Replying to
@Edent or in a human-readable date format, -rwxrwxrwxrwxrwxrwxrwxrwxrwt
Remote
sluttier lee edelman
@lw@mastodon.bsd.cafe
mastodon 4.7.3unfortunately, computers
213 Followers
53 Following
50 Posts
Joined May 04, 2024
i see people fleeing GitHub for Codeberg, which is understandable, but Codeberg is only one good DDoS away from being down for a week.
the fix for this isn't to move from one provider to another, it's for people to host their own projects, or ask a friend to host it. Git has a built-in email workflow that works great, you do not need "pull requests" to accept contributions.
tfw you just regurgitate your LLM slop on a mailing list like it has any value at all https://lists.freebsd.org/archives/freebsd-hackers/2026-July/006651.html
i think we should ban people from the lists for this shit
fun networking fact: routing tables don’t actually do anything. they’re a sham.
packets are ferrous and are moved around by electromagnets. the reason so many packets go to Google and Facebook is because they can afford the largest magnets.
so-called “AI datacentres” are a cover for even larger electromagnets, which is why they need so much power and water. (the water is used to filter the packets once they arrive.)
some areas of the country are reporting up to 25% packet loss compared to this time last year, and once those packets are lost they can take decades to regrow.
and all of this is being subsidised by YOU, the American taxpayer!
TIL: the FreeBSD pkg(8) was written by @_bapt_@mastodon.social.
the original plan was to call it "bapt-get".
"A missing underscore sent innocent man to prison for 18 months": https://arstechnica.com/tech-policy/2026/07/police-missed-one-underscore-and-sent-the-wrong-man-to-prison/
wow... could it be that the problem with the police is not just "a few bad apples" but the entire system that allows this?
surely not, because "honest cops" would never send someone to jail for two years for a crime they didn't commit.
fuck the state
fuck idiots who defend it
GPL-licensed code still in FreeBSD:
* gcov, https://cgit.freebsd.org/src/tree/sys/gnu/gcov
* bwn phy, https://cgit.freebsd.org/src/tree/sys/gnu/dev/bwn/phy_n
if you try building a GENERIC kernel for powerpc.powerpc64, you get an extremely unhelpful error:
--- all_subdir_accf_data ---
ld: error: accf_data.o is incompatible with elf32ppc_fbsd
*** [accf_data.kld] Error code 1
i feel like we should be able to do better here... somehow.
#freebsd
Replying to
@lucasmz although, perhaps this is related to the host itself having a ULA, so it prefers the IPv4 address. but this doesn't seem to be configurable in musl :-/
on FreeBSD it just works, since it realises the host has no IPv4 connectivity.
bought a new power amp and it came with an entire booklet full of warnings:
- Danger to life due to electric current! (from high voltages)
- Danger of life due to electric current! (from short circuits causing a fire)
- Risk of injury and choking hazard for children!
- Risk of death from electrical current!
- Possible hearing damage due to high volumes on speakers or headphones!
- Damage to the device due to high voltage!
- Risk of fire due to installation of a wrong fuse!
- Risk of fire due to covered vents and neighbouring heat sources!
and yes, all of them have an exclamation point at the end.
i'm wondering if i should return it, because it seems i'll be flirting with death every time i turn the thing on.
in case anyone was wondering whether PIM-SM works on RouterOS nowadays... no, it still doesn't.
pretty sure this has been broken since 7.0.
/bin/sh is a bad programming language, exhibit no. 94:
set -- $(nonesuch)
echo $?
-> 0
x="$(nonesuch)"
echo $?
set -- $x
--> 127
accidentally read a hacker news comment and now i need some sort of brain bleach
Replying to
GNU’s definition of “free software” is directly rooted in capitalism, because from the start, they envisaged companies would pay people to maintain GPL software. Big Tech is literally the FSF’s business plan.
well, now we have that. does anyone think Red Hat, or Oracle, or Canonical is an improvement over Sun, HP or DEC?
let’s reclaim the term “free software” and give it an actual meaning.
latest weird problem: on FreeBSD/ppc64, running as a KVM guest on Linux, 'tcpdump -ni vtnet0' doesn't show any packets.
the network is working fine, the packets just don't appear in tcpdump.
how strange.
coming soon, support for cross-building #freebsd pkgbase from Linux:
% uname -srm
Linux 6.19.12-200.fc43.x86_64 x86_64
% ../build.sh
==> Building for amd64.amd64.
bmake: /home/lexi/src/freebsd/src/dev/share/mk/jobs.mk:40:
@ 1777264542 [2026-04-27 05:35:42] Start buildworld-jobs
@ 1777264542 [2026-04-27 05:35:42] Start buildworld -j10 log=/home/lexi/src/freebsd/src/buildworld.log
@ 1777264586 [2026-04-27 05:36:26] Finished buildworld-jobs seconds=44
bmake: /home/lexi/src/freebsd/src/dev/share/mk/jobs.mk:40:
@ 1777264586 [2026-04-27 05:36:26] Start buildkernel-jobs
@ 1777264586 [2026-04-27 05:36:26] Start buildkernel -j10 log=/home/lexi/src/freebsd/src/buildkernel.log
@ 1777264592 [2026-04-27 05:36:32] Finished buildkernel-jobs seconds=6
bmake: /home/lexi/src/freebsd/src/dev/share/mk/jobs.mk:40:
@ 1777264592 [2026-04-27 05:36:32] Start update-packages-jobs
@ 1777264593 [2026-04-27 05:36:33] Start update-packages -j10 log=/home/lexi/src/freebsd/src/update-packages.log
@ 1777264771 [2026-04-27 05:39:31] Finished update-packages-jobs seconds=179
%
some people (i learned a while ago) depend on cross-building from Linux for their FreeBSD production deployments, so this is more useful than you might initially think.
but also, since we already support cross-building, and pkgbase is now the preferred way to distribute the system, we really should support it anyway.
the only remaining issue is you have to install pkg(8) on the host for this to work, but at some point we'll vendor pkg as a bootstrap tool and that won't be required.
remember when millennials were kids and the only way we could run computer programmes was to type in a programme listing from a magazine?
a modest proposal: all open source software should be distributed this way.
landing D56087 tomorrow. i don't care if someone finds a bug that causes your computer to explode when updating, this shit is going in the tree.
Replying to
... this is the one thing that makes me think i might one day leave and switch to NetBSD, but a) i'm not sure if they're really any better, and b) too many missing features for me right now. but i suppose i could fix the latter with work...
so, trying to understand #git...
let's say i have an internal ForgeJo instance at git.example.org. i pull src.git from git.freebsd.org and push it to git.example.org, and now i can pull it on all my dev VMs.
now i want to commit a change, but i don't want to give my dev VM access to my FreeBSD SSH key. so, i make the change on dev, test and commit it, then push it to git.example.org.
then, on my desktop, i pull from git.example.org and push to gitrepo.freebsd.org. but, in the mean time, someone has has pushed another commit to gitrepo.freebsd.org, so i have to rebase and push again.
now i can't push my rebased local branch back to git.freebsd.org without doing a force push, which will break all my clones.
what's the right way to do this?
Replying to
workaround:
iptables -A OUTPUT -j REJECT
not ideal, but at least installing packages works now...
day 2 of trying to build llvm
current problem: building llvm 22 with llvm 19's lld causes a SIGBUS, so i need to convince it to use GNU ld instead...
also, Microsoft have so completely fucked up how you renew 365 licenses with product keys that i have no idea if our email will even work tomorrow. good thing it's Sunday, i guess.
Replying to
@dexter the last time a government tried that, we got SFU, so they could continue to buy Windows and say it's POSIX-compliant...
Replying to
@js thanks. i do already do all my work on branches, i just don't push them to git.example.org, so i guess i simply need to shift the pull/rebase step to my desktop instead of doing it on dev.
Replying to
@js thanks! i believe i'm familiar with all of those basic concepts -- i do a lot of manual rebasing for various reasons
by feature branches, do you mean i'd push the dev changes to a new branch on git.example.org, then do the pull/rebase on my desktop before pushing back to main? (so, never push to main from dev.)
> Bluesky users are mastering the fine art of blaming everything on “vibe coding”
it’s easy (and satisfying) to blame this sort of thing on ignorant social media users who are incapable of applying any sort of critical thinking, but this feels like it’s falling into the same old trap of trying to change other people’s behaviour, which is just not practical on a large scale.
i’m going to file this as more evidence that humans are simply not capable of existing in a functional way in social groups of more than a hundred or so.
the Internet was a mistake. we simply aren’t adapted to make the effects of instant worldwide communication networks a positive rather than a negative.
- can we get pMTU discovery?
- we have pMTU discovery at home
pMTU discovery at home:
18:34:02.673455 IP6 fd00:0:0:1::d.40735 > 64:ff9b::141a:9cd8.443: Flags [P.], seq 1:1549, ack 1, win 259, options [nop,nop,TS val 3090622102 ecr 3839900097], length 1548
18:34:02.673490 IP6 fd00:0:0:1::12 > fd00:0:0:1::d: ICMP6, packet too big, mtu 1420, length 1240
18:34:02.675307 IP6 fd00:0:0:1::d.40735 > 64:ff9b::141a:9cd8.443: Flags [P.], seq 1:1549, ack 1, win 259, options [nop,nop,TS val 3090622104 ecr 3839900097], length 1548
Replying to
@litchralee_v6 yeah, i think it must be something like that. for now i'm going to leave it with my hacky OUTPUT REJECT rule and come back to this later :D thanks for the help!
Replying to
@adrian_blx thanks, but, i'd rather not do that since there's no need for namespaces here: wg only uses IPv4 and the network connectivity is only IPv6 (over wg). so i'd rather work out why wg-quick is breaking it.
i wonder why we build mandoc as a bootstrap tool. is this for catpages? (i don't think those exist anymore...)
#freebsd
Microsoft: "This account already exist with another Microsoft Service, Continue purchase Power Automate Premium _IUR for your organization"
uh, okay. i mean, i support non-standard grammar, but i wasn't expecting Microsoft to lead the way there.
what type of Grafana graph would you use to visualise CPU usage on a system with 176 CPU threads? i usually like heatmap, but i don't think that's going to work unlses the graph is three miles tall.
Replying to
@litchralee_v6
hyacinth:~# ip -4 route
23.155.224.0/24 dev eth0 proto kernel scope link src 23.155.224.95
176.126.243.164 via 23.155.224.1 dev eth0
hyacinth:~#
that /32 route is the Wireguard endpoint, no default route.
Replying to
@dch yeah, i have 3 amd64 VMs for 16, 15, and 14, and another six for ports (main and quarterly for each version), and now i need to set up a bunch more for ppc testing, so force pushes are quite annoying :-)
(and my desktop runs Linux, so i can't easily test the build there as cross-build appears to be broken right now - that's on my todo list...)
day 3 of trying to build llvm.
the build finished, but clang-22 is 5.1GB and it runs out of disk space (with ~25GB free) trying to install. so i guess doing a RelWithDebInfo build was a bad idea. time to delete the build directory and start over again...
none of those pictures are flans
my fix for building stable/15 on main broke building main on stable/14. ah, the natural cycle of diffs. it's a very normal give-and-take sort of thing.
Replying to
@tfb merge commits aren't allowed in FreeBSD src.git (except for vendor merges) so i have to rebase in case of conflicts.
Replying to
@litchralee_v6 okay so, bizarrely, the packets are going over the Wireguard interface.
hyacinth# ping bbc.co.uk
PING bbc.co.uk (151.101.192.81): 56 data bytes
---
hyacinth:/etc# tcpdump -ni rose dst host 151.101.192.81
tcpdump: verbose output suppressed, use -v[v]... for full protocol decode
listening on rose, link-type RAW (Raw IP), snapshot length 262144 bytes
13:35:52.189219 IP 23.155.224.95 > 151.101.192.81: ICMP echo request, id 8050, seq 57, length 64
13:35:53.189802 IP 23.155.224.95 > 151.101.192.81: ICMP echo request, id 8050, seq 58, length 64
13:35:54.189848 IP 23.155.224.95 > 151.101.192.81: ICMP echo request, id 8050, seq 59, length 64
but there's no route, so this shouldn't be happening. perhaps wg-quick is doing some "helpful magic" here?
(i usually use plain wg, but that seems to be a bit of a pain on Linux...)
What's new in 7.24beta1 (2026-May-26 13:47):
*) l3hw - added HW offloaded support for VLAN interfaces created directly on Ethernet for CRS8xx series switches;
ooh! does this mean we finally don't have to create fake vlans for routed ports?
buildbot-worker does not support IPv6 connections by hostname?? is it not the year of our lord 2026
on an IPv6 router, when configuring a routed /64 for clients (on an access port, vlan, whatever), do you prefer to:
(a) assign an IP address from the /64 to the router interface, or
(b) do not assign an IP address and instead add a /64 interface route.
i used to do (a), but i'm starting to prefer (b); fewer IP addresses on the router makes management easier, and there are no "special" IP addresses in the subnet.
one downside seems to be that many routers in this configuration won't receive packets destined for the subnet-routers anycast address, which may technically be an RFC violation, but i've never seen anything that relies on that.