#fdroid-dev

/

      • [gibot] joined the channel
      • ShapeShifter499 joined the channel
      • relan joined the channel
      • relan
        Many apps in Requests For Packaging cannot be added due to JARs, AARs and other executable blobs. Anyone minds adding a "includes-binary-builddep" tag?
      • pserwylo
        hey relan
      • i wonder if that is better targeted towards a ubuntu-multiverse-style repo, which is disabled by default
      • relan
        pserwylo, I don't like the idea of a switch that turns F-Droid into a proprietary cesspit.
      • I know that krt has a different opinion, though.
      • Anyway, I'm talking just about triaging in rfp.
      • Besides, some projects with binaries included do replace them when we ask. I'd prefer to work with them to resolve those issues, not just pushing them into a ghetto.
      • NicoAlt joined the channel
      • pserwylo
        relan: oh, i thought you meant a tag, as in an antifeature. Never mind then :)
      • relan
        Yeah, I definitely don't want such dubious apps in the main repo. :) Sorry for confusion.
      • ShapeShifter499
        hi
      • larjona has quit
      • I have a suggestion: F-Droid should have a away to sort by repository so I can see what each repository has
      • larjona joined the channel
      • _hc joined the channel
      • CiaranG has quit
      • CiaranG joined the channel
      • krt[m]
        relan: see includes-binary-builddep
      • relan
        thanks, will use it!
      • krt[m]
        you're welcome. if you need anything else or if you have a better description, just ping me
      • _hc
        I was thinking that we should include just the repo signing keys for lots of repos in F-Droid, then they would be trusted by default, but they wouldn't show up in the repositories screen
      • so if a user adds them, the key will already be in place and trusted, instead of Trust On First Use
      • then we also make adding repos really easy
      • then we don't need to include anything non-free by default
      • pserwylo
        ShapeShifter499: Agreed. Check out c#966 for some ideas that I've floated for the future.
      • NOTICE: [client] #966: Sketch for improved "Add Repository" experience - https://gitlab.com/fdroid/fdroidclient/issues/966
      • _hc: finally here for the thursday meeting! in 10 mins or 70 mins?
      • _hc
        70 minutes
      • pserwylo
        cool, i'll pop back in before then.
      • _hc
        cool, glad you can make it!
      • I'm looking at !534 now
      • NOTICE: [client] !534: Tests and slight improvements for multi-sig suppo… - https://gitlab.com/fdroid/fdroidclient/merge_requests/534
      • FYI the test failed..
      • pserwylo
        thanks, might have some time to address any feedback.
      • i'll go investigate the test thanks
      • _hc
        just retry connected24
      • :-|
      • gitlab-ci has just totally fallen down running that
      • pserwylo
        nah, this isn't a connected24, this is the good + trusted tests :(
      • ShapeShifter499
        pserwylo: Your suggesting a preview mode before the repo is added? I'm talking about being able to "sort" the list of apps in F-Droid by repository
      • like you can sort by type/category, I sometimes want to be able to sort by repository
      • pserwylo
        ShapeShifter499: one of the ideas (don't think it is in that issue I linked) is to add each repo as a pseudo "category" on the categories screen.
      • ShapeShifter499
        yea that would work
      • pserwylo
        needless to say, we all know that the UX around adding/browsing specific repos is pretty sucky right now :(
      • ShapeShifter499
        yea agreed
      • pserwylo
        _hc: Wait, it was a connected24 test. Originally i broke the good tests, then pushed a fix before I left for dinner, now it is that sad connected24.
      • _hc
        yeah, its frustrating. the LocalizationTest needs to run there, and its been really helpful recently
      • catches a lot of crashers that we'll never see since we never use fdroid in all the languages
      • pserwylo
        "Found 1 connected device(s), 0 of which were compatible." Hmm, I wonder what it is about the device which makes it not "compatible" with the test runner?
      • at least it is recognizing the emulator, that is a start I guess
      • I might go try and fire up gitlab-runner and see if I observe the same thing
      • _hc
        pserwylo: don't bother, its a glitch in their runner
      • its intermittent
      • retrying fixes it
      • its also a known issue out there for other pepole
      • pserwylo
        okay, so I should indeed retry the test until it is successful then. The last few small MRs I merged I didn't bother with it. Obviously it is useful, but I wasn't sure how broken it was.
      • _hc
        pserwylo: the PACKAGE_REMOVED for org.fdroid.fdroid coming so late is totally bizarre (!530)
      • NOTICE: [client] !530: Fix incorrect persistent + annoying notification … - https://gitlab.com/fdroid/fdroidclient/merge_requests/530
      • pserwylo
        indeed, I can't think of a reason why. But I'm quite confident that is what is happening.
      • _hc
        hmm, if its a "sticky" intent, then Android will hold onto it until F-Droid is running again
      • pserwylo
        the example logging I put in the MR is a paraphrasing of the actual logging I had.
      • _hc
        so perhaps its just slow to deliver it then
      • pserwylo
        or it intentionally buffers them, because it knows there is an app with a broadcast receiver listening to package manager, in the process of being restarted.
      • _hc
        yeah, that would explain it
      • I just had a though about !530
      • there seems to be nothing preventing F-Droid from being added multiple times. InstalledAppProviderService.insert() does not prevent duplicates from being added to the database. If F-Droid ever receives a PACKAGE_ADDED about itself, it'll add a duplicate entry to the database.
      • NOTICE: [client] !530: Fix incorrect persistent + annoying notification … - https://gitlab.com/fdroid/fdroidclient/merge_requests/530
      • pserwylo
        I'm not convinced it ever did receive a PACKAGE_ADDED. The whole thing is very strange to me.
      • _hc
        I put more comments inline in !530 I'm going to grab lunch now before the meeting
      • NOTICE: [client] !530: Fix incorrect persistent + annoying notification … - https://gitlab.com/fdroid/fdroidclient/merge_requests/530
      • pserwylo
        _hc: thanks, I'll go take a look.
      • _hc has quit
      • NicoAlt joined the channel
      • NicoAlt
        Hi there!
      • pserwylo
        Evening.
      • NicoAlt
        pserwylo: I hope to be able to review your MR(s) in the next days because I'm away the whole next week. Again, I'm still astonished how good it works.
      • pserwylo
        don't say that, you'll curse it and then it wont work :)
      • I don't think it is going out with the website launch, so even if it lives there for a week it isn't the end of the world.
      • NicoAlt
        OK, cool.
      • pserwylo
        FYI _hc is back soon for the meeting. Just getting lunch as per the logs above
      • NicoAlt
        Ah, OK, thanks
      • uniq[m]
        hey everyone thezero_ Tovok7 cdesai NicoAlt mimi89999_0 CiaranG
      • _hc will join us shortly
      • NicoAlt
        uniq[m]: Hi, thanks for the info.
      • _hc joined the channel
      • _hc
        sorry, computer troubles, I'm here
      • and read the log
      • NicoAlt
        Hi :)
      • There isn't much log yet.
      • _hc
        sounds like pserwylo started already :)
      • pserwylo
        well, not really, but I can start...
      • _hc
        go!
      • pserwylo
        most of my time was spent on adding search to the website.
      • any feedback is welcome in the single MR in fdroid-website.
      • NicoAlt
        *jekyll-fdroid, or?
      • pserwylo
        right, jekyll-fdroid, thanks NicoAlt
      • The search is dependent on JS, as is of course required for a static site.
      • NicoAlt
      • pserwylo
        there is quite a lot of moving parts, but the seem to gel together quite well in the end.
      • i would have liked to use an upstream plugin jekyll-lunr-js-search, but it was too speceific to the notion of searching "pages" and "posts".
      • So i ended up including some key classes from that project and adapting them to "packages" and "package summaries"
      • anyway, the rest of my week has been on ironing out kinks in the F-Droid client, for which a few MRs have been submitted, reviewed, and merged.
      • the last was about adding better support for multiple apks of the same version, with different signatures.
      • it is not fully supported yet, but a few more small MRs should be enough to get it working.
      • _hc: is there any app you have in mind I could use to test multi-sig stuff?
      • either in F-Droid, testy.at.or.at, or other public repos?
      • NicoAlt
        We should do a blog post about the possibility to use developer signatures/reproducibility, btw
      • pserwylo
        finally, _hc and I have a difference of opinion about enforcing line lengths in F-Droid client. I don't know how interesting/uninteresting that is to anyone here, but I'm happy to go with the majority if you just drop a comment or vote in !532.
      • NOTICE: [client] !532: enforce max length 118 - https://gitlab.com/fdroid/fdroidclient/merge_requests/532
      • _hc
        pserwylo: doesn't really matter, I have just been using the APK from gplaycli and fdroid
      • pserwylo
        seems like the right place ask in a dev meeting :)
      • _hc: fair enough.
      • _hc
        NicoAlt: yes, but first we need to get the details all worked out
      • pserwylo
        so that is all from me, happy to answer any questions about app search in website, or client improvements.
      • otherwise that is all.
      • NicoAlt
        pserwylo _hc: We have a length limit of 100 chars in Repomaker, fyi.
      • pserwylo
        oh, uni is finished now, so I'll probably be doing almost 5 days a week for the next few weeks.
      • NicoAlt
        I can go afterwards then, if nobody else wants.
      • _hc
        for me, its not about line lengths, its about what the tools we use expect
      • pserwylo: no public repos have APKs with duplicate packageName/versionCode because all existing versions of the client would throw an exception when parsing the index into the database
      • NicoAlt
        We use pep8 to check the style, which has the default of 79 chars ^^ https://www.python.org/dev/peps/pep-0008/#maxim...
      • _hc
        that means that all 0.103 versions will be broken once public repos have such APKs
      • index.xml will never allow more than one APK for a given packagename/VersionCOde
      • so 0.102 and older will always work
      • pserwylo
        _hc: right, so that is what you were refering to in our last meeting (removing duplicates from fdroiddata or something?)
      • _hc
        yeah
      • I added some to testy, but just got the exception
      • NicoAlt: I guess then people can vote for line length enforcing with the thumbs up/down on the merge request
      • NicoAlt
        OK, will do later then.
      • pserwylo
        _hc: I'd also like to bring c#1052 to your attention for later on when you have time.
      • NOTICE: [client] #1052: Index-v1 sets suggestedVersionCode instead of ups… - https://gitlab.com/fdroid/fdroidclient/issues/1052
      • now I'm really done :)
      • NicoAlt
        _hc: Can I start?
      • _hc
        pserwylo: I saw that
      • yes!
      • NicoAlt
        As usually I continued to work on Repomaker.
      • We successfully made the UI Test Milestone last week: https://gitlab.com/fdroid/repomaker/milestones/1
      • I finished quite some styles in RM, like the share and info view in repos.
      • Also I improved the functionality to add apps from remote repos.
      • It is now done via Ajax requests in the background, if JS is available.
      • Another big thing was drag and drop which works now in repo and app edit view.
      • Today I fixed some minor bugs/style issues, improved the app adding with JS and added the ability to remove repos.
      • Torsten also did quite some stuff, here's an excerpt:
      • We now have the F-Droid.org repo as a default remote repo in RM which gives the users the possibility to choose from over 2000+ apps. That was really great!
      • He revised the whole SCSS styles, improved the process of background tasks and fixed a lot of other tinier things too.
      • He's currently working on making RM accept other media types besides apps.