libera/#devuan-dev/ Wednesday, 2021-08-18

fsmithred## google's nameservers01:41
masonI mean, on the plus side, it lets them do a better job building out our profiles so they can sell the data to companies who want to sell us things we're likely to buy. =cough=01:43
fsmithredwhy is it contaminating /etc/resolv.conf in live isos made with live-sdk?01:43
rrqthe squasfs needs to have something so a live run works without getting dhcp-offered servers01:45
rrqI would have thought they come from your build01:45
rrqtime for "fgrep -r /var /usr" on your build system01:47
rrqpossibly /etc as well01:48
fsmithredthis is what I get in a refracta iso:01:48
fsmithredcat mnt2/etc/resolv.conf01:48
fsmithredcat: mnt2/etc/resolv.conf: No such file or directory01:48
fsmithredwell, the system that was used to build those isos is gone01:50
fsmithredbut the chimaera isos were built on this box01:50
fsmithredwhere would someone connect that they don't get dns servers?01:50
fsmithredand I think network-manager would deal with it anyway. It loves to stick its fingers in resolv.conf01:51
rrqsomeone with static IP setup might "forget" setting up their DNS at the same time01:53
rrqit seems has found its way into a small handfull of binaries01:55
fsmithredyeah, I think I remember reading something about this a few months ago (or maybe in 2019)02:00
fsmithredchromium, t-bird and ff-esr are some hits02:01
fsmithreddirmngr? could that do it?02:01
fsmithredI still don't get it. live-sdk starts with debootstrap and then installs packages in chroot.02:02
masonOh, I remember it being included in an NTP setting.02:02
fsmithredwell, it may be difficult to track down, but I can fix it easily. Options are to either remove resolv.conf or populate it with a different address.02:04
rrq"If this file does not exist, only the name server on the local  machine will  be  queried; ..."02:07
rrqalso, it's typically offered be dhcp servers and will then be created02:08
fsmithredI think n-m and wicd also create it. I have one on a booted refracta iso, but it's not inside the squashfs.02:12
rrqafair wicd uses dhclient which would write resolv.conf upon receiving the dhcp domain-server option, I would guess n-m has the same; it's part of the "standard handling" in Linux of the DHCP protocol02:28
rrqthus, it's not the networking tools themselves, although they may also have code that messes with the file02:29
rrqn-m is a redhat thingy so probably does many things that it deems it knows best about02:31
fsmithredn-m messes with resolv.conf and leaves a comment saying it was there and is in control.02:34
fsmithredand I need a reality check, please02:35
fsmithred-rwxr-xr-x = 0755   does it not?02:35
XenguyThe last 3 digits are correct for sure, but I don't know about the first one, as I usually only chmod in 3's02:38
fsmithredempty is zero02:38
XenguyWhat's the first slot for?02:38
XenguyOr maybe I can look that up02:39
fsmithredsticky bit and something else02:39
fsmithredalso symlinks get 'l'02:39
XenguyOh yeah, /tmp and those sticky bits02:39
fsmithredreleasebot is claiming that the file is 1363, but when I look at it I see 075502:40
fsmithredrrq, can you make any sense of this error message about postinst?
bgstack15the first octet is for stick bit, setuid, and setgid.02:58
bgstack15I use this for reference:
Xenguybgstack15, Nice03:56
Xenguybgstack15, The last bit (pun intended) is confusing to me:  Basically, if it’s worth setting set-gid, you should throw in sticky bit.04:01
Xenguychmod 03770 thisdir04:01
XenguySeems like an orphan disconnected from the previous content, unless I'm missing some obvious logical connection04:02
XenguyPerhaps the '3' comes from '2 + 1' ?04:03
XenguyBut the initial '0' removes?  Not sure.04:04
bgstack15the leading 0 in that "03770" is a fifth octet which means even something different to GNU chown. And definitely 3=2+1 in base 8 arithmetic.04:29
adhocthe leading 0 means the following is octal?04:30
fsmithred3 would be write and execute, but not read04:38
onefangA leading 0 usually means "this is a octal".  Probably the only thing octal is still good for.05:46
rrqfsmitred: the octal numeral 1363 coinsides with decimal numeral 75506:44
bgstack15GNU chown always interprets the integer as octal, pretty sure.13:15
onefangThink you meant chmod?13:28
bgstack15Yes, that one.13:49
bgstack15chown/chmod; do you ever need one without the other?13:49
onefangThey change different things, you don't always need to change both.  shrugs13:50
bgstack15Yes, I just always need to modify both. I really should use install(1) more...13:51
gnu_srsWDYT about creating a Devuan GNU/Hurd distro, enabling non-merged-/usr as default (with merged-/usr as an option as for Devuan GNU/Linux)?18:35
masongnu_srs: I'd tend to think Debian GNU/HURD would be a lot like Devuan out of the gate, since systemd won't run there. I bet it'd be relatively low-hanging fruit.21:30
masongnu_srs: I've been mucking with vanilla Debian lately, and for the software I run, the most significant difference is udev vs eudev.21:31
gnu_srsmason: Still, Debian will not provide non-merged-/usr in the future.21:52
gnu_srsadditionally, I did try to get eudev into Debian but not succeeded yet :(21:53
masongnu_srs: I was wondering about that. It looked like the only missing piece. I didn't see an RFP. Is there anything to read out there?21:54
masonbrb, getting ready for an errand21:56
gnu_srsIt's an rfp
gnu_srsHaven't been working on it lately, due to a missing sponsor from debian-init-diversity :(21:58
* mason nudges LeePen 22:03
gnu_srsLeePen: I'm all for it. Seems to be like only changing the debian-installer defaults? And debootstrap23:34

Generated by 2.17.0 by Marius Gedminas - find it at!