<qyliss>
Status update written, but now I have to try to edit it down a bit because it's currently more than twice as long as the (already extremely long) last one
<MichaelRaskin>
Just don't overdo the editing effort, remember cost/benefit
<MichaelRaskin>
Like, how would people have acquired the background necessary to read it after the first paragraph if we were slow readers?
<MichaelRaskin>
This times URL-as-a-statement is more precise than the previous time.
<Shell>
qyliss: I like msg_socket2.
<Shell>
just read through it a bit. :)
<qyliss>
:)
<qyliss>
Was very fiddly to write
<qyliss>
But I'm pretty pleased with it.
<qyliss>
The thing in it I'm really happy with is that it uses borrows of file descriptors to make sure they aren't closed before the message is sent
<qyliss>
(This was added after the original version, which did not do that, crashed because file descriptors were closed before the message was sent :P)
<MichaelRaskin>
Hehe
<Shell>
qyliss: quite nice. :3
<pie_[bnc]>
qyliss: do you by any chance know how to start a second x session on nixos
<MichaelRaskin>
I wouldn't expect too much success from asking people who have apparently abandoned X11 in favour of Wayland by now…
<pie_[bnc]>
ah
<pie_[bnc]>
should I do that
<pie_[bnc]>
it still sounds complicated given how much people talk about configuring wayland
<pie_[bnc]>
then again they talk about any other window desktop whatever manager too
<MichaelRaskin>
Getting a kind of working Weston is not hard.
<MichaelRaskin>
I say it as someone who dislikes Wayland and did not invest much effort.
<pie_[bnc]>
i imagine the problem on wayland is also going to be getting anything to talk to eachother
<MichaelRaskin>
> This got the number of open PRs below 1800 for a while, but it’s now right back where it started.
<MichaelRaskin>
The pain it's always there to stay
<Shell>
pie_[bnc]: getting what to talk to what? dbus is still a thing
<pie_[bnc]>
Shell: say I want to start something in display :1 from a terminal in :0, my nonexistent wayland knowledge tells me that wont work
<MichaelRaskin>
pie_[bnc]: that is easy actually
<pie_[bnc]>
with X you can just set DISPLAY
<MichaelRaskin>
Same with Wayland
<pie_[bnc]>
ok
<pie_[bnc]>
so what I currently happened to want to do before utterly f***ing my session as im ranting to MichaelRaskin in another channel, was run the test suite for wxpython in another x11 session so i can keep using my computer
<MichaelRaskin>
Modulo what specific vars you need to set, of course
<pie_[bnc]>
some things dont work in xephyr
<MichaelRaskin>
This being #spectrum, boot a CrosVM and launch X there?
<pie_[bnc]>
*shrug* :P
<xantoz>
pie_[bnc]: just switch VTs and startx?
<pie_[bnc]>
xantoz: thats what fails horribly
<xantoz>
pie_[bnc]: works here
<pie_[bnc]>
on nixos?
<xantoz>
yeah
<pie_[bnc]>
dude whats your magic sauce
<pie_[bnc]>
give me
<xantoz>
migth be some wierdness beacuse I use home manager. but it works
<xantoz>
pie_[bnc]: do you use a login manager?
<xantoz>
I disabled that, and just do startx for my zeroth sessino
<xantoz>
I have both .xinitrc and .xsession because of a little hack I had to do to get home manager to work ok when just using startx to start your session
<xantoz>
not sure .xprofile gets used how I start things, actually
<xantoz>
basically I told home manager that ".xsession" is my window manager
<pie_[bnc]>
i see
<xantoz>
is the problem with using startx to start your other X session that it doesn't load the right things?
<xantoz>
because startx will do things differently than your login manager, so might just be that
<xantoz>
.xinitrc is what gets executed by startx
<xantoz>
by default
<xantoz>
but home manager wants to write it
<pie_[bnc]>
not sure yet i barely know how to startx let alone do any of this stuff myself (yet?)
<xantoz>
I've just always used startx, so I don't really know what files login managers typically load
<xantoz>
I assume .xprofile?
<pie_[bnc]>
i need to go dig through the nixos source
<xantoz>
ah. .xprofile actually does get loaded in the .xinitrc that home manager made
<pie_[bnc]>
xantoz: i forgot to clarify, i have startx *running* after enable the exportConfig thing, the behaviours are just screwed up
<pie_[bnc]>
at least compared to what startx would get you a decade ago
<pie_[bnc]>
given how nix works i imagine i can learn all the stuff i need by looking at the serviceconfiguration
<pie_[bnc]>
modulo crosstalk related stuff
<xantoz>
pie_[bnc]: the X11 server starts, but the session doesn't?
<xantoz>
missing .xinitrc, probably/
<xantoz>
a decade ago I would've expected X11 to blow up if more than one instance ran at the same time...
<pie_[bnc]>
i dnt know ubuntu managed somehow
<xantoz>
maybe I'm thinking of even older
<xantoz>
might also depend on what drivers
<pie_[bnc]>
i was 16 or whatever and i was like whoa this is cool didnt know you could do this
<pie_[bnc]>
and then i would run full screen wine apps on a second desktop so it didnt screw up my resolution
<xantoz>
I had a xinerama setup with two graphics cards. I had to play around a lot to find two graphics cards whose drivers wouldn't rip each others' throats out
<pie_[bnc]>
there were still dome problems so i didnt always use it, i dont remember
<pie_[bnc]>
yeah i never had money for graphics cards lol
<pie_[bnc]>
or rather, my parents didnt buy me any
<pie_[bnc]>
my.
<pie_[bnc]>
god.
<pie_[bnc]>
oh.
<pie_[bnc]>
> this sysrq operation is disabled
<pie_[bnc]>
i thought people meant sysrqs all on or all off, i didnt know they could be turned off individually ;_; , i thought alt-sysrq-h would be sufficient to test.
<pie_[bnc]>
turns out ive been spouting bullshit all night
<pie_[bnc]>
ffffuuuu
<pie_[bnc]>
the 19.09 release notes have this hidden gem in them
<pie_[bnc]>
This also configures the kernel to pass core dumps to systemd-coredump, and restricts the SysRq key combinations to the sync command only. These sysctl snippets can be found in /etc/sysctl.d/50-*.conf, and overridden via boot.kernel.sysctl (which will place the parameters in /etc/sysctl.d/60-nixos.conf). "
<pie_[bnc]>
" We now install the sysctl snippets shipped with systemd.
andi- has quit [Remote host closed the connection]
andi- has joined #spectrum
<hyperfekt>
yay! a status update :D
<hyperfekt>
qyliss: s/reult/result/
thefloweringash has joined #spectrum
<hyperfekt>
good job pushing through on the status update!
<hyperfekt>
i'm looking forward to the streams
leah2 has quit [Remote host closed the connection]
leah2 has joined #spectrum
grufwub has joined #spectrum
<hyperfekt>
"the device a VM" also needs fixing
cole-h has joined #spectrum
multi has quit [Remote host closed the connection]
multi has joined #spectrum
<MichaelRaskin>
qyliss: re: external contributions not blocked by virtio work — maybe some experiments with small kernel builds could be done?
<MichaelRaskin>
Fast boot will eventually be needed…
edef has quit [Quit: leaving]
edef__ has joined #spectrum
edef__ is now known as edef
<qyliss>
Oh yeah that's a good idea
<qyliss>
If anybody wants to have a go at that that would be fantastic
<MichaelRaskin>
Maybe write it somewhere more permanent
<MichaelRaskin>
Re: ML — if you wanted ML to be more active, you could have explained which parts of the layout I got wrong…
<qyliss>
Ah, yes
<qyliss>
I think I started replying, but then realised there were things I didn't know, and then never finished.
<MichaelRaskin>
Or posted something where you want to see what ideas arise form a slow discussion
<MichaelRaskin>
Of course, not to want an active ML right now though might be a reasonable choice
<MichaelRaskin>
> there were things I didn't know — oh well, I tried to build a full plan that would be (badly) implementable without unpredictable foundational investments (like fixing virtio), what I expected is to indeed sketch a few alternative roadmaps to have terms for big-picture discussions of local decisions
<MichaelRaskin>
But yeah, if you decided to commit to doing virtio right, this was useless back then