On Sun 14 Nov 2021 at 19:35:57 (+0100), Thomas Schmitt wrote: > Greg Wooledge wrote: > > > unicorn:~$ sudo lsblk -o +UUID > > David Wright wrote: > > You shouldn't require root for either blkid or lsblk, > > though the former needs a path, and does include a > > warning that the information may be read from cache. > > This depends highly on the age of the system. > On Debian 8 you don't get much UUID from the system disks without the > permission to read them. > On Debian 10 lsblk seems to depend mostly on udev's assessment.
I should have mentioned that I'm using buster here. > > > ├─sda1 8:1 0 260M 0 part /boot/efi 4C30-7972 > > The universe must be small where this FAT UUID is unique. I don't think there's any need for it to be more than locally unique. I think I can live with odds of 1:4294967296. What's more important is that ID_PART_ENTRY_TYPE=c12a7328-f81f-11d2-ba4b-00a0c93ec93b (or 0xef on MBRs), which of course is anything but unique. > > One question about lsblk, though: why does it always > > use the xterm width to format the output, regardless of > > whether you redirect the output and the value of COLUMNS. > > For me on Debian 10 it does not care about the terminal size. > If i request a few sometimes lengthy fields from lsblk -h i get 206 bytes > per line (plus newline) printed to stdout: > > $ lsblk -o > NAME,KNAME,FSTYPE,LABEL,UUID,PARTTYPE,PARTLABEL,PARTUUID,MODEL,SIZE | head -1 > | wc -c > 207 Ah yes, you're right. The redirected output is always the minimum required for columnating the headings and data. It's the /screen/ output that I can't control, ie the amount of overlap employed is always that which just fills the lines in the xterm. > In the newest version of the man page > https://sources.debian.org/src/util-linux/2.37.2-4/misc-utils/lsblk.8/#L179 > i see an option which is not in the Debian 10 man page: [-w, --width number] Yes, not in buster's version at all. I must play with that in bullseye. Thanks. Cheers, David.
