bennofs__ has quit [Read error: Connection reset by peer]
bennofs_ has joined #nixos-dev
<sterni>
yorick: that is more trouble than its worth because we not only package stackage, but also stuff from hackage
<sterni>
it has been the experience that it is more work getting hackage to work with lts than nightly with hackage
<sterni>
qyliss: yorick: haskellPackages were more or less marked as broken automatically since forever, we now just have a script instead of some kind of emacs based workflow peti had
<qyliss>
yorick and I were talking about automatically marking unbroken yesterday
<qyliss>
with peti's workflow, at least, fixed packages were never marked unbroken
<sterni>
yeah we have some improvements to that
<sterni>
for one we don't mark reverse dependencies as broken anymore, so when unmarking something as broken you don't have to track down the reverse dependency tree anymore
<maralorn>
yorick: The main argument to use LTS was some kind of "It wouldn‘t be right to do not lts bumps on stable".
<maralorn>
yorick: But the reality is, we just freeze the packageset for stable on branch off and I can‘t think of another solution for that.
<maralorn>
So it does not really matter if we freeze a nightly or a lts.
<sterni>
well we could make a lts set on branch-off
<maralorn>
Normally after branch-off people start to manually edit hackage-packages.nix so after that we can‘t reasonably update.
<sterni>
and just keep updating with the same lts solver
<maralorn>
sterni: Yeah, but then we have to decide if we update hackage or not.
<sterni>
maralorn: I mean with the new scripts that could be a thing of the past :)
<sterni>
maralorn: yeah that is the big open question how to partially freeze the hackage packages
<maralorn>
I am not sure. I feel like bumping stackage but staying on an old hackage snapshot will fail.
<maralorn>
And regularly updating hackage on stable will break lots of stuff and double maintenance efforst.
<maralorn>
Besides I think CVEs in in Haskell packages are rather rare. If one happens I am sure we can find a manual solution to update them.
<sterni>
we would have to generate pvp upper version bounds for the hackage pin
<sterni>
and then bump select beyond breaking bounds which should happen rarely, but we would get non breaking updates
<maralorn>
sterni: Let’s say I wouldn‘t oppose a PR with a well thought out and easily maintainable solution.
<maralorn>
But I don‘t see the benefits.
<sterni>
well having maintained haskellPackages in stable
<sterni>
the question is does anyone care about haskell in stable enough to keep maintaining it…
<maralorn>
sterni: And I think especially for Haskell "maintained" is a really strong word. There is no guarantee whatsoever that any Bug in any Hackage package will be fixed in a minor version bumps. I have never heard about a backage which worries about fixing as much on 1.4.x before bumping to 1.5.0.
<maralorn>
*package^^
<maralorn>
Otoh I have to admit I am kinda arguing against the existence of stackage lts here. Which is probably a strong claim.^^
<sterni>
yeah, I mean I can't really judge this too well since I have no prior experience with lts, does it do breaking bumps for certain packages?
<sterni>
I think we would just have to do it if it was something we were interested in
vcunat has quit [Ping timeout: 260 seconds]
vcunat has joined #nixos-dev
orivej has quit [Ping timeout: 265 seconds]
<maralorn>
If someone wants to champion it I am all for it.
<siraben>
even with nix-prefetch-github, I'm extracting the relevant URL parts manually and there's hundreds of occurences
<yorick>
domenkozar[m]: oh, neat. I'm more interested in armv7l images on the rpi4 though
<domenkozar[m]>
hmm, why is that?
<yorick>
domenkozar[m]: because I'm using it as a builder for armv7l packages
<yorick>
raspbian is also armv7l by default
<domenkozar[m]>
it should be almost the same, plus all the waiting
<yorick>
ooh, waiting
<yorick>
domenkozar[m]: I'd recommend diffing lsblk instead of looking at dmesg
<domenkozar[m]>
hmm, why?
<yorick>
my dmesg has a lot of spam, for instance
<domenkozar[m]>
like all the time?
<yorick>
not all the time, but I have periods where it prints about 5500 instances of 'kernel: Bluetooth: hci0: SCO packet for unknown connection handle 0' per second
<yorick>
system works fine otherwise
<yorick>
so I prefer looking at lsblk :)
<domenkozar[m]>
:O
<yorick>
(you asked :D)
<yorick>
I'll try it later today (famous last words)
<domenkozar[m]>
yeah thanks, I'll keep dmesg and the user can just retry or grep away. diffing sounds error prone so I prefer events
<yorick>
domenkozar[m]: does the image have ssh enabled?
<domenkozar[m]>
nope
tomberek has quit [Quit: Connection closed]
jonringer has joined #nixos-dev
cole-h has joined #nixos-dev
teehemkay has quit [Remote host closed the connection]
teehemkay has joined #nixos-dev
<Ericson2314>
siraben: I added you
<siraben>
Ericson2314: thank you
orivej has joined #nixos-dev
Reventlov has quit [Read error: Connection reset by peer]
<qyliss>
Ericson2314: add me too?
<Ericson2314>
gladly!
<qyliss>
ty :)
<Ericson2314>
np!
<Ericson2314>
sterni: I see you requested; added you too
justan0theruser has quit [Ping timeout: 260 seconds]
vikanezrimaya has quit [Quit: Connection closed]
<thibm>
das_j: why the option is internal?
<ivan>
open-vm-tools is busted in master, error: template with C linkage
<yorick>
misuzu: do you know how it compares to an rpi4?
vikanezrimaya has joined #nixos-dev
<sphalerite>
Mic92: I can't make it to the RFC meeting this Thursday
<Mic92>
sphalerite: ok
<Mic92>
sphalerite: hope the others come
<Mic92>
kloenk is also not there.
<kloenk>
<Mic92 "Finn Behrens is also not there."> I don’t know anything about that. I should have time to join xD
pmy has quit [Ping timeout: 240 seconds]
pmy has joined #nixos-dev
<sphalerite>
lol
<das_j>
thibm: No real reason, I thought I'd stick to the unit option. Also it's probably a bit more advanced so a less experienced user shouldn't find it in the option index since the option probably does something they didn't expect
<radvendii>
I ran "nixos-rebuild build-vm" and then ./result/bin/run-foo-vm and got "gtk initialization failed". Couldn't find anything about it online
<radvendii>
this is all on a headless server
<cole-h>
run-vm starts up a qemu instance IIRC
<cole-h>
so that's probably why
<radvendii>
i'm not understanding the connection. is this a situation where qemu doesn't work?
<radvendii>
i'm not super familiar with qemu
rajivr has quit [Quit: Connection closed for inactivity]
<cole-h>
sorry, not qemu, it spawns a GUI window with the VM
<cole-h>
I don't remember what the tool is
<radvendii>
ohh
<cole-h>
but it uses gtk, so that's why you're seeing that
<radvendii>
is there a way to run it headless and just get dropped into a prompt?
<qyliss>
maybe QEMU_OPTS=-nographic
<qyliss>
cole-h: it is QEMU
<symphorien[m]>
nixos-shell does this iirc
<cole-h>
jfkdlsa
<cole-h>
sorry, not-not qemu :P
<radvendii>
QEMU_OPTS=-nographic gets me past that error, but now my terminal is just blank and unresponsive. it becomes responsive again if I kill qemu
<radvendii>
(C-c doesn't help)
<radvendii>
oh it turns out that was totally unrelated. also, i totally thought i was in #nixos, sorry for posting questions in this chat thanks for the solutions!
aleph- is now known as Batty
<clever>
radvendii: you have to tell nixos to launch a serial getty, one sec
<clever>
radvendii: virtualisation.graphics = false; will do the -nographic for you, and also fix login
<jonringer>
Is there a way to get hydra to essentially `--impure` when building something?
<jonringer>
to get hydra to do an impure build* as, in `--impure` when using the nix commands*
Valodim[m] has quit [Ping timeout: 245 seconds]
Valodim[m] has joined #nixos-dev
fuzzypixelz49 has joined #nixos-dev
<fuzzypixelz49>
can you point me to a derivation that deals with a nested CMake projet? i.e. one that needs to be compiled separately and linked to the main one?