Conversation

saleem

⍼

@saleemrash1d

Sep 12, 2021

there are two flowcharts, each with three steps from left to right.

the first is labelled "Maladaptive pattern" and goes "I run Arch Linux" then "I overwrite mkinitcpio.conf with mkinitcpio.conf.pacnew 2 months ago and today it wouldn't boot" then "Depression" (this step is in red).

the second is labelled "After cognitive restructuring" and instead says "I run NixOS" then "I could roll back to a configuration that worked" then "Everyone (deservedly) clowns on me for running NixOS" (this step is in green).

read image description

ALT

why do i keep experiencing consequences from this one mistake

Quote Tweet

current status: booting with init=/bin/zsh because i thought it'd be a good idea to clean up all the *.pacnew files but i accidentally ran my rename command in /etc and whacked my sudoers file with one that didn't include my user's group

Show this thread

Daniel Micay

@DanielMicay

Replying to

@saleemrash1d

Updates often result in data being migrated to new incompatible formats. There isn't a good substitute for having some form of checkpoints/snapshots. Android started using source.android.com/devices/tech/o in order to be able to roll back an A/B update even after mounting userdata as rw.

source.android.com

User Data Checkpoint (UDC) | Android Open Source Project

5:08 AM · Sep 12, 2021Twitter Web App

Replying to

and

It's inherently broken to be updating the currently running OS in-place in the first place beyond just being unable to roll back reliably. Typical software isn't designed/implemented to be able to handle the files being replaced with new versions one-by-one as it's running...

Replying to

and

It doesn't really mean having to go with the NixOS approach though. You mostly just need an active slot and an inactive slot. If you really wanted, you could use a traditional Linux distribution that way by separating your data partition and having 2 separate root partitions.