for use of the signing feature (to allow preliminary testing by power-users, without affecting the experience of "normal" users)08:39
jpetrellchriadam_: very true08:39
chriadam_so long as we somehow document that, and have a plan for how to go forward08:39
dcalistejpetrell: yes, see https://github.com/dcaliste/sailfish-office/commits/title08:39
jpetrellas such its not far from experience of doing signing on desktop linux, where you often do some bits from command-line interface08:39
jpetrelldcaliste: thanks, will have a look08:40
dcalistejpetrell: for email stuff, particularly if we put every thing behind sub-packages that power-users will to pkcon install by hand08:41
dcalistejpetrell: the branch was working well and implement all from design mockups that was available at that time, except that when zoomed in the page transition was not working anymore.08:42
dcalisteAt that time I didn't find the reason why…08:42
dcalisteAnd I put it aside because removing the drawer was very sensitive, making the link opening not working after rotation… (corrected since)08:43
dcalisteBefore working on it again, I wanted to make the regression tests work, see https://github.com/sailfishos/sailfish-office/pull/10108:43
dcalisteBut the regression tests themselves have some issues ;) As discussed with pvuorela at that time.08:44
pvuorela_lots of highlight here. hi!08:48
chriadam_is that something that could be split up into small chunks?  in recent community meetings, have seen some expressions of interest to help with coding.  maybe this is a good opportunity for someone to help, e.g. fixing some regression tests?08:48
dcalistepvuorela, hi! chriadam_, it's old code and I don't remember every bit, but I thing the issues with regression test were that they were too sensitive to numbers.08:51
dcalisteI mean, the regression test was mimicking device rotation, scrolling and we were testing if pointing at some place would trigger the link there after scroll or device rotation.08:51
dcalisteI don't remember exactly, but this imply some fuzzy comparison with too high bracketing for it to be reasonable.08:52
dcalisteThe problem with the regression test was that they play the scrolling, device rotation… at device speed, and some redraw maybe skipped and it's dificult to understand what happen.08:53
dcalisteI remember adding lot of sleep here and there but even like that some issues remains.08:53
dcalisteBut if people would like to try and contribute, yep that would be great. At least try to see if regression tests are always true or if they find some conditions when running the test make them false.08:54
dcalisteI can rebase the PR on github on master which would be a good strating point !08:56
chriadam_great, I will raise that at the next sailfish os community meeting08:56
chriadam_and see if we get volunteers to help with this effort.  good stepping stone, hopefully.08:56
chriadam_sledges: ^ topic for next meeting ;-)08:57
dcalistejpetrell: reading the commits for the 'title' branch, I remember that I try to showcase it first for the PDF document only, so the latest commit removing the drawer is doing not at DocumentPage.qml but directly at PDFDocumentPage level, which will have to be adapted for Office documents also.08:58
dcalistejpetrell: but the most sensitive commit is the previous one, the one introducing a header with the document title. It is sensitive because before all assumption on Y offset was based on top of page being 0, which is not the acse anymore and can introduce a lot of bug in context menu position calculations.09:00
dcalisteHopefully I did it right at that time, but that's why I wanted the regression tests to work at that time.09:00
chriadam_makes sense09:06
chriadam_good to have working test suite, in order to have confidence when making changes, definitely.09:07
jpetrelldcaliste: here is my first rough proto https://github.com/jpetrell/sailfish-office/tree/jb4107109:07
dcalistechriadam_: sure and this thing is very boring to check by hand at each modification, because you have to test at 0 zoom, zoomed in, after device rotation, after second rotation, checking that it should go back to initial position, that links open at the right place, if they are at the bottom, that they push the view a bit up…09:09
dcalistejpetrell: great looking at it…09:09
dcalistejpetrell: I will try it later this week. You may open a WIP PR in Github, I may comment there on code changes…09:14
dcalistelike import Sailfish.Silica.private 1.0 ;) But I like the DocumentOverlay unification, I'll look at it in action later.09:17
chriadam_dcaliste: I had nothing else - thanks again for your time and effort09:45
chriadam_if you need anything from me, or if I've forgotten something, please ping me :-)09:46
dcalistechriadam_, sure, I'm testing the regression test branch after rebase now.09:49
*** TheOnlyJoey|BNC is now known as TheOnlyBouncer16:43
*** svartoyg_afk is now known as svartoyg17:42
*** svartoyg is now known as svartoyg_afk19:05
*** SpeedEvil is now known as Guest9476921:58
*** BitEvil is now known as SpeedEvil23:03

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