ChanServ changed the topic of #robotnix to: Robotnix: https://github.com/danielfullmer/robotnix || Channel logs: https://logs.nix.samueldr.com/robotnix
<ajs124> well, my device can't seem to be convinced to do anything useful right now. any ideas how I could get my adb keys into it through the recovery or fastboot?
<ajs124> flashing the old version (2021.02.21.02) through fastboot doesn't even start adb
<danielrf[m]> I'm not aware of any way to get adb keys in easily
<ajs124> danielrf[m]: have you tried what happens when flashing vanilla over grapheneos or vice versa?
<danielrf[m]> btw, I wanted to ask, did you set `buildDateTime` on your build?
<ajs124> nope
<danielrf[m]> Like, you had already sideloaded the version built from ae32ae5ec112bd6e2ead1978ffa1c5ba8ca850b7
<danielrf[m]> and then tried sideloading the slightly updated version c7a54b5e2b338fe8a7b973593653fac555932e68
<ajs124> that version I installed through the ota updater
<danielrf[m]> but they both would have the same buildDateTime
<ajs124> ae32 through ota, c7a5 sideload
<danielrf[m]> kk
<danielrf[m]> I would have thought that adb sideload might have failed because you didn't update the buildDateTime
<danielrf[m]> the OTA updater definitely would have
<danielrf[m]> Could be worth a shot to rebuild with a newer buildDateTime and `adb sideload` again
<danielrf[m]> Never tried going between vanilla and grapheneos, but definitely would be surprised if something didn't break
<ajs124> sideload is complaining in a lot of interesting ways right now… if only so many apps didn't block backups, then I'd just nuke and restore.
<danielrf[m]> sorry you're encountering this! I think this is the first time we've had issues of this magnitude for a device someone regularly uses
<ajs124> no problem, I'm sure I'll figure something out somehow. although this really reminds me of what I hate about android devices. like, just give me a root shell somewhere somehow so I can look around and poke at things.
<samueldr> while grapheneos is cloer to vanilla than e.g. LineageOS is... I would assume too that it won't work to "dirty flash" any on the other
<samueldr> is closer to*
<samueldr> ajs124: did you try switching the slot?
<samueldr> not sure how you would do it
<samueldr> fastboot --set-active=_ # apparently
<samueldr> might want to identify the slot beforehand, so you're not blindly trying to do a no-op
<samueldr> fastboot getvar current-slot # I believe
<samueldr> but as danielrf[m] said, it's surprising it didn't reset to the previous slot after too many failed boots
<danielrf[m]> As we run out of other things to try, we could eventually build and flash a version with `variant="eng";`, which should allow you to get adb access (and root) without needing user interaction
<danielrf[m]> kinda destroyes the android security model, but eventually that would be the route I'd try
<danielrf[m]> after trying to swap slots, and another fresh build with an updated `buildDateTime`
<samueldr> (btw, I have nothing specific to robotnix or recent android builds to say, using generic android as a platform knowledge :))
<ajs124> samueldr: yeah, neither slot works. not sure how that happened, tbh.
<ajs124> danielrf[m]: I'll give those things a try, but it'll take a while, consindering I'm building these on my ryzen 5…
<danielrf[m]> After I finish the thing I'm currently working on, I want to see if I can reproduce your issue on my sunfish
<danielrf[m]> e.g. flash the broken build, then sideload the updated version without updating buildDateTime
<danielrf[m]> Just pushed vanilla-2021.03.04.01 and grapheneos-2021.03.04.01
<danielrf[m]> No need to update on grapheneos if you're not on sunfish
<danielrf[m]> vanilla also includes a fix for the OTA updater
<danielrf[m]> ajs124: Sorry, I was not able to reproduce your issue.
<danielrf[m]> ajs124: I did have a broken touchscreen on GrapheneOS with the previous tag
<danielrf[m]> ajs124: But sideloading the update (even without updating `buildDateTime`) fixed the issue for me
<ajs124> danielrf[m]: hm. I'm wondering how I managed that…
<ajs124> meanwhile, I've flashed an eng build and figured out it crashes with "java.lang.RuntimeException: There must be exactly one installer; found []"
<ajs124> which feels like something I've seen before, but don't remember how I solved
<danielrf[m]> ajs124: I'd be interested if you could upload the whole log somewhere
<danielrf[m]> that specific error message originates from frameworks/base/services/core/java/com/android/server/pm/PackageManagerService.java
<samueldr> different configuration of google services. vs. no google services?
<samueldr> IIRC if you freshly flash e.g. lineageos, with gapps, do stuff, "update" by freshly flashing lineageos again, but this time without gapps, it breaks things... it is only a total guess
<samueldr> though I did see someone with that exact error message in the google search results