#AndroidAppRain at https://apt.izzysoft.de/fdroid today with 15 updated and 1 added apps:
- InkOS - Minimal Eink Friendly Launcher: nomen est omen 😉 a fork of mLauncher/oLauncher, optimized for eInk displays 🛡️
Current RB status: 637 apps (48.3%)
Enjoy your #free #Android #apps with the #IzzyOnDroid repo :awesome:
3 Comments
IzzyOnDroid@floss.social · 1 pts · 1y
Speaking of RB:
DavDroid 4.5.1 unfortunately failed RB. Which shows the thin line between "deterministic" and "reproducible":
We were able to build the app umpteen times, and got the very same, byte identical APK on each build: deterministic. So, it was reproducible, right? Well: no. It didn't match the APK built by the developer. A very slight difference in this case, an "off-by-one" in the baseline (so don't you worry, it's just the optimizer).
#reproducibleBuilds #IzzyOnDroid (1/2)
IzzyOnDroid@floss.social · 1 pts · 1y
(2/2) We could have claimed: "Reproducible!" – as we were able to reproduce the very same APK over and again, on multiple builders even. But: it was not the APK "upstream" shipped we reproduced, just our own builds.
Hence, if you wonder about the warning shield at DavDroid v1.5.1: We could not reproduce THAT APK. But, the difference was minimal, and not in the code itself: it was just the baseline profile, which differed in number of "hot" (important) methods: https://developer.android.com/topic/performance/baselineprofiles/overview
IzzyOnDroid@floss.social · 1 pts · 1y
And now you wonder why the shield turned green, right? Well: when after about 15 tries we always got the byte-identical APK, but not one matching the original, fate seemed to have been sealed. But I forgot to commit recipe & logs – and today, next to the "off-by-one" we got the "one time lucky", and succeeded reproducing the original APK.
Life's a bitch. And so are RBs, sometimes… Sometimes, they are like ferrets: you try to shoo them in front of you, but they come around and nip your calf 🙈