<{^_^}>
swaywm/sway#4091 (by grahamc, 50 weeks ago, closed): `exec sway` is security-critical, but underdocumented
<gchristensen>
"We're not saying that we're fine with sway crashing." lmao
<gchristensen>
and it crashes sooo much lol
<andi->
Yeah :/ I mean I used to do exec startx back in those days
<gchristensen>
yeah
<andi->
It is not their fault (entirely). There is/was no proper way to launch it without another graphical thing.
<gchristensen>
that is not the point of my issue
<gchristensen>
or the reason it is stuck in my craw
<gchristensen>
tbh the reason that issue is stuck in my craw is every time it crashes I'm like "good thing you didn't want for this to happen!" and then it crashes very often
<andi->
Mhm, it hasn't crashed for me in probably months
<gchristensen>
this is also not the point of my issue
<gchristensen>
the frequency of it crashing is not the point of my issue
<arianvp>
aanderse: I see you added apache ACME support. I wasn't aware we supported that :P
<andi->
I know.. It will happen and they should update the docs..
<gchristensen>
nahhh its fine
<arianvp>
aanderse: does it do a similar dance with snakeoil certs?
<gchristensen>
no need to put in a last-ditch safety measure, we don't want it to crash, it is fine
<gchristensen>
we are good C programmers, it won't crash probably
<andi->
gchristensen: I tried to submit a patch to aerc to let it read the config from the nix store. He considers public readable files as footguns there .
<andi->
Reminds me.. I did type 'host foo' on the current OpenBSD snapshot and it segfaulted...
<{^_^}>
emersion/mako#247 (by grahamc, 3 weeks ago, merged): criteria: always init its list
<gchristensen>
s/that link to sway/a link to mako/
<gchristensen>
as you can see this is seriously stuck in my craw lol
<andi->
Part of the reason I play with OpenBSD is so I see (again) what good stuff we have over here and what I miss. Makes me see things more positive :)
<gchristensen>
:D
<andi->
Think positive! They are doing this amazing work and probably get a few sh*t tons of trash reports every week. They do not reflect properly anymore. (They being sway people)
<gchristensen>
yeah
<gchristensen>
I am usually good at being very positive about things
<andi->
It is my great weakness...
<gchristensen>
something about their culture makes it very difficult for me to be positive
<gchristensen>
the "back to the olden days" stuff drives me nuts. okay I need to stop thinking about this, and try to be positive :P
<gchristensen>
sorry for ... ^
<andi->
We have a great saying in Germany: Früher war alles besser.
<gchristensen>
lol
<aanderse>
arianvp: ahhh now i understand... you fixed nginx issue and were kind enough to port fix to httpd :)
<aanderse>
arianvp++
<{^_^}>
arianvp's karma got increased to 6
<arianvp>
:)
<gchristensen>
arianvp++
<{^_^}>
arianvp's karma got increased to 7
Deknos has quit [Ping timeout: 260 seconds]
Deknos has joined #nixos-systemd
<arianvp>
merging the tests now flokli
Deknos has quit [Quit: Deknos]
<flokli>
I rebased my systemd-mainline branch on top of latest master, hoping to pick this up after having lost some motivation
<flokli>
would appreciate some help in figuring out if some of the failing tests are true regressions, or just have been flaky before.
<flokli>
^ andi- arianvp hexa- NinjaTrappeur
<andi->
I'll try to go through that now as I have a few hours of silence... Can't be bothered with people right now ;)
<flokli>
so, I refactored the flannel test a bit, it's now python, and a bit more bulletproof
<flokli>
(will push that separately)
<flokli>
but indeed the packets don't appear on the other side of the vxlan
<flokli>
I'm tcpdumping a bit
<arianvp>
im still confused about the fact that we need to explicitly list unit files we want from upstream but things like udev rules and link files are read from the package
<arianvp>
(im still not sure if this statement is actualyl entirely true either)
<flokli>
arianvp: can you elaborate?
<arianvp>
in systemd modules we explicitly copy upstreamUnits to /etc/systemd to make systemd pick them up
<flokli>
no
<arianvp>
we do right?
<flokli>
everything that is added to systemd.packages is added
<arianvp>
yeh but the `systemd` package itself has their unit files in a folder named "example"
<arianvp>
and hence they're not added :P
<flokli>
the systemd package itself is a bit different
<flokli>
there's code in the module system which "copies" upstream-provided units
<arianvp>
that's literally what I was saying :P
<flokli>
then I don't understand the question
<arianvp>
"the package" referred to "the systemd package"
<arianvp>
im just confused some systemd config (that is shipped in the systemd package) is treated differently than others
lovesegfault has joined #nixos-systemd
<flokli>
arf.
<flokli>
so, the flannel test has already been broken before
<arianvp>
but is flannel broken or the test broken ;)
<flokli>
I'm suspecting either the nixos firewall, or the way qemu does networking
<arianvp>
the patchset looks nice floran, esp if we can drop the micro-optimisation patches
<flokli>
thanks :-)
<flokli>
so, even with the firewall disabled, the machines don't want to talk with each other. mhhh
<arianvp>
hmm
<flokli>
so, it works with the udp backend
<flokli>
I assume multicast doesn't work properly with the qemu software networking
<flokli>
not sure if it /did/ work once, and broke