F-Droid Devs

133 Members
#fdroid-dev F-Droid development discussion only | Use #fdroid:f-droid.org for general, app- and repo-related matters | Meetings: https://gitlab.com/fdroid/wiki/-/wikis/Weekly-Office-Hours | This channel is publicly logged at https://matrix.f-droid.org/alias/%23fdroid-dev:f-droid.org23 Servers

Load older messages


SenderMessageTime
8 Nov 2022
@andreas:schildbach.deAndreasWe could just have one file per app which gets replaced for each version. History would still be available in git.22:10:53
@_oftc_jochensp:matrix.orgjochensp Andreas: I very much agree but we need to rework fdroidserver to make that happen 22:23:34
@_oftc_jochensp:matrix.orgjochenspcurrently fdroiddata defines which apps and versions are available in the client22:24:02
@grote:matrix.orggroteWow is this an actual repomaker user? https://fdroid.fedilab.app/23:27:58
9 Nov 2022
@rdfg77:kde.orglinsui
In reply to @_oftc_Izzy:matrix.org
linsui: what redirects? Any input needed from my end?
Git repo redirect. Do you check them with a script?
01:24:42
@_oftc_Izzy:matrix.orgIzzyIn a way. I have a script I use to find new apps for my repo. It ignores URLs already present in the indexes. So when an app is shown despite of being in the index, the URL must have changed…07:37:06
@_oftc_Izzy:matrix.orgIzzy Our repos have no issue with that AFAIK: my updater pulls fine despite the redirect, and our build servers build fine as well as far as I can tell. I didn't get if there's a real issue with that? 07:38:11
@rdfg77:kde.orglinsuiNo, not really. I thought it makes sense to add some lint for the urls.07:39:58
@rdfg77:kde.orglinsuiOr in the checkupdates code? When the repo is redirect it's updated automatically.07:40:48
@_oftc_Licaon_Kter[xmpp]:matrix.orgLicaon_Kter[xmpp] linsui: didn't get a chance to dig deeper into krita make issue. Buuuut, last time I added te `build:` this `export MAKEFLAGS=-j$(nproc)` it build it really fast, so pls add it asap so it 07:59:24
@_oftc_Licaon_Kter[xmpp]:matrix.orgLicaon_Kter[xmpp] ...gets ta be used in the next cycle, if I'm not done 08:00:16
@rdfg77:kde.orglinsuiTesting...08:02:47
@eighthave:matrix.org_hc
In reply to @andreas:schildbach.de
It's kind of weird that the base of the build process changes without a way to pin that dependency. All the old builds in my metadata files will never build again.
yeah, it would be nice to have base dependency management. I think there are no objections, someone just needs to implement it. For now, the way to do it would be to set up the buildserver based on an fdroidserver commit used to build the APK.
08:22:51
@rdfg77:kde.orglinsuiThen we should include the fdroidserver commit in the build metadata?08:23:33
@_oftc_jochensp:matrix.orgjochenspin the build log08:23:54
@eighthave:matrix.org_hcyeah, having it in the archived build log would be enough. Ideally, we'd have some kind of structured data, like Debian's .buildinfo to make rebuilding more automated.08:25:33
@eighthave:matrix.org_hc
In reply to @grote:matrix.org
Wow is this an actual repomaker user?
https://fdroid.fedilab.app/
nice, though I'd recommend to them to not make their repomaker instance publicly available, unless they mean it to be their offering.
08:25:46
@rdfg77:kde.orglinsui
In reply to @_oftc_Licaon_Kter[xmpp]:matrix.org
linsui: didn't get a chance to dig deeper into krita make issue. Buuuut, last time I added te `build:` this `export MAKEFLAGS=-j$(nproc)` it build it really fast, so pls add it asap so it

I still see gmake[4]: warning: jobserver unavailable: using -j1. Add '+' to parent make rule. with

    build: MAKEFLAGS=-j$(nproc) packaging/android/androidbuild.sh -p=all --src=$(pwd)
        --build-type=Release --build-root=$(pwd)/build --ndk-path=$$NDK$$ --sdk-path=$$SDK$$
        --android-abi=arm64-v8a

Is it necessary to export it?

08:28:09
@_oftc_Licaon_Kter[xmpp]:matrix.orgLicaon_Kter[xmpp] linsui: yes, but it built in 1h on the 22 core VM, while without it yesterday I killed it after 3h when I realised I forgot to put that fix 08:29:41
@_oftc_Licaon_Kter[xmpp]:matrix.orgLicaon_Kter[xmpp] I've read about + and +(MAKE) will see 08:30:27
@rdfg77:kde.orglinsuiIt seems like gcc148 is much slower that gcc 147.08:54:10
@eighthave:matrix.org_hcthey're the same hardware AFAIK, so I imagine it depends on what people are running on them. Or maybe a KVM misconfig.08:54:44
@eighthave:matrix.org_hcyou can have root on it if you want to look around.08:55:01
@rdfg77:kde.orglinsuiNot sure what I should look...08:55:23
@eighthave:matrix.org_hc

On a separate note, I just saw a nice side benefit of proactively disabling SHA1 (https://gitlab.com/fdroid/ansible-role-secure-sshd) SSH bots fail earlier:

sshd[1685297]: Unable to negotiate with x.x.x.x port 12345: no matching key exchange method found. Their offer: diffie-hellman-group1-sha1,diffie-hellman-group14-sha1,diffie-hellman-group-exchange-sha1 [preauth]
08:56:06
@_oftc_Licaon_Kter[xmpp]:matrix.orgLicaon_Kter[xmpp] linsui: `w` and `last|head` and `htop` ? 09:01:41
@rdfg77:kde.orglinsuiLooks like everything is normal. But after half an hour the build on gcc148 bubu is still downloading tarballs...09:04:19
@rdfg77:kde.orglinsui I still see gmake[4]: warning: jobserver unavailable: using -j1. Add '+' to parent make rule. with the export. 09:05:05
@_oftc_Licaon_Kter[xmpp]:matrix.orgLicaon_Kter[xmpp] linsui: I've already answered that :) 09:06:14
@rdfg77:kde.orglinsuiAh, I didn't understand your yes.09:06:58

Show newer messages


Back to Room ListRoom Version: 6