<samueldr>
so if you know the moment you put it in service, you could approximate the time it spent flyings vs. not
<cole-h>
You forgot the 12.385s
<samueldr>
still over an elephant, still not a mars orbital period
disasm has quit [Ping timeout: 260 seconds]
evanjs has quit [Read error: Connection reset by peer]
evanjs has joined #nixos-chat
<colemickens>
I decided to tell myself that the LCM people are still around the area and that the equipment is just in storage. Trying to find some silver.
<colemickens>
If I were a parent, I'd offer to buy all my kids' old electronics and just store them.
drakonis has joined #nixos-chat
<cole-h>
Holy shit
<cole-h>
Power just flickered and one of my monitors went out and then came back
<cole-h>
Scary shit
<cole-h>
(Excuse the language, I forgot where I was for a moment)
<pie_>
if any scary looking girls come out of your monitors, try inviting them for tea
<cole-h>
I hope cider is acceptable.
<cole-h>
It happened again 😨
<cole-h>
I don't know how well ZFS handles ungraceful shutdowns due to power...
<cole-h>
... outages...
<pie_>
i would like to assume pretty well
<pie_>
run a scrub?
<pie_>
also maybe take a break from the computer for a bit
<cole-h>
It hasn't gone out yet
<cole-h>
Yeah, that's the plan
<samueldr>
break? from the computer?
<samueldr>
I only break for loops
<cole-h>
I also don't have any redundancy, so if my pool dies... :'(
<pie_>
samueldr: well, unless your laptop has a battery :P
* cole-h
quickly commits everything that would leave him devastated if lost
<samueldr>
cole-h: and push
<cole-h>
Done, thanks.
<cole-h>
And now to take that fabled "break from the computer."
cole-h has quit [Quit: Goodbye]
puzzlewolf has quit [*.net *.split]
aterius has quit [*.net *.split]
ohhaimark[m] has quit [*.net *.split]
thefloweringash has quit [*.net *.split]
johanot has quit [*.net *.split]
elvishjerricco has quit [*.net *.split]
ldlework has quit [*.net *.split]
rummik has quit [*.net *.split]
kalbasit has quit [*.net *.split]
rardiol has quit [*.net *.split]
bridge[evilred] has quit [*.net *.split]
abathur has quit [*.net *.split]
AluisioASG has quit [*.net *.split]
genevino has quit [*.net *.split]
numkem has quit [*.net *.split]
tilpner has quit [*.net *.split]
energizer has quit [*.net *.split]
nckx has quit [*.net *.split]
c74d has quit [*.net *.split]
eyJhb has quit [*.net *.split]
evanjs has quit [*.net *.split]
srhb has quit [*.net *.split]
lovesegfault has quit [*.net *.split]
lukegb has quit [*.net *.split]
infinisil has quit [*.net *.split]
makefu has quit [*.net *.split]
jared-w has quit [*.net *.split]
drakonis has quit [*.net *.split]
raboof has quit [*.net *.split]
zimbatm has quit [*.net *.split]
jw358 has quit [*.net *.split]
tazjin has quit [*.net *.split]
Cynthia has quit [*.net *.split]
rizary has quit [*.net *.split]
LnL has quit [*.net *.split]
lucus16 has quit [*.net *.split]
Ashy has quit [*.net *.split]
spacekookie has quit [*.net *.split]
ky0ko has quit [*.net *.split]
hexa- has quit [*.net *.split]
f0x has quit [*.net *.split]
gchristensen has quit [*.net *.split]
samueldr has quit [*.net *.split]
viric has quit [*.net *.split]
danielrf[m] has quit [*.net *.split]
Peetz0r has quit [*.net *.split]
edef has quit [*.net *.split]
liff has quit [*.net *.split]
andi- has quit [*.net *.split]
misuzu has quit [*.net *.split]
Jackneill has quit [*.net *.split]
das_j has quit [*.net *.split]
kini has quit [*.net *.split]
ottidmes_ has quit [*.net *.split]
ekleog has quit [*.net *.split]
hl has quit [*.net *.split]
dadada_ has quit [*.net *.split]
evils has quit [*.net *.split]
adisbladis has quit [*.net *.split]
monsieurp has quit [*.net *.split]
avn has quit [*.net *.split]
clever has quit [*.net *.split]
danderson has quit [*.net *.split]
kraem has quit [*.net *.split]
kcalvinalvin has quit [*.net *.split]
Arahael has quit [*.net *.split]
ornxka_ has quit [*.net *.split]
pie_ has quit [*.net *.split]
colemickens has quit [*.net *.split]
Gaelan_ has quit [*.net *.split]
bqv has quit [*.net *.split]
vesper11 has quit [*.net *.split]
fpletz has quit [*.net *.split]
flokli has quit [*.net *.split]
etu has quit [*.net *.split]
savanni_ has quit [*.net *.split]
leonardp has quit [*.net *.split]
steveeJ has quit [*.net *.split]
mudri has quit [*.net *.split]
lopsided98 has quit [*.net *.split]
Taneb has quit [*.net *.split]
{^_^} has quit [*.net *.split]
Shados has quit [*.net *.split]
hax404 has quit [*.net *.split]
aranea has quit [*.net *.split]
ashkitten has quit [*.net *.split]
Valodim has quit [*.net *.split]
joepie91 has quit [*.net *.split]
wirew0rm has quit [*.net *.split]
immae has quit [*.net *.split]
jackdk has quit [*.net *.split]
julm has quit [*.net *.split]
kgz has quit [*.net *.split]
aszlig has quit [*.net *.split]
arianvp has quit [*.net *.split]
lejonet has quit [*.net *.split]
arianvp has joined #nixos-chat
lejonet has joined #nixos-chat
aszlig has joined #nixos-chat
kgz has joined #nixos-chat
julm has joined #nixos-chat
jackdk has joined #nixos-chat
genevino has joined #nixos-chat
das_j has joined #nixos-chat
Jackneill has joined #nixos-chat
kini has joined #nixos-chat
ottidmes_ has joined #nixos-chat
ekleog has joined #nixos-chat
hl has joined #nixos-chat
edef has joined #nixos-chat
liff has joined #nixos-chat
andi- has joined #nixos-chat
misuzu has joined #nixos-chat
danielrf[m] has joined #nixos-chat
feepo has quit [Ping timeout: 270 seconds]
Peetz0r has joined #nixos-chat
JJJollyjim has quit [Ping timeout: 244 seconds]
vaibhavsagar has quit [Ping timeout: 256 seconds]
jneplokh has quit [Ping timeout: 240 seconds]
Irenes[m] has quit [Ping timeout: 240 seconds]
philipp[m] has quit [Ping timeout: 244 seconds]
jtojnar has quit [Ping timeout: 244 seconds]
emily has quit [Ping timeout: 244 seconds]
balsoft has quit [Ping timeout: 240 seconds]
worldofpeace has quit [Ping timeout: 240 seconds]
tokudan[m] has quit [Ping timeout: 240 seconds]
aanderse has quit [Ping timeout: 240 seconds]
arcnmx has quit [Ping timeout: 246 seconds]
leons has quit [Ping timeout: 256 seconds]
viric has joined #nixos-chat
CRTified[m] has quit [Ping timeout: 244 seconds]
rycee has quit [Ping timeout: 244 seconds]
flokli has joined #nixos-chat
etu has joined #nixos-chat
fpletz has joined #nixos-chat
Gaelan_ has joined #nixos-chat
vesper11 has joined #nixos-chat
bqv has joined #nixos-chat
mudri has joined #nixos-chat
leonardp has joined #nixos-chat
ashkitten has joined #nixos-chat
Valodim has joined #nixos-chat
lopsided98 has joined #nixos-chat
hax404 has joined #nixos-chat
immae has joined #nixos-chat
Shados has joined #nixos-chat
savanni_ has joined #nixos-chat
joepie91 has joined #nixos-chat
wirew0rm has joined #nixos-chat
aranea has joined #nixos-chat
Taneb has joined #nixos-chat
{^_^} has joined #nixos-chat
betawaffle has quit [Ping timeout: 244 seconds]
leonardp has quit [Ping timeout: 260 seconds]
steveeJ has joined #nixos-chat
ornxka_ has joined #nixos-chat
pie_ has joined #nixos-chat
Arahael has joined #nixos-chat
thefloweringash has joined #nixos-chat
puzzlewolf has joined #nixos-chat
ohhaimark[m] has joined #nixos-chat
johanot has joined #nixos-chat
ldlework has joined #nixos-chat
rummik has joined #nixos-chat
steveeJ has joined #nixos-chat
steveeJ has quit [Changing host]
energizer has joined #nixos-chat
AluisioASG has joined #nixos-chat
numkem has joined #nixos-chat
c74d has joined #nixos-chat
tilpner has joined #nixos-chat
nckx has joined #nixos-chat
eyJhb has joined #nixos-chat
drakonis has joined #nixos-chat
jared-w has joined #nixos-chat
Cynthia has joined #nixos-chat
ky0ko has joined #nixos-chat
Ashy has joined #nixos-chat
gchristensen has joined #nixos-chat
LnL has joined #nixos-chat
jw358 has joined #nixos-chat
raboof has joined #nixos-chat
hexa- has joined #nixos-chat
zimbatm has joined #nixos-chat
lucus16 has joined #nixos-chat
samueldr has joined #nixos-chat
tazjin has joined #nixos-chat
spacekookie has joined #nixos-chat
f0x has joined #nixos-chat
rizary has joined #nixos-chat
feepo has joined #nixos-chat
lukegb has joined #nixos-chat
infinisil has joined #nixos-chat
lovesegfault has joined #nixos-chat
makefu has joined #nixos-chat
evanjs has joined #nixos-chat
srhb has joined #nixos-chat
kcalvinalvin has joined #nixos-chat
kraem has joined #nixos-chat
danderson has joined #nixos-chat
clever has joined #nixos-chat
feepo has joined #nixos-chat
feepo has quit [Changing host]
adisbladis has joined #nixos-chat
evils has joined #nixos-chat
dadada_ has joined #nixos-chat
avn has joined #nixos-chat
monsieurp has joined #nixos-chat
bridge[evilred] has joined #nixos-chat
abathur has joined #nixos-chat
rardiol has joined #nixos-chat
kalbasit has joined #nixos-chat
puzzlewolf has quit [Ping timeout: 246 seconds]
elvishjerricco has joined #nixos-chat
thefloweringash has quit [Ping timeout: 246 seconds]
ohhaimark[m] has quit [Ping timeout: 246 seconds]
jared-w has quit [Ping timeout: 245 seconds]
raboof has quit [Ping timeout: 245 seconds]
rizary has quit [Ping timeout: 245 seconds]
betawaffle has joined #nixos-chat
danielrf[m] has quit [Remote host closed the connection]
pasukon has quit [Ping timeout: 244 seconds]
jared-w has joined #nixos-chat
pasukon has joined #nixos-chat
raboof has joined #nixos-chat
rizary has joined #nixos-chat
colemickens has joined #nixos-chat
Irenes[m] has joined #nixos-chat
thefloweringash has joined #nixos-chat
jtojnar has joined #nixos-chat
puzzlewolf has joined #nixos-chat
danielrf[m] has joined #nixos-chat
JJJollyjim has joined #nixos-chat
jneplokh has joined #nixos-chat
rycee has joined #nixos-chat
worldofpeace has joined #nixos-chat
emily has joined #nixos-chat
arcnmx has joined #nixos-chat
ohhaimark[m] has joined #nixos-chat
aterius has joined #nixos-chat
aanderse has joined #nixos-chat
tokudan[m] has joined #nixos-chat
leons has joined #nixos-chat
balsoft has joined #nixos-chat
CRTified[m] has joined #nixos-chat
philipp[m] has joined #nixos-chat
leonardp has joined #nixos-chat
vaibhavsagar has joined #nixos-chat
<abathur>
anyone use/d hugo manage to succesfully bridge a big version update gap? I'm finding the version notes and velocity daunting...
* abathur
hears that dumb little voice that whispers "you could rewrite this in bash faster than you can figure out what APIs changed over 30 releases...
<samueldr>
and you know it's right
<abathur>
also can't imagine anyone who came up with semver imagining a project on version 0.73.3 or whatever with 45k stars and 5k forks or whatever effectively still shirking any responsibility to be stable and keep a clear list of breaking changes (assuming I'm not just too dumb to find it...)
<samueldr>
fear of commitment is a thing
<abathur>
probably just need to start a new site from scratch and then bring in my template and then my content
<abathur>
It may be less annoying to build now, IIRC I've seen things I had to hack around before becoming straightfoward features
<abathur>
0.73.0, sorry for exaggerating :P
* abathur
spends more time reading comparisons of static site generators than it would take to rebuild the site in modern hugo *and* rewrite it in bash
<jackdk>
This is one reason I disagree with semver's rule for <1.0 versions
<abathur>
if you wanna do that just use year.month and spare us the suspense :D
<samueldr>
but then you'll be lambasted because you're not "following semver"
<abathur>
bog there's a market for a java/kotlin static site generator?
<abathur>
ah, doc focused, I guess I'll allow it
<abathur>
also to be fair it only just barely has more java than CSS
<abathur>
> Gollum - Gollum is a simple wiki system built on top of Git. - #Ruby
<{^_^}>
error: syntax error, unexpected '-', expecting ID or OR_KW or DOLLAR_CURLY or '"', at (string):318:62
<abathur>
oops; anyways--none of those things sounds independently simple
<jackdk>
samueldr: too bad, I follow PVP. =)
<jackdk>
,cookie
<samueldr>
jackdk: PVP?
<samueldr>
(the acronym might be hard to search for)
<abathur>
I have some little fantasy about making a weird little test-driven version protocol, but I haven't had time to fiddle tooling enough to really seriously try to adopt it in one of my own projects
<jackdk>
haskell's package version policy. Predates Semver, similar in some ways, pretty much confined to the Haskell world, which is where I do most of my stuff
<abathur>
I guess I'll have to write a pure-bash static site generator made from nothing but bash aliases, builtins, and fairy-dust
<abathur>
hmm, going to be a pita to generate good json from pure bash :(
<jackdk>
abathur: have you seen nanoblogger?
<abathur>
this seems like a common windmill to tilt at recently, github has *3* bash static site generator repos in 2 different *human* languages that have been created in the last 2 weeks, and 2 of them even have code
<abathur>
it lists one of its dependencies as: Self-loathing. Why else would you be considering using a static site generator written in Bash.
rajivr has joined #nixos-chat
<abathur>
I like that nanoblogger lists this: cons: no comments (only available through unoffical add-ons)
drakonis has quit [Quit: WeeChat 2.8]
<ashkitten>
hmm samueldr do you think it'd be reasonable to use robotnix to try and port lineage to a new device?
<samueldr>
possibly
<ashkitten>
(i don't remember who makes robotnix, if they're around)
<samueldr>
danielrf[m] does
<ashkitten>
oh cool
<samueldr>
though the hard part is actually the port I guess
<ashkitten>
right
<ashkitten>
it's got a slightly outdated unofficial omnirom build i guess
<ashkitten>
that's the only custom rom available
<samueldr>
oh, if there is already something to base it off of, it's already something more, but even then, when I tried last time I never was able to make it work
<ashkitten>
(it's the moto one hyper)
<samueldr>
I suspect it was related to the vendor firmwares
<ashkitten>
hmm
<samueldr>
let's just say that I was basically as confused about how to properly port to a device when I stopped than when I started
<ashkitten>
yeah
<samueldr>
so I can't be of much help :)
<samueldr>
though once you know it builds with whatever funny stuff linageOS uses, it should be buildable using robotnix
<ashkitten>
i'd love to just port mobile-nixos but it'd be even more useless than all the other devices without hybris/halium
<ashkitten>
this is all the info ky0ko has compiled about the device, i guess
<samueldr>
>> Downstream stock rom's kernel uses an out-of-tree module called focaltech_mmi to support the touch panel
<samueldr>
ugh
<samueldr>
such GPL non-compliance
<samueldr>
(I assume the module is a binary)
<ashkitten>
that's why i said it'll be even less useful without hybris
<samueldr>
AFAIK hybris won't help with kernel modules
<ashkitten>
oh
<samueldr>
it's about the userspace programs
<samueldr>
daemons that run and then talk over proprietary (or not) interfaces to the kernel
<ashkitten>
hm
<ashkitten>
so, not promising that she couldn't get the kernel module loaded
<samueldr>
I don't know about kernel modules and what makes them compatible or not with a build
<ashkitten>
maybe i'll just run the unofficial omnirom build and not bother
<ashkitten>
until mobile-nixos has hybris and stuff so there's incentive to get it working
<samueldr>
I wonder if for -def it's as simple as transplanting another focaltech_mmi driver from another kernel where it's shipped with
<samueldr>
I guess it must not be as simple as that
<samueldr>
it would be too easy
<ashkitten>
hmm
<ashkitten>
dunno if ky0ko has tried
bqv has quit [Ping timeout: 260 seconds]
bqv has joined #nixos-chat
<ky0ko>
hybris does do kernel modules if the kernel built is compatible
<ky0ko>
many of the userspace daemons wouldn't even work without the modules
<samueldr>
I still need to brush up on hybris, but what does it do for modules?
<samueldr>
I thought the module would be loaded by the kernel
<etu>
eyJhb: I've had troubles with slow prompt rendering
<etu>
eyJhb: But that was because I had a really stupid prompt function that re-did everything all the time
<ky0ko>
the kernel doesn't load them on its own. usually the modules are modprobed by the userspace daemons.
<ky0ko>
so when you start up the service that does e.g. the wifi setup, that service goes and modprobes the icnss driver
<ky0ko>
which is sitting alongside it on the vendor partition
<ky0ko>
if the kernel was built with the same sources, same version of compiler, and not too many options were changed, generally the module will still be loadable
<ky0ko>
for -def, this will mean building with clang (8.0.1 iirc)
<samueldr>
right, though for my understanding, the modprobe here does not depend on hybris, right?
<samueldr>
assuming the kernel is built in a compatible way (compiler, etc), if I go and load the module manually using insmod or modprobe, without hybris in sight, it would load
<ky0ko>
yes
<ky0ko>
so if you had the kernel module for e.g. focaltech_mmi added in separately then yeah, you could get touchscreen without hybris
<ky0ko>
not so sure about the wifi
<samueldr>
good, I was worried I was building my knowledge on top of wrong assumption
<samueldr>
(I think it's possible matrix is laggy tonight)
<samueldr>
(which I believe you're using)
<samueldr>
AFAIUI wi-fi could work without hybris, but the right startup sequence is assumed to be handled by the daemon, which is basically where I'm at once I'm done with the yak shave that'll help me look into it
<samueldr>
(and done with not-working-on-mobile-nixos)
<ky0ko>
simply because the userspace daemon is needed to talk to the module
<eyJhb>
samueldr: what have you done regarding work, before starting nixos-mobile? Self employed or what?
<samueldr>
self-employed
<samueldr>
but I wasn't a linux or low-level dev
<samueldr>
uh, not self-employed duh
<samueldr>
doing contract work for a tiny firm
<samueldr>
web dev
<samueldr>
eyJhb: is that what you wanted to know?
<eyJhb>
samueldr: so you moved from webdev to low-level linux stuff? - Just if you had a job, said screw this, and then went from a office setting to working from home :)
<samueldr>
oh, it wasn't office setting :)
kalbasit has quit [Ping timeout: 264 seconds]
<samueldr>
and basically I was spending my time off learning and doing that lower-level stuff, or learning and using linux, well before nixos, and then nixos basically made this easier
<eyJhb>
Cool!
<eyJhb>
But then I guess you do not miss the office setting, etc.
<samueldr>
since I have A/C in my office room at home, no I don't
<eyJhb>
*people? Noo.. A/C? Yes!* :D
<eyJhb>
And cats as far as I remember samueldr ?
<samueldr>
it gets quite toasty here during summer
<samueldr>
cat (1)
<samueldr>
pun not intended
<eyJhb>
Where were you located again? - Ah, my memory is broke. So just excited I could remember something :D
<samueldr>
no snow accumulation during winter? that's not fair
<eyJhb>
Well.. This year we basically had no snow...
<eyJhb>
But yeah, not really any exsessive amounts of snow here
<eyJhb>
off-topic from weather: I hate people who will CC your boss, to half yell at you without knowing what they are talking about
<eyJhb>
F those people, especially when they know shit
<eyJhb>
`Exceptional weather conditions can bring as much as 50 cm of snow to Copenhagen in a 24-hour period during the winter months` samueldr , guessing you have a lot more
<samueldr>
in one day, I wouldn't think so
<samueldr>
there's a finite amount of water in the clouds!
<samueldr>
>> The event included the worst 24-hour snowfall on record in the city of Montreal with 43 centimeters
<samueldr>
but the thing is, if we get 30cm, it doesn't thaw until spring :)
<samueldr>
>> Quebec City also had their worst season in 1970–1971 with 460 centimeters (180 inches) before it was broken in 2007–2008 when just over 500 centimeters (200 inches) fell
<samueldr>
imagine, 2007, 5 meters of snow during all of winter
<ar>
nice
<ar>
how did people who live on the 1st floor deal with it?
<ar>
i imagine they had things like blocked doors
<energizer>
it's like a foot a week, not too hard to deal with
<samueldr>
5 meters over a span of months
<samueldr>
but sometimes, with wind, it happens that people's door get totally covered with a normal snow fall
<samueldr>
"normal"
rtjure has joined #nixos-chat
<bqv>
My efforts to declaratively break out of the sandbox have been fruitless
<arianvp>
manveru: looked through bitte ! Looks nice!
<arianvp>
Though not for the fair heart currently
<arianvp>
Hehe
<eyJhb>
samueldr: I both wish and do not wish for that amount of snow
<eyJhb>
I am just said I changed to winter tires, for nothing...
<eyJhb>
And then took me waaay to long to get the summer ones on
<manveru>
arianvp: yeah, there's still do much to do...
<manveru>
Also got bitte-cli for the deeply
<manveru>
Damn,I can't type on mobile
rtjure has quit [Ping timeout: 246 seconds]
<manveru>
arianvp: but I'd be very happy about any kind of feedback or recommendation, I'm still pretty new to that whole ecosystem
<arianvp>
One problem I have with having all module options deeply embedded into NixOS is that suddenly you have a burden of keeping them up to date.
<arianvp>
But that's less of s problem with stable software
<eyJhb>
Anyone knows of alternatives to Amazon Polly? Seems nice, but it feels off
<arianvp>
Google has a competing project? Not sure if that's less off
<arianvp>
Windows has an amazing speech Synthesizer built-in offline
<bqv>
__noChroot to the rescue
<eyJhb>
arianvp: well, sadly I can't script up against that
<eyJhb>
More like, I can't do it in a easy way from Linux, when I do not really own a Windows PC :p
<manveru>
arianvp: yeah... i've been thinking of automating the option creating a bit more, i basically used a little script to convert their docs pages into nixos options
<arianvp>
At least Terraform schemas can be introspected. I wrote a bit of code for that before that is very rudimentary
<arianvp>
Using the grpc plugin interface
<manveru>
terraform has been alright, terranix doesn't fully specify that other than the top-level attributes like resource/data/output/...
<manveru>
but i ended up making a huge specialized module for it anyway because writing terraform is rather painful...
<manveru>
and making the options 1:1 encoded in nix doesn't seem like a huge benefit
<manveru>
i don't really expect the vault/consul/nomad module to be included like this in nixos either unfortunately
<manveru>
they're still too opinionated :)
<bqv>
TIL terranix
<manveru>
and the whole bootstrap circus is so hard to understand, i wish i could visualize systemd services
<manveru>
ended up writing a module for specifying haproxy configs yesterday, and i fear i'll have to overhaul all the cert stuff for the fifth time :(
<manveru>
it's a huge PITA to get those certs all into the right places at the right time with the righ permissions
<srk>
/o\
<manveru>
so what i really want is to get an ACME cert into vault and let vault generate certs for everybody
<manveru>
but to start vault, you need certs...
<manveru>
and for vault storage, you need consul, which needs certs...
<philipp[m]>
Can't you bootstrap with self signed certs?
<manveru>
that's basically what i do
<manveru>
i create certs on the deployer and copy them over before doing the initial nixos-rebuild
<manveru>
then i have a list of services that take the certs from /run/keys and copy them into the corresponding systemd services runtime directores
<manveru>
*directories
<manveru>
but that's all very hardcoded and brittle
<manveru>
i hate when order matters, there's no way to roll back
<philipp[m]>
Yeah, that's very annoying.
<manveru>
and in the end, i have only self-signed certs everywhere :P
<manveru>
then i'll have to coordinate all services to switch to the ACME cert without actually losing connectivity to each other, or vault stops working
<manveru>
i actually had three phases in my earlier implementation, where i'd first start all services without TLS enabled and then run nixos-rebuild three times to get to the end state
<manveru>
but that was way too brittle
parsley936 has joined #nixos-chat
<philipp[m]>
Yeah, so it sounds like what you actually would need is another layer of trusted connections? I'm not sure about your architecture, but it kinda sounds like you need a vpn or something like it.
<manveru>
i have no clue how others do this, the amount of production ready open source cluster setups out there is pretty close to zero
<manveru>
yeah, been thinking about adding glusterfs and wireguard to the mix :P
<bqv>
manveru: I just switched back to traefik in the end, lol
<bqv>
But I too had a haproxy module
<manveru>
i've got no exp with traefik
<manveru>
i'm usually a nginx guy :)
<manveru>
i'm trying envoy after all now, given that it's already used by consul connect and i have packaged it...
<manveru>
but damn that's some verbose config they got there...
<manveru>
thankfully it's yaml, which means it's json, which means i can generate it with nix :)
<eyJhb>
JSON is YAML, but YAML is not JSON as far as I remember :p
<eyJhb>
Isn't it like that?
<manveru>
yep
<adisbladis>
eyJhb: Yep
<adisbladis>
And YAML isinsanity so don't use that
<eyJhb>
I WOULD NEVER! EVER USE YAML! (uses it for all configs atm. in this project)
<adisbladis>
=(
<eyJhb>
Any alternative for something easily human editable?
<adisbladis>
I'd much rather write json than yaml tbh
<adisbladis>
eyJhb: toml
<eyJhb>
Ohh, I hate that ..
<eyJhb>
And like it at the same time
<manveru>
nix generated json ftw :)
<adisbladis>
This ^
<adisbladis>
The _only_ good thing about yaml parsers is that I can generate a file with nix
<eyJhb>
I have some fun implementation facts for you in Go regarding JSON vs YAML adisbladis
<adisbladis>
Mhm?
<eyJhb>
If you have a byte slice in Golang, then when you marshal it to JSON it will encode it as a base64 string. For YAML it will however make a array of all the decimal values of each entry
<eyJhb>
I know that is "what it is", but I so much prefer base64 encoded... It does not fill a shit load that way
<manveru>
well, you could always wrap it into a custom type for encoding
<manveru>
but yeah, i've had my share of struggles with Go and JSON
* manveru
still shivers when thinking of JSONAPI
<adisbladis>
Go+JSON is really not ergonomic
<manveru>
we also did the documentation of the API using YAML/Swagger...
<adisbladis>
Yay...
<manveru>
which was even more of a pain, we resorted to writing it by hand :P
<adisbladis>
manveru: Fwiw I've had pretty good success with grpc + swagger
<manveru>
didn't have much choice in that case, was a contract over a few millions :)
<adisbladis>
I even managed to get grpc & the rest api on the same port
<manveru>
so they hashed out all the details beforehand and chose "common standards"
<manveru>
i'm just glad it wasn't XML
<adisbladis>
=)
<bqv>
Write directly in bson like a true pro
<Valodim>
also switched from go with json to grpc
<Valodim>
acceptable experience so far
<srhb>
srk: Thanks!
<philipp[m]>
adisbladis: I did some digging around eshell again. It seems like there are plugins that give you fish completion in eshell via pcomplete. https://gitlab.com/ambrevar/emacs-fish-completion
<philipp[m]>
Suddenly I'm interested in eshell again.
<eyJhb>
manveru: JSON and Go works just fine, what problems have you had?
<philipp[m]>
adisbladis: Played around with eshell now. I think this is all I need for missing features on top of fish completion modes: https://github.com/manateelazycat/aweshell
<__monty__>
infinisil: Filtering on the server rather than the client seems a bit weird?
<infinisil>
__monty__: Since I control the server that's fine for me
<infinisil>
It also means I don't have to configure the client(s)
<infinisil>
So I have the same thing happening e.g. on my phone
neeasade has joined #nixos-chat
FRidh has joined #nixos-chat
waleee-cl has joined #nixos-chat
kalbasit has joined #nixos-chat
drakonis has joined #nixos-chat
<polezaivsani>
want to fix a bug in docs - what might be a plausible KDE specific package to use NixOS `environment.systemPackages` option?
<colemickens>
god I love that important conversations happen on twitter and then just disappear
<colemickens>
moxie wrote up this entire explanation for signal and pins and its just some damn twitter thread
<cole-h>
Twitter is good for posting links and not much else IMO
<colemickens>
I guess I can't complain since I have a pile of blog topics that I never write about
<colemickens>
I think people can still read paragraphs without needing to see reply/rt/heart buttons every 140 chars
<colemickens>
But how will people know which sentence I agree with
* colemickens
ok thats enough
<pie_>
colemickens: pls archive
<drakonis>
there's no guide on how to build a system with flakes yet, right?
<drakonis>
i mean, i want to use nix from the nix flake instead of nixpkgs
<drakonis>
the current nix flake has working completion instead of being half broken
<joepie91>
colemickens: "things that I never write about" is precisely the value of Twitter for blogging
<drakonis>
it is so good
<joepie91>
it's a very low barrier of entry, there's zero expectation of you writing more than one sentence
<joepie91>
so it's easy to get over that barrier of "I don't want to write a whole thing about this" and just keep writing until you're out of motivation
<joepie91>
until some sort of "proper" blog system manages to address that problem, people are gonna keep blogging stuff in Twitter threads
<ashkitten>
when are we gonna have a decentralized megaforum software tbh
<qyliss>
ah, usenet
<drakonis>
i want that again
<drakonis>
decentralized megaforum means i can bring the magic of ye old good days of internet forums
<ashkitten>
qyliss: correct me if i'm wrong but usenet was more federated than fully decentralized right?
<qyliss>
yeah
<drakonis>
with the something awful drama taking place
<drakonis>
there has been some pretty heavy fragmentation
<ashkitten>
i've really turned around wrt federated software, i think it's a bad idea that leads to massive social issues
<drakonis>
and i dont think they'd like to be under someone else's rule for the upteenth time
<philipp[m]>
If only gnunet wasn't stuck in limbo, it would be pretty easy to build something like this.
<ashkitten>
imo social software should be p2p or nothing
<philipp[m]>
Briar also has forum capabilities but it's also very alpha without much development going on (looking from the outside) https://briarproject.org/
<ashkitten>
hm
<ashkitten>
pgp, ew
<joepie91>
ashkitten: yeah, like I said, bit weird :D
<joepie91>
the whole thing is a bit... yeah
<joepie91>
it does all the things it claims, just everything is slightly awkward and slow and GPG
<joepie91>
and it's a bit obscure so I'm not sure how resistant its network design actually is
<ashkitten>
i've heard nice things about briar
<ixxie>
ashkitten: can you elaborate on why you think federated systems are so problematic?
<ashkitten>
i've been on the fediverse for almost 4 years now and there's always been fighting about oh this instance is bad, we should block it, now their users are alienated and have to find a new instance if they want to be seen, even just the requirement of having to find an instance to sign up on before you even get to see what the fediverse is like... it's all bad
<ashkitten>
because activitypub doesn't do e2ee or anything, we can't have nomadic identity, and our choice in server is both incredibly impactful and long-lasting
<ixxie>
so there is no real mobility
<ashkitten>
there's always just so much meta discourse that wouldn't happen with a p2p system
<ashkitten>
someone can spin up an instance and have more power than the average user just because they have the hardware, and now anyone who signs up on their instance is subject to their rule
<joepie91>
I think "no nomadic identity" is the crux there
<ashkitten>
partly
<ashkitten>
but how can you safely implement nomadic identity when instances themselves must be trusted separately?
<ashkitten>
matrix gets it right because they're e2ee by default
<ashkitten>
activitypub has no e2ee at all
<ashkitten>
so they've settled for this half-solution where you can "migrate" your account by setting your old account to redirect to the new one, and then clicking a button to force everyone who follows your old one to instead follow the new one
<ashkitten>
note that this doesn't let you follow them back in the same way
<ashkitten>
because forcing them to send their posts to an untrusted instance is dangerous
<ashkitten>
because there is absolutely no protection from any admin reading your posts if it reaches their instance
* pie_
mumbles something about email mixnets
* pie_
doesnt know anything about them
<ashkitten>
matrix is looking into p2p design and nomadic/multi-home identity and i think they'll get it right, because matrix has always been relatively good wrt privacy and user experience
* adisbladis
read nomadic as monadic and got confused
<ixxie>
ashkitten: I'm wondering if one could build a richer social media experience on top of matrix
<ashkitten>
possibly
<ashkitten>
i'd be into that
<ixxie>
is it designed for extensibility?
<ashkitten>
yes
<__monty__>
Pretty sure. It's "just" an eventually consistent DB, right?
<joepie91>
sort of
<joepie91>
it's designed as an eventually-consistent event store, more or less
<joepie91>
but certainly with a focus on chat/IM features
<ixxie>
I guess its hard to say before trying exactly what the painpoints would be
<joepie91>
ixxie: fwiw, there was some research recently on the robustness of Matrix' state resolution protocol, which is core to its distributed design
<joepie91>
ixxie: going from memory, in this context "without finality" basically means "without the certainty at any point in time that there doesn't exist a branch that you are unaware of"
<joepie91>
ie. the DAG can always retroactively have stuff attached
<joepie91>
no point in it is ever 100% certainly "done"
<joepie91>
(but the paper will probably define it more precisely)
<ixxie>
joepie91: I think it goes deeper than I am willing to go at this point; I'm trying to get a feel for which tech stack to bet on, and this tells me matrix is promising for a decentralized future
<joepie91>
ixxie: so my cliff's notes on Matrix as a decentralized tech stack: 1) it has a lot of potential and seemingly a baseline sensible design for chat usecases, 2) it still needs a lot of polishing (of eg. the spec) to really be robust, but it's probably viable, 3) the reference implementations are currently junk from a performance perspective, but this does not seem to be inherent to the protocol design, and 4) I am less
<joepie91>
certain about its suitability for general-purpose distributed data persistence; I would say that it can *probably* do it well, but I wouldn't bet a company on it without first understanding the tradeoffs very well
<joepie91>
oh, and 5) I am a bit more skeptical about the pure-P2P aspect of it, and how well it will actually work in practice. but work is ongoing on that
<FRidh>
drakonis: you can use nixFlakes from nixpkgs to build the flakes branch of nix, and then use that build.
FRidh has quit [Quit: Konversation terminated!]
<drakonis>
its outdated
<drakonis>
oh
<ixxie>
joepie91: well, my general feeling has been that for the time being its not really realistic to build a decentralized platform
<ixxie>
joepie91: I was thinking to bet on some kind of federated model and wait until some tech matures before going fully decentralized
<ixxie>
joepie91: ashkitten has given me food for thought, and I'm wondering if it may be wiser to start even fully centralized to avoid the mess
<ixxie>
although, I am not entirely sure whether or not there are solutions to many of those issues
<__monty__>
ixxie: Open federation is decentralized. You probably mean "distributed."
<ashkitten>
i think matrix's approach has been fairly good, tbh
<ixxie>
__monty__: indeed, lost that word
tazjin has quit [Quit: Byeee]
tazjin has joined #nixos-chat
<ixxie>
ashkitten: yeah but its quite a bet to think you can build a complex arbitrary stack on top of it
<ixxie>
although maybe you can treat it like a WAL for your arbitrary stack, I donno
<ixxie>
the problem is, the kind of social platform I am interested in building is complicated in other ways (involving low-code modular apps that users can use to augment their social spaces)
<ixxie>
so I would like to solve that first, and let others improve the decentralized stuff, and then adopt it
<joepie91>
__monty__: federation is also distributed :P
<ashkitten>
might be hard to move from a centralized platform to a decentralized one
<joepie91>
ixxie: one thing to consider is that a P2P system will be much more difficult to design correctly since you don't have *any* system that you can centralize trust on even partially
<ixxie>
there is also a very valid critique from the signal creators about the issues regarding protocol governance
<joepie91>
ixxie: yes; that of flagged experimental out-of-spec implementations in the wild and then standardizing based on that, ie. how basically all the web standards were being developed before Google just completely took over
<joepie91>
ixxie: which also immediately brings me to one of the many, many errors in Moxie's post: "We got to HTTP version 1.1 in 1997, and have been stuck there until now. "
<joepie91>
this is false
<joepie91>
there have actually been many different extensions to it since, but he's conveniently only considering the base protocol
<energizer>
i have a computer and the console says it's out of space. when i try to log in, it won't go. do i need to remove the disks and pull data off manually?
<gchristensen>
is this why XMPP is such a successful project?
<joepie91>
which is a bit like saying "yeah well the Signal wire protocol hasn't changed in 6 years"
<joepie91>
gchristensen: XMPP has gone off the rails due to mismanagement
<joepie91>
people love to cite XMPP when complaining about federated protocols, but that's more than a little cherry-picking
<drakonis>
xmpp still sees usage
<drakonis>
despite having google smother it
<joepie91>
yes, XMPP is a nightmare, but it has a very easily explainable reason: there was never a coordinated attempt to test the standards around a user-friendly implementation
<joepie91>
XMPP clients have sucked since basically forever
<ashkitten>
whereas matrix.org is heavily focused on user experience ime
<qyliss>
Until fairly recently
<joepie91>
yeah, this is the main reason I see a future for Matrix and not XMPP, UX is an actual thing that is being taken seriously
<qyliss>
Do you really not see UX being taken seriously by Conversations?
<joepie91>
qyliss: the problem is that it has never been accounted for on a protocol level
<philipp[m]>
Yeah, it's really amazing how xmpp has been trying to implement crypto properly since before matrix was around and it's still a mess while matrix seems to have it figured out now.
<__monty__>
joepie91: I know but the most common terminology seems to be centralized (mainly politically, i.e., facebook's chat is centralized), decentralized usually implying (open) federation (again largely politically though there's a minimum technical aspect) and distributed aka fully distributed aka peer to peer.
<joepie91>
UX isn't just glossy buttons in the UI, it's a complex system of interactions and behaviours that requires sensible access to certain data in a certain format
<qyliss>
XMPP cryptography works fine and has for years
<ashkitten>
i'll be honest, having to decide what xmpp server to use is a big reason i've never used it. that's one of the big reasons matrix is looking at p2p, actually
<joepie91>
if your protocol is an archaic nightmare to work with and just outright doesn't support a ton of things that you need to smooth out the edges in your client, then you're not going to be building a client with good UX
<gchristensen>
ashkitten: same reason I really like Signal
<joepie91>
__monty__: distributed is a separate axis entirely
<philipp[m]>
xmpp crypto is a mess in multi device setups.
<joepie91>
centralized, federated, P2P are a social axis, whereas monolithic vs. distributed exist on a technical axis
<philipp[m]>
And most people have at least a computer and a phone.
<ashkitten>
i only started using matrix because it boasts e2ee to such an extent i felt comfortable signing up on matrix.org, the flagship instance
<qyliss>
In what way?
<joepie91>
and yes, OTR has always been unworkable with multi-device
MichaelRaskin has joined #nixos-chat
<qyliss>
I use it with multiple devices, as does everybody I talk to
<ashkitten>
(i eventually started self-hosting)
<joepie91>
OMEMO seems to have fixed that, but most clients don't support it
<qyliss>
And I've never had an OMEMO problem
<qyliss>
Most usage supports OMEMO though. Doesn't matter if the long tail of clients doesn't.
<joepie91>
this is starting to feel a lot like an "it works for me so therefore there is no problem" conversation
<joepie91>
which I'm not particularly interested in
<qyliss>
I'm not sure who it doesn't work for though
<philipp[m]>
No key signing at all.
<joepie91>
the vast majority of people, in some way, in my experience.
<joepie91>
I have actively used XMPP for many years, I have tested many clients, I have developed on a client
<ashkitten>
qyliss: i was looking at xmpp clients and asked my friends which i should use from a few that looked good, and they told me "none of those" because they all have problems or use the same backend library with some major issue (generally related to e2ee)
<pie_>
well this is new
<pie_>
i cant seem to use google without js anymore
<qyliss>
There are many XMPP clients, and most of them are bad
<qyliss>
There are also many web browsers, and most of them are bad
<joepie91>
it was an incredibly complex protocol (certainly not helped by the XML-ness), with three competing implementations for many features, with every client supporting a different one
<qyliss>
Or many email clients, most of which are bad
<joepie91>
and there was basically no reasonable common set of features that would work across all platforms with reasonable UX
<qyliss>
This is what you get with successful open protocols
<qyliss>
If I develop a Matrix client that doesn't support E2E, that's not an argument against Matrix, it's an argument against my bad Matrix client.
<joepie91>
qyliss: until a few years ago, there were 0 (zero) good XMPP clients, and there have been zero for most of its lifespan
<joepie91>
this is not comparable to other protocols
<qyliss>
I'm not really interested in what the situation was in the past
<qyliss>
If there are good clients now, the situation five years ago is irrelevant to me
<joepie91>
then that's your problem, because the past situation 100% matters for understanding how XMPP got to where it is today, and why people are looking at Matrix instead
<ashkitten>
qyliss: i was told there is one single "good" xmpp client i could use on my computer, and one single "good" xmpp client for my phone
<qyliss>
that is correct
<ashkitten>
i have a problem with that
<qyliss>
how many good matrix clients are there for your phone?
<ashkitten>
because i don't want to use that xmpp client
<ashkitten>
qyliss: 3 and counting
<qyliss>
oh that's a very pleasant surprise to hear!!
<qyliss>
what are the two that aren't riot?
<joepie91>
qyliss: even many of the heavily-under-development-and-definitely-not-done-yet Matrix clients already present better UX than nearly every XMPP client
<ashkitten>
qyliss: ditto, and one other i forget the name
<ashkitten>
ditto is also cross-platform with react native
<ashkitten>
the other one uses... flutter, i think?
<qyliss>
That's very cool!
<ashkitten>
i'm not sure about the status of all features but they do tend to reach parity fairly quickly because there are good platform libraries
<qyliss>
fwiw, I don't think "people are looking at Matrix instead" is anywhere near universal
<joepie91>
the point that I am trying to make here is that in XMPP-land, there never has been a strong culture around UX - which manifests in many different ways, including awkward protocol design, bad clients, and so on. one single "good" XMPP client doesn't really change that, if you are trying to treat it as an open protocol - because it means that in practice, the good client exists *despite* the state of XMPP, not *because* of
<joepie91>
it
<qyliss>
The people who got me to use XMPP are people who gave up on Matrix and then looked for an alterantive
<ashkitten>
cross-signing is a sore spot with most clients that aren't riot, i think nheko is the only one that supports that right now, but i'm confident everything else will catch up quickly
<joepie91>
qyliss: let me guess, the issue they have with Matrix is performance
<qyliss>
No, the issue they had with Matrix was that encryption never worked
<joepie91>
what
<ashkitten>
qyliss: i think you'll be very surprised by the state of encryption now, with cross-signing
<qyliss>
They'd join a chat and be bombarded with "you don't have a key for this user", 50 times
<qyliss>
The last time I heard of this happening was a few months ago
<ashkitten>
yeah, that issue is completely resolved now with cross-signing
<ashkitten>
which was rolled out a few months ago
<ashkitten>
the encryption ux is so much better now
<qyliss>
hearing that sort of story doesn't exactly fill me with confidence that matrix is designed "with UX in mind" or whatever
<qyliss>
It sounds like in both cases, there was historically bad UX, and now there is much better UX
<joepie91>
ashkitten: months? has it really been months already?
<joepie91>
qyliss: the bad pre-cross-signing UX was exactly why E2EE was not enabled by default...
<joepie91>
which was specifically because they wanted good UX
<ashkitten>
qyliss: i don't see your point, cross-signing is a feature they've been working on basically since matrix was first conceived
<joepie91>
there is a difference between "do not take UX into account" and "have not gotten around to building it right yet"
<ashkitten>
a lot of these features are
<qyliss>
If all of this has truly been fixed then that's great news I was not aware of
<ashkitten>
it is
<joepie91>
this is precisely why I am hammering on "culture around UX" though
<qyliss>
But when I came to these ecosystems around six months ago, from my perspective XMPP had working encryption OOTB, and Matrix had encryption that was disabled by default and didn't work properly.
<joepie91>
the current state of something only tells you what something is like right now, the culture around something tells you what it will be in a year or two
<qyliss>
I'll be happy to reëvaluate Matrix now that that's apparently been solved
<joepie91>
the culture around Matrix is "let's build something with good UX", the culture around XMPP is "noone really cares about UX except for this one client dev", that is why I do not see a future in XMPP
<joepie91>
(that's certainly not to say that the Matrix ecosystem is perfect, but it at least tries to get there)
<joepie91>
well, correction: it is why I do not see a future in XMPP as an open protocol
<qyliss>
Is Fractal likely to get me the current state-of-the-art Matrix experience?
<qyliss>
Also, how easy is to run a Matrix server with NixOS, and what's its resource consumption like?
<ashkitten>
no, because i don't think it supports e2ee
<joepie91>
Riot is the reference implementation on desktop that has All The Features, with Nheko trailing fairly closely behind AFAIK
<ashkitten>
qyliss: i run a matrix server, lemme go pull up the nixos-config
<cole-h>
cc colemickens -- how's mirage shaping up?
<joepie91>
qyliss: resource consumption is still junk. running Synapse is trivial though
<qyliss>
Riot is some sort of web thing though, right?
<joepie91>
config-wise
<joepie91>
qyliss: riot is a HTML/CSS/JS client, can be run either in a browser or as a standalone Electron thingem
<qyliss>
So will have poor Electron's poor Wayland support :(
<qyliss>
Are there other current desktop clients?
<ashkitten>
nheko
<joepie91>
that would be aforementioned Nheko
<joepie91>
Qt, not sure how nicely that plays with Wayland
<ashkitten>
nheko is the most featureful non-riot client atm
<ashkitten>
it's missing some features but catching up very quickly
<ashkitten>
i haven't used it though
<joepie91>
qyliss: Riot works fine in Firefox, btw
<joepie91>
so if the Wayland issue is a Chromium one, you're not stuck with that
<qyliss>
I really don't want to be using a web-based chat program
<qyliss>
Qt is fine with Wayland
<MichaelRaskin>
Re: Synapse — Synapse resource consumption as actually fine as long as everyone on your server only participates in the rooms where they personally know everyone
<joepie91>
🤷♂️
<qyliss>
> The motivation behind the project is to provide a native desktop app for Matrix that feels more like a mainstream chat app (Riot, Telegram etc) and less like an IRC client.
<{^_^}>
error: syntax error, unexpected ',', expecting ')', at (string):318:129
<qyliss>
Is this implying there is a native desktop app for Matrix that feels like an IRC client?
<ashkitten>
i've never used riot-desktop, and don't really have any incentive to, because nixpkgs' riot-desktop doesn't build with seshat which would give it e2ee room search
<qyliss>
Because if so I'd like to hear about that
<energizer>
tbh i'm increasingly more comfortable using web based tools these days since linux doesn't do application sandboxing by default
<joepie91>
qyliss: there's weechat-matrix
<cole-h>
qyliss: The IRC bridge :P
<joepie91>
also pretty full-featured afaik
<qyliss>
cole-h: you can't bridge IRC -> Matrix, can you?
<qyliss>
I thought it was only the other way around
<qyliss>
joepie91: INTERESTING
<cole-h>
Indeed.
<cole-h>
Matrix -> IRC
<cole-h>
(AKA I misunderstood)
<joepie91>
cole-h: qyliss: there's some matrix-ircd project sitting on a shelf rotting away somewhere
<qyliss>
energizer: yeah, working on that :)
<energizer>
qyliss: in spectrum?
<qyliss>
yeah
<qyliss>
Sounds like there probably is a Matrix client for me, then
<ashkitten>
also, for e2ee support in clients that don't have it, you can technically use... i forget the name, matrix-org wrote a mitm thing which transparently handles e2ee for clients that don't support it
<MichaelRaskin>
pantaleimon
<ashkitten>
^
<joepie91>
pantalaimon
<qyliss>
Unlikely I have the capacity to run synapse though if it's as heavy as people say it is, which is a shame
<joepie91>
but that doesn't give you a first-class cross-signing experience AFAIK
<energizer>
qyliss: that would be cool.
<energizer>
i'd be pretty happy with just
<joepie91>
qyliss: if you never join large channels it'll be fine
<energizer>
i'd be pretty happy with just firejail applied to nixpkgs
<qyliss>
energizer: you might want to talk to MichaelRaskin about that
<joepie91>
qyliss: but join a bridged IRC channel or a large Matrix room and you are in for a painful experience, currently
<energizer>
MichaelRaskin: you have something going with limiting capabilities of nixpkgs apps?
<qyliss>
I'm unlikely to do that I think
<ashkitten>
dendrite is rapidly catching up to synapse though, and they will provide a migration path from synapse to dendrite
<MichaelRaskin>
qyliss: I run a small Synapse, it federrates with matrix.org (for small rooms) and consumes very little
<MichaelRaskin>
energizer: well, there is a catch
* joepie91
is looking at Conduit with interest, currently
<energizer>
MichaelRaskin: i just want to be able to run rando software and not have it read my homedir or connect to the internet
<MichaelRaskin>
I use nsjail and some code for generation of parameters to it and for socat-based proxying of sockets
<MichaelRaskin>
(so, I might have to let WebRTC-intended Firefox/Chromium instances connect to Internet unfiltered, but some instances have only Squid access)
<MichaelRaskin>
I am not sure how to auto-jail
<MichaelRaskin>
My wrappers have their whole own CLI logic
<MichaelRaskin>
So that when I launch a program I can tell what to give it
<MichaelRaskin>
A ton of time there is a current directory with relevant files and only them, and that's what you want to give to the program, for example
<ashkitten>
MichaelRaskin: every time you talk about your systems i'm awed by how weird and neat it is, and how much of a pain it must be to use them
<ixxie>
hmm... could I license literature with GPL 2/3
<joepie91>
ixxie: could, yes. but the GPL is only really designed to work with code. possibly CC-BY-SA is what you're looking for, which is a more general copyleft license for cultural works
<qyliss>
There's also the GFDL but CC-BY-SA is probably better
<ixxie>
I wanna force people to contribute back
<ixxie>
but yeah, the share-alike clause would do the trick I guess
<abathur>
ixxie: curiosity unrelated to answering your question; what kind of lit?
<cole-h>
Well, you can see how well GPL does with that (/me looks at various OEMs not releasing their kernel sources)
<joepie91>
ixxie: keep in mind that "forcing people to contribute back" is certainly not guaranteed by making your license require it
<ixxie>
abathur: well, I have this idea for a simple publication site, where people can evolve and remix the material
<qyliss>
cole-h: GPL enforcement in that sort of case generally does work when somebody takes it up
<qyliss>
D-Link, etc.
<ixxie>
abathur: I'm thinking of somehow potentially leveraging git to allow people to fork and remerge work, but its all sort of vague in my mind
<ixxie>
abathur: if I can somehow make it work without needing authentication it would be neat, but I think it might be necessary
<ixxie>
abathur: I acquired the domain fractal.press :D
<abathur>
nice
<ixxie>
this is somehow the lower priority side project though
<MichaelRaskin>
qyliss: on the other hand, VmWare; and in the ODM case, how many iterations did it take once Tesla agreed in principle until the code from NVidia they blindly relayed started to actually build (let alone be anything related to the deployed code)
<abathur>
always more plates/irons/burners than hands
<MichaelRaskin>
There is something aesthetically impressive about how TP-Link approaches GPL obligations
<joepie91>
ixxie: personally I am not convinced that demanding share-alike in a license actually does anything more at all than just nicely asking
<pie_>
MichaelRaskin: what do they do
<joepie91>
in fact, I suspect that the latter can work a lot better, because it places a social expectation, rather than a rule that can be rule-lawyered around by a shitty company
<ixxie>
joepie91: I can do both
<joepie91>
(which means social consequences, rather than "well they technically adhered to the license" apologism)
<joepie91>
ixxie: if you want the social consequences for bad open-source citizens, that seems mutually exclusive with demands-through-license to me
<ixxie>
joepie91: in any case, I can also add a no commercial use clause to avoid companies getting involved, or at least discourage them
<joepie91>
as the point of asking for something more than the license demands, is to signal "hey we expect you to generally not be a shitbag"
<joepie91>
and that doesn't work if your license contains all the requests :)
<joepie91>
ixxie: no-commercial-use makes it not open-source, fwiw
<joepie91>
(and not just due to a technicality)
kalbasit has quit [Ping timeout: 272 seconds]
<joepie91>
aside from the general problem with defining what exactly constitutes "commercial", which is... a fuzzily explored topic, at best... it also suffers from the general problem of license-specific usecase restrictions, which is that the more people combine works, the less people can use it, every iteration
<MichaelRaskin>
pie_: well, they recognise the obligation to provide full source, and tooling sufficient to do the things no doable by standard tooling
<pie_>
ah
<ixxie>
joepie91: how does that happen?
<MichaelRaskin>
They cleanly separate the underlying boot image and the proprietary webapp (that nobody should want, IMO)
<joepie91>
ixxie: say that you have demographics 1, 2, 3, 4, and 5. now project A disallows demographic 1 from using their thing, project B disallows demographic 2 and 3, project C disallows demographic 4 and 5
<joepie91>
ixxie: if you now compose something from parts of project A, B and C, noone can use it, because every demographic is disallowed by *some* part of the work
<MichaelRaskin>
For the base image they actually kept the source together. I think even separating upstream sources and patches. And object files. And the patched GCC binaries to build these. And the patched GCC sources. And whatever.
<joepie91>
(this is the same reason that "ethical use licenses" don't work and are not open-source)
<MichaelRaskin>
All this adds up to ~1GiB of stuff for any given device
<ixxie>
joepie91: it depends on your definition of open-source; I can see how they are unfree, but since everything needed to recreate the binary is available, its still open-source to me
<MichaelRaskin>
Now the beautiful part: they just tar all of it. No, they do not compress.
<joepie91>
ixxie: open-source means much more than "can reproduce the binary"
<samueldr>
well... they do sell networking gear
rajivr has quit [Quit: Connection closed for inactivity]
<samueldr>
so it's in their interest to keep the bytes flying
<pie_>
ha
<MichaelRaskin>
So they put 1 GiB tarball of everything, uncompressed, on a moderately slow FTP server.
<ixxie>
joepie91: well, I don't really subscribe to prescriptivist views of language, I don't believe you can define a word; meaning follows use in practice, so I believe that in order to understand what a word means you need to empirically survey how people use a word, generally speaking.
<ixxie>
joepie91: I can accept all of these things are important, and that I should reconsider my license approach, but I'm not sure what open-source means to everybody in the community
<MichaelRaskin>
pie_: It does seem they achieved the lowest effort necessary to provide something that will not get complaints about insufficiency
<joepie91>
ixxie: that's a very superficial view of descriptivism. descriptivism doesn't mean "it means whatever people use it to mean", it also encompasses active efforts to retain a specific distinct meaning, especially for commonly-abused terms
cole-h has quit [Quit: Goodbye]
<MichaelRaskin>
And well, there _are_ some benefits in knowing you get this one tarball and that's everything included
<joepie91>
ixxie: as is the case with open-source; there are plenty of commercial parties trying to devalue the term "open-source" such that they can benefit from the goodwill it generates, without actually making the corresponding contribution to the public domain
<joepie91>
ixxie: it is not prescriptivist to speak out against that and refuse to recognize that coopting of the term
<joepie91>
(especially since people *do* still associate certain benefits with "open-source" that would not be gotten if it were not free to use as desired)
cole-h has joined #nixos-chat
<ixxie>
joepie91: I'm not saying 'give equal weight to every position'
<MichaelRaskin>
And a gigadump is among the better cases OpenWRT has to deal with…
<ixxie>
joepie91: but lets say we focus on the good meaning open-source community and survey *those* people about the meaning of open-source
<ixxie>
joepie91: I am not convinced you will get exactly that definition
<joepie91>
TL;DR: arguing for precise usage of a term is not prescriptivism
<joepie91>
prescriptivism, rather, is the ideology that language is immutable
<samueldr>
MichaelRaskin: yeah, what you're describing isn't top tier reproducible nix derivation... but it's also pretty good to *also* have the compiler patches
<samueldr>
much better than is often seen
<joepie91>
ixxie: of course you won't, because people don't typically think in definitions. but if you survey people's *expectations*, you will find that those align pretty much with the definition I linked
<MichaelRaskin>
samueldr: patches _and_ binaries
<samueldr>
like some mediatek phones requiring a binary off of the mediatek release, binary based on dtc, dtc is gpl too... no sources in sight
<ixxie>
joepie91: I'm no sure about that
<samueldr>
(binary at build time)
<ixxie>
joepie91: I wonder if somebody did a survey
<joepie91>
ixxie: I have, for many years :)
<samueldr>
so right now no way to build those kernels on aarch64 without expending some effort in reversing that binary
<joepie91>
(informally)
<samueldr>
though, tbf, I assume that reversing would be a cinch... debug symbols are present, and the original source for dtc is known (it's android's fork)
<MichaelRaskin>
samueldr: it is pretty plausible that MediaTek itself doesn't have a consistent set of sources (in the sense of just grabbing object files that were scattered around)
<samueldr>
plausible
<samueldr>
which doesn't excuse it!
<MichaelRaskin>
Well, sure
<MichaelRaskin>
Neither does it improve the confidence in all that mess…
<MichaelRaskin>
(Tesla experiment hints that Nvidia is also a mess)
<samueldr>
oh, it looks like a mess even for their "open" platforms
<ixxie>
anyway, thanks for food for thought joepie91
<MichaelRaskin>
TP-Link is a bit funny in the sense of «put everything in a single tarball _and never compress_», but they seem to actually have some idea what is they build process, and I guess any improvement would actually be costly and not provide much improvement
<joepie91>
ixxie: I do recommend doing an informal survey over time, btw :P
<MichaelRaskin>
I mean, their proprietary webapp on top also has its share of mess…
<ixxie>
joepie91: I'll keep it in mind :D
<drakonis>
MichaelRaskin: entirely unsurprising, given nvidia's history
<samueldr>
MichaelRaskin: it wouldn't surprise me, if they do ODM things, that this is also the way they provide things to their clients
ixxie has quit [Quit: Lost terminal]
<samueldr>
and changing it means *also* changing documentation about that
bkv has joined #nixos-chat
bqv has quit [Ping timeout: 246 seconds]
bqv has joined #nixos-chat
bkv has quit [Ping timeout: 272 seconds]
Ashy has joined #nixos-chat
polezaivsani has quit [Remote host closed the connection]
__monty__ has quit [Quit: leaving]
<bqv>
lmao classic
<bqv>
i installed a browser, tried to browse to a url
<bqv>
"TLS/SSL support not available; install glib-networking"
* bqv
facepalm
parsley936 has quit [Remote host closed the connection]