| *** kimmoli_ is now known as kimmoli | 03:36 | |
| poetaster | Thaodan, Fahrplan I've almost fixed 2 further parsers. Need to modernize. | 07:25 |
|---|---|---|
| poetaster | piggz, I'm not maint. in harbour-multimodal, testing. | 15:23 |
| piggz | oops | 15:23 |
| piggz | done and in main too | 15:25 |
| poetaster | yipee! | 15:29 |
| poetaster | It was a bit of work to get the dev on board, but I'm paranoid about keeping up with the apis in fahrplan. | 15:30 |
| poetaster | a very elaborate backup plan. | 15:30 |
| poetaster | Funny thing, the deutsche bahn apis used in Fahrplan are ancient binary apis. no api tokens, fast, hopefully eternal. | 15:31 |
| Thaodan | I think that is the blessing and the curse about fahrplan. | 15:45 |
| Thaodan | the other uses efa which is quite broadly used | 15:46 |
| Thaodan | There's another oss public transport app called öffi while its jave based many of the backends could be easily ported. | 15:46 |
| Thaodan | poetaster: but back to bahn. From what I know the bahn has multiple apis some for internal use like the one for train station signs to shown the journey data. | 15:48 |
| Thaodan | Then they have ones that are publicly available and the one fahrplan uses is that one. | 15:48 |
| Thaodan | The API is used by many apps since the output is so good and works over whole Germany and partially also outside of Germany. | 15:49 |
| poetaster | Thaodan, yes there has been an explosion of apis: https://api.deutschebahn.com/store/ | 15:51 |
| poetaster | Thaodan, fahrplan currently lacks, but for the netherlands, json based parsers. | 15:51 |
| poetaster | Thaodan, I think the binary api will remain in service, since I believe it's used internally. | 15:52 |
| poetaster | Thaodan, I suspect it's not limited because nobody (except smurfy and co.) has bothered with binary parsing. | 15:52 |
| poetaster | muli-modal is using the news apis from that 'store', but he's also storing a lot of station data locally. | 15:53 |
| Thaodan | If you speak German there's this funny talk about the Deutsche Bahn's data: https://media.ccc.de/v/36c3-10652-bahnmining_-_punktlichkeit_ist_eine_zier#t=1322 | 15:54 |
| poetaster | Thaodan, heh. Habe ich 'live' gesehen :) | 15:54 |
| Thaodan | poetaster: Well I guess the issue is that to decide which is "canon". | 15:54 |
| Thaodan | Often those new APIs don't support everything. | 15:55 |
| poetaster | yeah, intentionally. 'business case prototyping' | 15:55 |
| Thaodan | A thing that's worth to look into is how others e.g. öffi or kpublictransport handle it. | 15:56 |
| Thaodan | If i would do again I wouldn't write it again e.g. multi-modal but use kpublictransport. | 15:56 |
| poetaster | I've been going through the kpublic code wanting to move backends in that direciton | 15:56 |
| Thaodan | Also the license is an issue for me. | 15:56 |
| poetaster | MIT? | 15:57 |
| Thaodan | business case prototyping sounds more like malicious beta testing. | 15:57 |
| poetaster | exactly! | 15:57 |
| Thaodan | Yes I'd prefer something that has copyleft .e.g. lgpl or gpl. | 15:57 |
| poetaster | kpublictransport backends is the direction I'd like to go. | 15:58 |
| Thaodan | However as that I think of kpublic transport I don't really like to reinvent backends. | 15:58 |
| Thaodan | Parsing is harder if there's no parser. | 15:58 |
| Thaodan | E.g. vrr backend was so easy to do. | 15:58 |
| poetaster | I tried GPLv3, but he's a bit alergic because of some previous dev encounters. | 15:58 |
| Thaodan | but writing a parser for the binary vrs data was such a hurdle for me. | 15:59 |
| Thaodan | poetaster: like? | 15:59 |
| Thaodan | VRS = Verkehrsverband Rhein Süd | 15:59 |
| poetaster | did that ever get merged? | 15:59 |
| poetaster | In a pique (smurfy wasn't answering) I split my repo because I was diverging too much. I've moved a bunch of merge requests into branches. | 16:02 |
| poetaster | Worked ok for ninetwo backend. | 16:03 |
| Thaodan | poetaster: I think he is not active anymore | 16:03 |
| Thaodan | poetaster: Are you sing tags for changelog generation? | 16:03 |
| Thaodan | *using | 16:03 |
| Thaodan | forking as the best idea I think | 16:04 |
| poetaster | Thaodan, no, because I'm too sloppy at present. I should though. | 16:05 |
| Thaodan | please do it makes this stuff seamless | 16:05 |
| Thaodan | tar_git takes care of that for you | 16:05 |
| Thaodan | Please remove %if chum from +BuildRequires: qt5-qttools-linguist. | 16:06 |
| poetaster | Thaodan, It can remain a dependancy in any case, or? | 16:06 |
| Thaodan | Locally in the SDK it won't break anything and just clearly specific the build dependencies. | 16:06 |
| poetaster | Thaodan, | 16:06 |
| poetaster | oops, sure thing. | 16:07 |
| Thaodan | It will break anyone building in a clean build env outside of chum. | 16:07 |
| Thaodan | e.g. me :D | 16:07 |
| Thaodan | np | 16:07 |
| Thaodan | btw good think that you fixed the calendar export | 16:07 |
| Thaodan | that bugged me for long | 16:07 |
| Thaodan | You removed pkgconfig(kcalcore-qt5) from the buildrequires, what is the new calendar feature using as a backend? | 16:08 |
| poetaster | done. | 16:09 |
| poetaster | no backend. just xdg. | 16:09 |
| Thaodan | ah interesting. | 16:10 |
| Thaodan | I assumed kcalcore or something. | 16:10 |
| poetaster | I have a version that goes that way, but pvuorela suggested, why not let it be handled higher up. | 16:11 |
| poetaster | so other calendar providers could step in. It also reduces dependancies and means I don't have to futz with priv. files. | 16:11 |
| poetaster | HOWEVER, the kcalcore method allows nicer formatting. | 16:12 |
| poetaster | Ah, looking at it now, I'd call it a hack https://github.com/poetaster/fahrplan/blob/master/src/calendar_sfos_wrapper.cpp | 16:14 |
| poetaster | In my defense, I've only been getting back into c++ in the past year and need a lot more practice. | 16:14 |
| Thaodan | Provider as in not just local calendars? | 16:15 |
| poetaster | In theory something else could handle vcal/ical | 16:17 |
| poetaster | non-trivial issues considering system wide cals and sailjail. I was hoping to a qml plugin abstraction to allow writing to arbitrary calendars and at the same time protect system relevant ones. | 16:20 |
| poetaster | s/hoping\ for/ hoping\ to/ | 16:20 |
| Thaodan | SFOS needs something like Akonadi I guess | 16:22 |
| Thaodan | transparent storage of different types of media | 16:23 |
| Thaodan | https://github.com/poetaster/fahrplan/pull/12 | 16:23 |
| Thaodan | poetaster: ^ | 16:23 |
| poetaster | Thaodan, thanks! | 16:23 |
| poetaster | Thaodan, I wonder if I should do a name change (fahrplan3?) and push it to harbour. | 16:27 |
| Thaodan | poetaster: You can ask to want to maintain the app I think | 16:27 |
| Thaodan | I did the same for commmuni | 16:27 |
| Thaodan | although I had the blessing of the authors | 16:27 |
| Thaodan | But since fahrplan is gpl anyway | 16:28 |
| poetaster | Thaodan, true, that was the same for the apps I took over from tobia. | 16:28 |
| poetaster | yeah, I'll ask smurfy. maybe he finds the time ... | 16:28 |
| Thaodan | poetaster: You can also just merge the commit instead of creating a merge commit. For simple changes its less clutter. | 16:30 |
| Thaodan | The PR will be still referenced. | 16:30 |
| poetaster | Thaodan, that was just my default 'looks at the github interface' behaviour. I also still haven't got a good squash strategy. | 16:31 |
| poetaster | Thaodan, I'm often way noisy with minor commits that I end up pushing when I should squash. | 16:32 |
| poetaster | shit. I just did, exactly, that. | 16:34 |
| Thaodan | I suggest to use branches to work on stuff | 16:35 |
| Thaodan | there you can be all dirty and rebase | 16:36 |
| Thaodan | Magit made git so easy for me that these things go so fast | 16:36 |
| poetaster | Thaodan, of course branches are key. master, hotfix, release, feature, feature ... | 16:37 |
| poetaster | so VRR is functional | 16:38 |
| Thaodan | I've been thinking of how to fix communi, either I notifications in harbour are less functional in harbour now or I find some hack. | 16:39 |
| poetaster | I just started looking at notifications for my stopmotion app today. I'm thinking of working around it. | 16:40 |
| Thaodan | poetaster: I think they are less used because there are not that many good git uis out there | 16:40 |
| Thaodan | poetaster: I have a workaround to but it doesn't work in harbour | 16:41 |
| poetaster | Thaodan, in my case the work-a-round is just like that calendar hack. Instead of a notification, when the file operation (video encode) finishes, open the stream . | 16:43 |
| Thaodan | You mean the file that the app handle? | 16:45 |
| poetaster | yes, not stream, file. | 16:50 |
| poetaster | Thaodan, but karry has a lower level library in the works and it's all just a mock up (with ffmpeg) at the moment. | 16:50 |
| poetaster | Thaodan, re: vrr shouldn't I find Solingen and Bochum?? | 16:51 |
| Thaodan | poetaster: ah ok that works nice for apps that use files, basic notifications work just led and notifications on lockscreen don't work from the qml side. | 16:53 |
| Thaodan | poetaster: It should | 16:53 |
| Thaodan | Last time I checked it did. | 16:53 |
| poetaster | Thaodan, yeah, I think there is an EFA version mismatch. I had to rework a number of them ( münich, ireland). | 16:54 |
| poetaster | Thaodan, yeah, for file's it's kind of 'easy' but I'm worried about the state of stuff at the dbus level. | 16:55 |
| poetaster | Thaodan, but it's a sandboxing issue, writ large. | 16:55 |
| poetaster | I'm too scattered to think clearly at the os architcture level at the moment. | 16:56 |
| poetaster | Thaodan, [D] ParserEFA::parseStationsByName:336 - Query Error: "name list" | 16:59 |
| Thaodan | poetaster: You can also rebase master to not include the vrr backend and we fix it first. | 17:00 |
| Thaodan | I assumed it still works fine. | 17:00 |
| poetaster | Thaodan, no worries. I'll do a bit of reading and get back to you. | 17:01 |
| Thaodan | okay | 17:02 |
| poetaster | Thaodan, I think it's the endpoint, I'll make a note and get back to you as I mentioned. | 17:06 |
| poetaster | Thaodan, the dbahn has a thousand defunkt endpoints that still work, in some sense. no wonder nothing runs on time. | 17:08 |
| poetaster | neat, the vrr interface alows you to switch between xml and json with a single flag. | 17:13 |
| poetaster | &outputFormat=rapidJSON | 17:14 |
| poetaster | Thaodan, I got it. It's further EFA changes. In the vrr case, I had to modify type_sf to any. | 17:20 |
| Thaodan | poetaster: sounds good | 17:36 |
| poetaster | Thaodan, there were a number of things implemented in the parent class that should'nt be there. time is a cruel mistress. | 17:38 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!