poetaster | attah, seems to be the case that behaviour of QImage.autotransform has changed from 3.4 to 4.4. I'm going to go through each step again to be able to present it thrusday. | 06:37 |
---|---|---|
poetaster | Flohack, thanks, I will do! | 06:37 |
poetaster | attah, where do you STORE 75 thermometers! | 06:53 |
attah | poetaster: just be sure to not discard any other differences between versions tested (you mentioned 32 vs 64 bits yesterday, right?) | 07:02 |
attah | on my kitchen table currently | 07:03 |
dcaliste | Good morning pvuorela. | 07:05 |
pvuorela | dcaliste: hey | 07:05 |
pvuorela | dcaliste: sorry again. been quite busy wrapping up things before my vacation and haven't gotten too much time to process your PRs. | 07:06 |
pvuorela | dcaliste: the silica one i was about to merge any time now. | 07:06 |
pvuorela | and think the import filtering was good too at least. | 07:07 |
dcaliste | I think so, no problem. Thanks anyway for what you can achieve. | 07:07 |
dcaliste | I restarted my tablet and play a bit with the calendar tab design. Without anything convincing yet though. | 07:10 |
pvuorela | excellent that you have it available. | 07:11 |
dcaliste | The current one implies a large day number with a background item with the month and year for selection. | 07:11 |
dcaliste | It match quite well with the current tab design which also has a month + year background item for selection. | 07:12 |
dcaliste | But as I have designed it at the moment, this month selection is linked to the tab header while the large day on tablet is linked with the month view only. | 07:13 |
dcaliste | I would have like to write something that is not month view specific though but didn't achieve something convincing yet. | 07:14 |
pvuorela | yea, no immediate suggestions from me either. | 07:16 |
dcaliste | The alignment is a bit boring also, it is two columns on landscape small screen and 2 columns on both portrait and landscape for large screens, but the header should be on first column in landscape large screen and in the second column in other cases... | 07:17 |
dcaliste | If creates a lot of 'ifs' that make the layout very difficult to read. No beautiful solution at the moment neither. | 07:17 |
dcaliste | I have a commit though that is not that nice design-wise, but works very well : it removes the background item from the large screen under the large day number. | 07:19 |
dcaliste | All the rest is unchanged. | 07:19 |
dcaliste | It may be a reasonable compromise if including a dedicated tab design takes too much time. | 07:19 |
pvuorela | what's the problem with the background item? | 07:20 |
dcaliste | I don't think I pushed it yet though, but I can do it today so you can have a look. | 07:20 |
dcaliste | The background item in large screen (the one under the large day number) has became redundant with the tab header one. | 07:20 |
dcaliste | If you run the current PR on tablet, you'll see that the layout is somehow correct but you have this month + year info in two places which looks a bit strange. | 07:21 |
pvuorela | pondering if that's a problem. | 07:21 |
dcaliste | If you find the time, you can try the current PR on tablet. Besides this redundancy it is working quite ok. | 07:22 |
pvuorela | above about access to month change page. | 07:22 |
pvuorela | i'll try at least changing the sizecategory handling manually. i have some tablet(s) but they've been breaking apart. | 07:23 |
dcaliste | I can add a screen shot in the PR for you to see... | 07:24 |
pvuorela | that would be nice. | 07:24 |
dcaliste | If we keep the current month selection in DatePickerPanel, I can simplify the code to use the YearMonthMenu from silica there also. | 07:27 |
pvuorela | dcaliste: hm, now having crash again changign to day view. wonder if the same old layout thing. | 07:31 |
dcaliste | Arf, I cannot get it myself :/ | 07:36 |
pvuorela | dcaliste: heh. added some debug to setHeight() and seems it first gets -32, does updateEarlierAndLater() which crashes. | 07:40 |
dcaliste | Ah, I'm stupid, I put the "if (m_cellHeight < 1)" for quick return in layout(), but didn't notice that updateEarlierAndLater() can be called directly... | 07:42 |
dcaliste | My bad. Of course, I'll add a safe guard there also. | 07:42 |
pvuorela | thanks | 07:47 |
dcaliste | Ah bug of screenshot on tablet ;) Image are empty... | 07:49 |
pvuorela | ah of course that one. | 07:50 |
pvuorela | dcaliste: hey and for the next time. i could be ok 9th of august or then week after that. or later. | 07:51 |
dcaliste | It will be the week after, I'll still be abroad on the 9th. | 07:54 |
pvuorela | dcaliste: ok, better for me as there can be some catching up. 16th then. | 07:55 |
dcaliste | I put photograph screenshots of the tablet layout rendering of the current implementation in the PR. So you can get an idea. | 07:55 |
pvuorela | dcaliste: ah, that way. yea, the redundant month label for example seems a bit off. | 08:00 |
dcaliste | One quick solution is thus to simply removed it. | 08:01 |
pvuorela | dcaliste: pondering also if the big label should be on the right side on both orientations for large screens. | 08:04 |
pvuorela | perhaps the tab controls could be more right aligned on tablet too. | 08:05 |
dcaliste | Do you mean right aligned in landscape for the large day number with the right of the calendar ? | 08:06 |
dcaliste | The best layout may be also to have no background item in the tab header for tablets and use the current background item ? | 08:08 |
dcaliste | The tab icons would then be centred aligned if there is no tappable labels on its right. | 08:08 |
pvuorela | hm, some option. but maybe continue this on the pr side, i need to finish the meeting soonish. | 08:10 |
dcaliste | Sure, so do I. Thanks for the feedback and enjoy your vacations to come. | 08:11 |
pvuorela | thanks, you too | 08:14 |
*** Dima2387 is now known as Guest4241 | 08:27 | |
poetaster | attah, I'll detail the differences 3.4 vs. 4.4. As I said, Volla 3.4 is an armv7 build, 4.4 is aarch64. the 4.4 behaviour is identical volla/Xperia 10ii | 09:21 |
attah | :) that narrows its down anyway | 09:25 |
poetaster | attah, indeed. I just can't find where in: Label the richttext rendering does loadimage nor in canvas, for that matter. should be in the qtquick sources? | 09:38 |
poetaster | attah, the issue within richtexts is constant from 3.4 to 4.4 | 09:39 |
poetaster | context2d | 09:41 |
attah | wait what... are you saying the build engine used yields different results on the same phone? | 09:44 |
poetaster | one GS290 with arm32 bit 3.4, one GS290 with aarch64 4.4. that's how piggz built them. | 09:51 |
attah | That's not how it reads... but maybe i'm just being thick | 09:53 |
poetaster | sorry, you mean in my poorly worded bug reports? | 09:53 |
poetaster | https://github.com/sailfishos/qtdeclarative/blob/003a2cbafe7d5c036fd2e6b32b9aaa1bd447dc86/src/quick/items/context2d/qquickcanvasitem.cpp#L1001 | 09:54 |
piggz | poetaster: wasnt volla always aarch64? | 09:55 |
poetaster | piggz, well, maybe :) But I always deployed arm to it till the 4.x versions. Was it maybe multi lib? In any case, I'm still building arm32 bit vor the 3.4 | 09:56 |
attah | Wasn't it until 4.x we started being even close to having 64-bit userspace? (as opposed to kernel) | 09:58 |
poetaster | piggz, I'm wrong: 3.4 Linux VollaPhone 4.4.146+ #16 SMP PREEMPT Sun Nov 22 10:20:28 UTC 2020 aarch64 GNU/Linux | 09:58 |
piggz | thought so ... i dont change arch for a port :) | 09:59 |
poetaster | it's still 32 bit userland libs though. | 10:00 |
piggz | is it? | 10:00 |
poetaster | ah, let me do a test. | 10:00 |
piggz | look in /etc/zypp config files | 10:01 |
poetaster | arch = armv7hl | 10:04 |
poetaster | piggz, an perspective on a camera fix? otherwise 4.4.0.64 is stable. and the volla 22, any progress? | 10:08 |
poetaster | network on 4.4 is sucky (timeouts on multiple requests in tight loops) in comparison to 3.4. Requires a whole new level of error correction foo. but that's yet another rabbit hole. | 10:10 |
piggz | cant be as bad as the networking on v22! | 10:15 |
poetaster | ah, networking is crap on the xperia 10ii, too. something in the network stack changed. but I can't even find this friggin QImage bug, ... | 10:16 |
poetaster | I've been collecting test cases for network. in c++ and python. oddly enough, it's worse in c++ | 10:17 |
HengYeDev[m] | what does `# This is about access to device's camera hardware, not data` mean in https://github.com/sailfishos/sailjail-permissions/blob/master/permissions/Camera.permission ? | 15:39 |
HengYeDev[m] | is webview camera not working for anyone else? tried with sandboxing=disabled also | 15:47 |
Tomin_ | HengYeDev[m]: it means that to save pictures you need another permission. Sandboxing=Disabled is case sensitive. If it's correct there isn't permissions in Settings -> Apps -> the application page (or no page at all more likely) | 16:11 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!