As I look upon `@world` `emerge` updates sometimes taking a while, I think that `btrfs` and/or `zfs` snapshots sure are convenient and necessary for such systems.
After all, what if the power went out for [20ms](https://en.wikipedia.org/wiki/Blinking#Central_nervous_system's_control) (long past what any ATX power supply is designed to cope with) while it's overwriting something important?
Then of course I remember that question isn't hypothetical and I did break an Arch Linux system in that exact way in the past. Binary packages may be faster, but they're nowhere near fast enough (<=11ms commit should work on compliant ATX supplies).
One could then think, okay, but before `btrfs` and `zfs` there were [LVM snapshots](https://web.archive.org/web/20260801154104/https://linux.die.net/man/8/lvcreate), right? Possibly, the feature was documented even back in 2007. Barely anyone talked about it though.
So what's with unix-likes mostly not having transactional updates (not that other OSes do much better, but it was expected for Windows to brick itself for basically no reason)? Is the expectation just that all users are privileged-enough to afford a UPS? Or to cope with arbitrary breakage when the power network has so much as a hiccup (or to be so privileged as to have reliable infrastructure)?
LisPi
npub132…qgzqx
2026-09-27 03:11:41 UTC
Author Public Key
npub1326y4p8mw4l750yc8e52n0xguvcl5j0dn88ul5r2gjva2sh2ay7qjqgzqxSeen on
wss://relay.momostr.pinkPublished at
2026-09-27 03:11:41 UTCEvent JSON
{
"id": "b0ad3e0f6d29b9c80052bc5fd613f46a183ac903123552ee45da447a41ce7ea2",
"pubkey": "8ab44a84fb757fea3c983e68a9bcc8e331fa49ed99cfcfd06a4499d542eae93c",
"created_at": 1790478701,
"kind": 1,
"tags": [
[
"content-warning",
"Gentoo thoughts and some I've had before"
],
[
"proxy",
"https://udongein.xyz/objects/42d62b53-9d9f-48b6-9cae-8d5a47ede6f9",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://udongein.xyz/objects/42d62b53-9d9f-48b6-9cae-8d5a47ede6f9",
"pink.momostr"
],
[
"-"
]
],
"content": "As I look upon `@world` `emerge` updates sometimes taking a while, I think that `btrfs` and/or `zfs` snapshots sure are convenient and necessary for such systems.\r\n\r\nAfter all, what if the power went out for [20ms](https://en.wikipedia.org/wiki/Blinking#Central_nervous_system's_control) (long past what any ATX power supply is designed to cope with) while it's overwriting something important?\r\n\r\nThen of course I remember that question isn't hypothetical and I did break an Arch Linux system in that exact way in the past. Binary packages may be faster, but they're nowhere near fast enough (\u003c=11ms commit should work on compliant ATX supplies).\r\n\r\nOne could then think, okay, but before `btrfs` and `zfs` there were [LVM snapshots](https://web.archive.org/web/20260801154104/https://linux.die.net/man/8/lvcreate), right? Possibly, the feature was documented even back in 2007. Barely anyone talked about it though.\r\n\r\nSo what's with unix-likes mostly not having transactional updates (not that other OSes do much better, but it was expected for Windows to brick itself for basically no reason)? Is the expectation just that all users are privileged-enough to afford a UPS? Or to cope with arbitrary breakage when the power network has so much as a hiccup (or to be so privileged as to have reliable infrastructure)?",
"sig": "d68b4f5da83bf563234d37164384df162a2bbf46255dd2cdda188114232ae9c6c698b6fa09956af10a6edf1da95fc156c737424783cdf0e4a947ae6e51a8f891"
}