| Solrac[m] | Hello. Pardon me asking silly questions but, are are there ports for Sony edo devices (1II, 5II)? If so, what might their status be? | 04:30 |
|---|---|---|
| dcaliste | Good morning pvuorela, how are you ? | 07:07 |
| pvuorela | dcaliste: good, thanks. having some networking issues so just managed to get here. | 07:07 |
| dcaliste | Oh, I hope it will be fine, networking issues can be annoying. | 07:08 |
| pvuorela | indeed. something funny on my home connection and then networkmanager also got totally confused about it. but finally managed to switch to phone hotspot as backup. | 07:11 |
| dcaliste | Good, I've always got reliable connection through the phone when needing it, also. I was even surprised that Gnome somehow detected that it was a phone and restrained from downloading updates at that moments ! | 07:13 |
| dcaliste | May I ask, about the signing placeholder in email application, what do you prefer ? Option 1, put a placeholder as in the current MR, or don't display the icon in the email list ? | 07:13 |
| pvuorela | was left a bit undecided. recall the email list icon got added with the generic support. in a way if the crypto thing is now an add-on it could be consistent if it's not shown on the email list either. | 07:16 |
| dcaliste | Ok, I can do like that, the idea being just to be consistent. | 07:17 |
| pvuorela | suppose if the signature can't be verified it's not much of use. | 07:18 |
| dcaliste | Yes, the idea of the placeholder would have been to warn the user that the email was signed, but capability was missing to check it. It can be confusing I agree. | 07:19 |
| dcaliste | You may not have enough time to look at it properly yet, but what do you think about mkcal separation between calendar related functions and storage related ones in https://github.com/sailfishos/mkcal/pull/28 ? | 07:27 |
| pvuorela | indeed didn't have much time yet to check that. well, no objections at least so far :) | 07:30 |
| pvuorela | but it's changing a lot so need some time to digest. | 07:30 |
| dcaliste | Yes, I understand of course. Building on top, I'm trying also to implement a sort of ExtendedStorage where the calendar is in the main thread, while the sqlite part would live in another thread. | 07:32 |
| dcaliste | This leads me to think about the passing data structures. Currently the (new) StorageBackend and thus SqliteStorage is using ::Ptr as arguments. | 07:33 |
| dcaliste | But as we discussed, these are not thread safe, so one cannot call QMetaObject::invokeMethod() on methods with such arguments (the main thread may decrease the internal ref counter, while the worker thread is increasing it for instance). | 07:34 |
| dcaliste | So I'm trying to find a way to express in the API that data are main-thread owned or worker-thread owned. | 07:35 |
| pvuorela | alright. let's see how that works out. suppose it might be tricky if were also tied to calendarcore api. | 07:40 |
| dcaliste | I'm wondering if I should use ::Ptr for in-thread API and duplicate some routines with Incidence* for out-of-thread ? Is there some better ways ? | 07:40 |
| pvuorela | what sort of things? | 07:41 |
| dcaliste | For instance the BackendStorage::storeIncidences() is having ::Ptr arguments, but this means that I cannot call it in-between thread. | 07:43 |
| pvuorela | yea, not sure. one thing worrying me is how to keep the rules clear on what can be called from where. | 07:45 |
| dcaliste | Indeed, I would like to make this clear from the API. | 07:46 |
| dcaliste | I'm trying different things at the moment. I will push it as a WIP so you can give a look when I will have something. | 07:49 |
| pvuorela | alright | 07:50 |
| dcaliste | Is there a Qt data structure where one can transfer ownership ? The main problem with QSharedPointer is that if you create an object in one thread, pass it to another, then the destructor will still decrease the ref count in the first thread, and this is not thread safe. | 07:53 |
| dcaliste | Passing a simple pointer Incidence* and mentioning ownership transfer in the API doc is not very satisfactory neither. | 07:54 |
| pvuorela | well, at least QSharedDataPointer but not sure if it helps much if otherwise using calendarcore. | 07:57 |
| dcaliste | KCalendarCore is not enforcing ::Ptr. It is doing so when you are using calendars to store the incidences. But incidences themselves can be treated as normal objects, passed by reference or pointed to. | 07:59 |
| dcaliste | I've a commit that remove all ::Ptr from SqliteFormat for instance. And I'm wondering if I should do the same from SqliteStorage in general. | 07:59 |
| dcaliste | The ::Ptr make sense at ExtendedStorage level because of the calendar storing. | 08:00 |
| dcaliste | I've arrived to the conclusion that this ::Ptr API is making sense only for routines that would like to keep the data at hand, but for routine that are working on the incidence data for the time life of the routine only, this is quite detrimental. | 08:02 |
| pvuorela | i.e. often for methods would be better to just pass reference or pointer? | 08:05 |
| dcaliste | Yes, that would make more sense in term of API intension, and it would not force the use of such method for stored incidences. Because at the moment if you have routine with a reference to an event for instance, you cannot call any of the sqlite routines for instance because they need a ::Ptr and you cannot create a ::Ptr from a reference, while the sqlite routine will actually not keep any event data for later. | 08:08 |
| pvuorela | agree | 08:08 |
| dcaliste | That's why, I'm trying to keep the ::Ptr at the ExtendedStorage/Calendar level because they are keeping data for later, but I'm trying to see if I can design an API for the (new) BackendStorage that is using only const reference and pointers, so it can be used behind a thread or not. | 08:10 |
| dcaliste | It is still at early stage of development though, trying many things and trashing most ;) | 08:13 |
| pvuorela | let's see :) | 08:14 |
| dcaliste | Indeed, but the current PR separating ExtendedStorage and StorageBackend, making SqliteStorage inherit of the second instead of the first is helping to adapt the API, while trying to be backward compatible. | 08:15 |
| dcaliste | I'm just a bit annoyed in this PR by the observer breakage… | 08:16 |
| pvuorela | dcaliste: anything else for today? | 08:24 |
| dcaliste | No, I'll keep you informed about my advance in that direction, maybe in the form of a PR if it's good enough. And in the mean time, we can begin to discuss the current separation of API in the opened PR when you'll have time to look at it. | 08:26 |
| pvuorela | sure. | 08:26 |
| nephros-nc-bridge | [nctalk] <nephros> Can anyone tell me how one configures the connman dhcp server to serve a certain ip range on a given interface? | 08:43 |
| lolek | some time ago I asked about problems with qtsensors and accelerometer to be exact. I didn't got any replies so I'd like to know if sailjail may affect normal application that is started using command line? And if sailjail may restrict access to Accelerometer | 09:34 |
| dcaliste | As far as I know, if you don't run the application with sailjail from the command line, there will be not restriction coming from it. | 09:36 |
| lolek | hmm so I'm at the starting point again :( | 09:37 |
| Mister_Magister | mal: can i ask you about audioresource? | 09:46 |
| Mister_Magister | i remember you did openal with audioresource so you know at least something about audioresource xd | 09:46 |
| nephros-nc-bridge | [nctalk] <nephros> sailjail does apparently not affect offlineStoragePath from QQmlEngine. Is that by design, or could it be considered a bug? | 10:29 |
| poetaster | oh dear. as if I wasn't confused already. | 10:32 |
| nephros-nc-bridge | [nctalk] <nephros> To be more precise, offlineStoragePath always has the values from .desktop and they do not change when set from CPP. | 10:33 |
| poetaster | ah, code? | 10:34 |
| poetaster | Thaodan, I was going to start with fahrplan memory leaks here: https://github.com/smurfy/fahrplan/pull/269/commits/9578c88b377a2c497d59032a712c202e3cc86d33 | 10:35 |
| nephros-nc-bridge | [nctalk] <nephros> I made this https://github.com/nephros/sailjail_test/blob/master/README.md to see what changes when. storagePath is the only one that never changes. | 10:37 |
| x2s | Finally I've got the time to debug my carddav problem with nextcloud. First: [W] unknown:0 - CardDAV sync for account: 33 finished with error: 301 | 10:52 |
| x2s | Doesn't look good. Now to find the logs where it complains in detail, what is wrong... | 10:52 |
| x2s | "The "PROPFIND" operation requested on the remote content is not permitted" | 11:02 |
| x2s | That's the problem. | 11:02 |
| x2s | But why... | 11:02 |
| mal | Mister_Magister: what about it? | 11:10 |
| x2s | Is there a way to add files to a forum topic? | 11:17 |
| x2s | .oO( just use a paste website, x2s, and post the link. Don't overcomplicate things here! ) | 11:22 |
| Mister_Magister | mal: after i release it, i cannot reacquire it in same app. i have to stop the process and restart it | 11:50 |
| mal | can you show the code | 12:01 |
| Mister_Magister | mal: https://github.com/Michal-Szczepaniak/microtube/blob/remake/src/player/player.cpp | 12:02 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!