<ndowens08>
this is a pain, python is now missing setuptools and some of the names of the packages is rediculously long lol
tvon has joined #nixos
<HappyEnte>
ndowens08, yeah it can be quite a pain :) but its worth it.
digitus has quit [(Quit: digitus)]
sziszi has quit [(Ping timeout: 255 seconds)]
sziszi has joined #nixos
kampfschlaefer has quit [(Ping timeout: 256 seconds)]
drakonis has joined #nixos
<drakonis>
gdm is broken right now, isn't it?
viric has quit [(Ping timeout: 255 seconds)]
georges-duperon has joined #nixos
markus1189 has joined #nixos
markus1219 has joined #nixos
markus1209 has quit [(Ping timeout: 252 seconds)]
<drakonis>
does gnome's display manager not boot up for anyone else?
markus1199 has quit [(Ping timeout: 256 seconds)]
viric has joined #nixos
Sonarpulse has joined #nixos
georges-duperon has quit [(Ping timeout: 252 seconds)]
systemfault has joined #nixos
mkoenig has quit [(Ping timeout: 240 seconds)]
modulistic has quit [(Ping timeout: 240 seconds)]
<gchristensen>
I'm having a weird issue, where I can't ssh in to a nixos machine as root with a password. I can log in as a user with a password, or root with a key, but with a password it doesn't work -- I always get authentication failures. I've set permitRootLogin to "yes". any ideas?
<pie__>
gchristensen, have you tried looking at the logs?
<pie__>
i dont actually know whatere/what, so jus saying
<pie__>
*where/what
sigmundv has joined #nixos
<sheenobu_>
i was hoping for something a bit more precise but i'm guessing it can't be helped
<gchristensen>
pie__: yeah, nothing in particular. auth failures messages from PAM and ssh
<gchristensen>
sheenobu_: what do you mean?
mkoenig has joined #nixos
<pie__>
i had a weird issue where i needed to lower my mtu or something, but im pretty sure that was broken independent of user i used
erasmas has quit [(Quit: leaving)]
hendrik[m] has joined #nixos
<Ralith>
gchristensen: they probably want to collect only the transitive closure of the result link
<Ralith>
(when applicable)
mudri has quit [(Quit: WeeChat 1.6)]
Shou has quit [(Ping timeout: 276 seconds)]
<gchristensen>
ah
<joachifm>
sheenobu_: nix-store --delete ./result
<joachifm>
might do what you want
* Ralith
takes notes
<Ralith>
joachifm: is there a transitive version of that?
mkoenig has quit [(Ping timeout: 264 seconds)]
<joachifm>
Ralith: you could nix-store --delete the result of nix-store --query with the appropriate flags
<joachifm>
Ralith: possibly there's a canned thing you can use, but I can't recall such a thing offhand :)
* Ralith
nod
mkoenig has joined #nixos
<Ralith>
wouldn't necessarily expect a canned thing, but nice to know the right primtiives are handy
Sonarpulse has quit [(Quit: Leaving)]
Sonarpulse has joined #nixos
Shou has joined #nixos
phreedom has joined #nixos
ryanartecona has quit [(Quit: ryanartecona)]
eacameron has joined #nixos
mkoenig has quit [(Ping timeout: 276 seconds)]
mkoenig has joined #nixos
<c74d>
I have a Perl module (perlPackages.Swim) installed on NixOS, and I notice that every time I run the Nix store garbage-collector, the tarball for that Perl module gets collected, even though the nixpkgs expression for the module isn't being updated, so why would it keep getting downloaded?
eacameron has quit [(Ping timeout: 240 seconds)]
<joachifm>
c74d: have you set gc-keep-outputs?
Phlogistique has quit [(Ping timeout: 255 seconds)]
<c74d>
joachifm: no, I've not
Phlogistique has joined #nixos
<joachifm>
c74d: that might explain it; by default, everything that isn't registered as a gc root is deleted; that'd include source archives
eacameron has joined #nixos
Unode has quit [(Ping timeout: 240 seconds)]
<c74d>
I want it to be garbage-collected, but I don't like that it keeps getting downloaded for no reason I know of
<c74d>
I suppose what I want to know is, why does Nix keep (I assume) re-installing that package, such that it keeps re-downloading the source archive?
<joachifm>
c74d: ah, I see what you mean now
<c74d>
(I have autoUpgrade enabled, so my system is rebuilt nightly)
Unode has joined #nixos
eacameron has quit [(Ping timeout: 255 seconds)]
eacameron has joined #nixos
eacameron has quit [(Ping timeout: 240 seconds)]
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/vDgSz
<NixOS_GitHub>
nixpkgs/master 4000489 Adam Beckmeyer: olm: init at 2.2.1
eacameron has joined #nixos
eacameron has quit [(Ping timeout: 264 seconds)]
bennofs1 has quit [(Quit: WeeChat 1.6)]
phreedom has quit [(Remote host closed the connection)]
systemfault has quit [(Quit: Bye!)]
modulistic has joined #nixos
georges-duperon has joined #nixos
roconnor__ has quit [(Ping timeout: 245 seconds)]
modulistic has quit [(Ping timeout: 258 seconds)]
georges-duperon has quit [(Ping timeout: 252 seconds)]
hamishmack has quit [(Remote host closed the connection)]
panostimos has quit [(Ping timeout: 255 seconds)]
hamishmack has joined #nixos
eacameron has joined #nixos
modulistic has joined #nixos
derjohn_mob has quit [(Ping timeout: 276 seconds)]
eacameron has quit [(Remote host closed the connection)]
<k0001>
What's the path that needs to be sourced from `.bashrc` or similar for a Nix install done with `https://nixos.org/nix/install` ?
eacameron has joined #nixos
systemfault has joined #nixos
eacameron has quit [(Ping timeout: 258 seconds)]
Somelauw has quit [(Quit: WeeChat 1.5)]
<gchristensen>
k0001: the installer should put it there // tell you after installing I think
derjohn_mob has joined #nixos
eacameron has joined #nixos
mkoenig has quit [(Read error: Connection reset by peer)]
nahamu has quit [(Remote host closed the connection)]
mkoenig has joined #nixos
s33se has quit [(Ping timeout: 245 seconds)]
sigmundv has quit [(Ping timeout: 240 seconds)]
roconnor__ has joined #nixos
<NixOS_GitHub>
[nixpkgs] grahamc opened pull request #22610: nixos: update default cases from KDM/KDE4 to SDDM/KDE5 (master...switch-to-kde5-by-default) https://git.io/vDgb5
nahamu has joined #nixos
Shou has quit [(Ping timeout: 245 seconds)]
mbrgm_ has joined #nixos
mbrgm has quit [(Ping timeout: 264 seconds)]
mbrgm_ is now known as mbrgm
ryanartecona has joined #nixos
<NixOS_GitHub>
[nixpkgs] bendlas opened pull request #22611: pypy: enable dbm support in db dependency (master...fix-pypy) https://git.io/vDgNA
roconnor__ has quit [(Ping timeout: 240 seconds)]
sophiag has joined #nixos
eacameron has joined #nixos
<sophiag>
can anyone advise me on how to setup up systemd to support efi? from the configuration man page it seems i need to add "boot.loader.systemd-boot.enable = true;" and "boot.loader.efi.efibootmgr.enable = true;" is there anything else? it's kind of a big deal because currently i have no ability to wake up my laptop if i set it to go to sleep...
mkoenig has quit [(Remote host closed the connection)]
<ToxicFrog>
sophiag: that should be sufficient, but note that that only affects *boot*; it won't have any effect on ACPI nonsense once the system is up and running.
<sophiag>
ah ok. any idea the settings for that or should i just look them up?
<ToxicFrog>
No idea, sorry :/ ACPI is a mess but the ways in which it's a mess are very system-specific.
dbmikus has quit [(Quit: WeeChat 1.7)]
<sophiag>
hah, ok. probably has to do with gnome...
davidak has quit [(Quit: Leaving.)]
<ToxicFrog>
(my wife's laptop, for example, claims to support sleep but the parts of the bios that would let it wake up from sleep are missing, because lenovo never bothered fully implementing S3, because they only tested it on windows and windows preferentially uses S0ix if it's available
<ToxicFrog>
(so the answer is to use "freeze" -- S0ix -- rather than sleep, but that's not well supported in general and it's not supported at all by GNOME or KDE, you have to do it from the command line or using acpid)
<sophiag>
yikes
<sophiag>
can she get it to freeze upon closing the lid?
<ToxicFrog>
Yeah, we ended up disabling all the "suspend on lid close" stuff in KDE and writing an acpid rule to freeze on lid close instead
<ToxicFrog>
But still. Frustrating.
<ToxicFrog>
Motherboard firmware claiming to support various ACPI modes that it either doesn't actually support or supports in really broken hacky unreliable ways is still sadly common. :(
alx741 has quit [(Quit: alx741)]
<Sonarpulse>
hmm is patchelf smart about cross compiling?
<sophiag>
ugh. i obviously disabled suspend lid on close as well in gnome, but it's not good to leave it that way in the long run if i can't freeze it
<sophiag>
is your acpid rule in configuration.nix?
<ToxicFrog>
I'd recommend trying freezing it manually first to make sure it actually wakes up from it, though, not everything supports S0ix
<ToxicFrog>
Hibernate might also be worth trying as an alternative (I think that's echo 'disk' instead of echo 'freeze', although gnome should support it natively)
<sophiag>
gnome does support hibernate. i can't remember if i tried it tho
* ToxicFrog
deploys the sleep. Good luck!
ebzzry_ has joined #nixos
ndowens08 has quit [(Ping timeout: 276 seconds)]
ebzzry has quit [(Ping timeout: 245 seconds)]
hexagoxel has quit [(Ping timeout: 256 seconds)]
alx741 has joined #nixos
<stepcut>
looks like the odroid-c2 is inching closer to having a usable mainline u-boot+kernel, which should make nixos support much easier
systemfault has quit [(Quit: Bye!)]
hexagoxel has joined #nixos
<Ralith>
somehow my /dev/null turned into a regular file with 0600 permissions
<Ralith>
what
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/vDgpM
<NixOS_GitHub>
nixpkgs/master a848643 Dan Peebles: pythonPackages.lmdb: fix on python 3 and add meta
<tnks>
"value is a function while a set was expected"
defel has quit [(Ping timeout: 245 seconds)]
viric has quit [(Ping timeout: 260 seconds)]
viric has joined #nixos
andymandias has joined #nixos
nahamu has quit [(Ping timeout: 258 seconds)]
nahamu has joined #nixos
andymandias has quit [(Ping timeout: 276 seconds)]
Fare has joined #nixos
<clever>
sophiag: also of note, a lot of hardware like the gpu needs special scripts to reset it after sleep mode, and if you directly write disk to things, it wont be able to resume them
<sophiag>
clever: is that a recommendation of sleep over hibernate or you're saying it could be an issue with either?
<sophiag>
*freeze i mean
<clever>
its better to use the correct scripts for it, when possible
<clever>
those handle saivng and restoring state
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/vDgj1
<NixOS_GitHub>
nixpkgs/master 03cab2d Dan Peebles: ecs-agent: init at 1.14.0
<sophiag>
would a window manager handle that correctly? otherwise i'm not sure how to implement what you're describing. if it's just the graphics card i'm not particularly concerned, but assume i could look it up
<clever>
i would expect the WM to do things properly
<sophiag>
well, i'm going to test hibernate in gnome once i finish up some work. but i'm not optimistic about it given it can't handle sleeping correctly
<unlmtd[m]>
Im getting into the nixpkgs/docs. Is xml chosen to make it more readable? Or was it written when markdown wasnt invented yet, or hasnt anyone used a conversion tool, or should I open an issue and maybe send a pr?
<clever>
sophiag: i also wound up testing hibernate by mistake on xfce
<clever>
sophiag: the laptop display was in the way, so i closed it, and it just hibernated, and woke up with zero problems
ryantm____ has joined #nixos
<sophiag>
clever: you're saying it woke up from hibernate, but not sleep? the default on gnome is sleep, but you can switch it to hibernate
<clever>
oh wait, it was probably standby, it woke up pretty fast
<sophiag>
(also tbh i'm embarrassed to say i don't know whether sleep and freeze are the same thing...)
<unlmtd[m]>
holy crap I cant read raw xml it makes my spirit go mad
<clever>
i didnt mean to send it intoanythng special, and was actualy trying to use a voice chat when it fell asleep, lol
<sophiag>
ah...standby == sleep
mguentner2 has joined #nixos
mguentner has quit [(Ping timeout: 240 seconds)]
alx741 has quit [(Quit: alx741)]
esmiurium has quit [(Ping timeout: 276 seconds)]
esmiurium has joined #nixos
Fare has quit [(Ping timeout: 245 seconds)]
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
jmeredith has quit [(Quit: Connection closed for inactivity)]
mk-fg has quit [(Ping timeout: 276 seconds)]
flemming has joined #nixos
mk-fg has joined #nixos
eacameron has quit [(Remote host closed the connection)]
eacameron has joined #nixos
systemfault has joined #nixos
eacameron has quit [(Ping timeout: 258 seconds)]
eacameron has joined #nixos
MercurialAlchemi has joined #nixos
viric has quit [(Ping timeout: 240 seconds)]
eacameron has quit [(Ping timeout: 276 seconds)]
eacameron has joined #nixos
ebzzry_ has quit [(Ping timeout: 264 seconds)]
eacameron has quit [(Ping timeout: 260 seconds)]
pi3r has joined #nixos
Itkovian has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
pi3r has quit [(Ping timeout: 256 seconds)]
flemming has quit [(Ping timeout: 245 seconds)]
viric has joined #nixos
<unlmtd[m]>
where should keep nixpkgs
reinzelmann has joined #nixos
pi3r has joined #nixos
jasom has joined #nixos
<jasom>
I'm having trouble finding an article that described how to sync the git version of a nixpkgs checkout so that you still got hydra builds to most packages while still being able to make local modifications...
pi3r has quit [(Read error: Connection reset by peer)]
<unlmtd[m]>
^^
<unlmtd[m]>
I think we need to write one, I need that too
<jasom>
I remember *reading* it but can't find it now
<jasom>
the next question is how hard is it to test 32-bit binaries on a 64-bit machine (I've only tested my expression with 64-bit native, and should probably test 32-bit before making a PR)
blahdodo has quit [(Remote host closed the connection)]
JagaJaga has joined #nixos
ryantm____ has quit [(Quit: Connection closed for inactivity)]
<unlmtd[m]>
I say f*** github, lets run nixos on gitea
thaega has joined #nixos
Itkovian has joined #nixos
athan has quit [(Ping timeout: 264 seconds)]
FRidh has joined #nixos
<musicmatze>
unlmtd[m]: +1 :-P
ekleog has quit [(Ping timeout: 255 seconds)]
rly has joined #nixos
ekleog has joined #nixos
<philipp[m]>
No, let's go gitlab and solve the problem of the repository size by losing data.
<philipp[m]>
Also open issues should vanish from time to time automatically when somebody deletes the db again.
Sibi__ has quit [(Quit: Connection closed for inactivity)]
sophiag has joined #nixos
rmeis has joined #nixos
<musicmatze>
philipp[m]: given the 1,4k issues we have... that'd be an option! :-P
<NixOS_GitHub>
[nixpkgs] FRidh pushed 4 new commits to staging: https://git.io/vD23e
<NixOS_GitHub>
nixpkgs/staging 7004601 Frederik Rietdijk: guitarix: use python2
<NixOS_GitHub>
nixpkgs/staging 2832472 Frederik Rietdijk: golly: use python2
<NixOS_GitHub>
nixpkgs/staging 616f358 Frederik Rietdijk: hamlib: use python2
Mic92 has quit [(Quit: WeeChat 1.7)]
Mic92 has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
systemfault has quit [(Quit: Bye!)]
derjohn_mob has quit [(Ping timeout: 256 seconds)]
<sophiag>
can anyone tell me what option under fileSystems i can use to remount a read only one (mounted from disk image) as writable? i see we're not supposed to edit fstab directly
<sophiag>
its from OS X, which i suspect is why i can't just remount it
georges-duperon has quit [(Ping timeout: 252 seconds)]
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to staging: https://git.io/vD23D
<NixOS_GitHub>
nixpkgs/staging bfb31d2 Frederik Rietdijk: ranger: fix tests
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python3 from e98efc2 to 9fbe535: https://git.io/vPPX6
<sophiag>
and perhaps i should add i don't intend to keep it in configurations.nix as a fileSystem, but really just want to copy it to a regular directory (cp -a throws an error part of the way through)
<ronny>
sophiag: on the rare occcasion a nixos-rebuild switch will not be enough you can reboot
indi_ has joined #nixos
IITaudio has joined #nixos
<ronny>
also if you plan larger changes/dm changes, its useufll to do a rebuild for boot, and then boot at your own choice
<IITaudio>
Hi, when will be released the next stable version of nixos? march/april?
<sophiag>
yeah, it's still saying read-only
<sophiag>
to be clear, though, i don't want to keep that setting in configuration.nix...just copy it to a regular directory. it's salvaged from my dead osx box
Itkovian has joined #nixos
sheenobu_ has quit [(Remote host closed the connection)]
indi_ has quit [(Remote host closed the connection)]
civodul has joined #nixos
sophiag has joined #nixos
indi_ has joined #nixos
<sophiag>
ok...so i tried rebooting as suggested and had to rollback updates for my first time using nixos :p
<sophiag>
given the problematic updates i don't think i need to pull up the log. it was clearly specifying that fileSystem i wanted to remount as rw
<sophiag>
i wonder if possibly because i didn't specify the boot volume as well and that overrides the default?
derjohn_mob has joined #nixos
<sophiag>
the whole volume i copied the disk image onto appears empty as well...that took around 6 hours to complete seeing as it was from usb...how is it possible that rolling back builds would delete files?
mudri has joined #nixos
<sophiag>
ok, it's late and i'm thinking clearly. just don't have my secondary internal drive set to mount at boot
jensens has joined #nixos
<sophiag>
now i have to remount the img anyway so i wonder if setting it as rw from the start will save me having to edit fileSystems in my configuration
indi_ has quit [(Remote host closed the connection)]
ixxie has quit [(Remote host closed the connection)]
indi_ has joined #nixos
<sophiag>
i can't find this in the man page, can anyone explain what mount -oloop does as opposed to -o?
indi_ has quit [(Remote host closed the connection)]
indi_ has joined #nixos
<sophiag>
oh, actually from reading a bit about loopback mounting that seems like it could be the entire reason rw isn't working...
<sophiag>
(talking to myself now)
indi_ has quit [(Remote host closed the connection)]
georges-duperon has joined #nixos
<sophiag>
wrong...in order to mount it normally i need to specify it as a fileSystem in configuration
<sophiag>
so presumably if i do that and also specify the boot volume it'll load and i'll be able to mount it as rw
<FRidh>
nix-build returns with exit status 1 but there's no error or whatever reported, nice.
sophiag has joined #nixos
<sophiag>
ok, so i clearly need some help at this point... this is what i added for fileSystems (both for the purposes of mounting my secondary drive "/dev/sda" at boot and allowing me to mount an OSX disk image as read-write): http://pastebin.com/VYbK0ATY
<sophiag>
the error on rebuild is: "Failed assertions: - The ‘fileSystems’ option can't be topologically sorted: mountpoint dependency path / -> / loops to /"
mojjo has joined #nixos
jgertm has quit [(Ping timeout: 264 seconds)]
<mojjo>
hi! is there anyon experienced with using stack under nixos? I'm having trouble with using the haskell package gloss, which needs some C-lib dependencies...
<hendrik[m]>
sophiag: i think fileSystems should be a set instead of an array
<hendrik[m]>
"array" -> "list" I mean
<domenkozar>
mojjo: you probably need to override the package to provide a system dep
mudri has quit [(Ping timeout: 252 seconds)]
<sophiag>
hendrik[m]: isn't that how i have it already? square brackets?
<goibhniu>
sophiag, also, should the first one be a partition e.g. /dev/sda1?
<mojjo>
oh, I did not reveal this: user error (unknown GLUT entry glutInit)
<domenkozar>
mojjo: instead of mesa try mesa_glu
<sophiag>
hendrik[m]: oh ok, lemme give it a try
<sophiag>
maybe then i can even just deal with the disk image situation without rebooting
rmeis has joined #nixos
<mojjo>
domenkozar: I did so, still the same. The only way I get it working is: nix-shell --pure -p "haskellPackages.ghcWithPackages (p: [p.gloss])"
<mojjo>
then inside ghci I load the code, and something is drawn to the window as expected.
mudri has joined #nixos
<sophiag>
well, the rebuild was successful but it still says read-only filesystem so i guess i need to try rebooting. the upshot is i took out the boot and swap volumes since they were conflicting with hardware-configuration.nix so at least it shouldn't go into safety mode this time
iMatejC has quit [(Remote host closed the connection)]
<sophiag>
ok so this is really killing me...i now have both my secondary drive and the disk image booting at startup...still running "mount -o remount,re MacintoshHD.hfsx MacintoshHD" does NOT change it from being read-only :(
<goibhniu>
can you mount it rw manually?
<sophiag>
plus in the course of having to reboot i tested my issues with suspend/freeze/hibernate...none of which can it come out of
<sophiag>
goibhniu: isn't that what i'm doing with that command?
<sophiag>
the only reason i added it to configuration was because that wasn't working when it was loopback mounted
<goibhniu>
gah, sorry ... I thought you were referring to how it gets mounted on boot
georges-duperon has quit [(Ping timeout: 252 seconds)]
<sophiag>
oh shit
<sophiag>
i did NOT do that
<sophiag>
although it's likely not to work if it doesn't on remount, right?
<sophiag>
lol, yeah i was told basically i may be screwed with my laptop
<sophiag>
it's horrible if you think about it tho...like i have to power down every time i want to bring it somewhere or the batter will die?
<goibhniu>
I've had success with TuxOnIce on a machine that I couldn't otherwise get to work before ... not sure if that's an option any more
<sophiag>
someone on here was kind enough to give me a config that fixes it for lenovo (mine is asis), but that's assuming you can come out of freeze...i can't
<goibhniu>
aye, not having some form of suspend is unbearable
eacameron has joined #nixos
<sophiag>
and then i was told you really don't want to do it ad hoc anyway since it can screw up hardware driver data
mojjo has joined #nixos
<sophiag>
i basically need to ask on a mailing list or something to find other nixos users with similar asus laptops
<goibhniu>
... any other linux users would do too
<goibhniu>
hopefully there's some magical kernel parameter that does the trick
<sophiag>
true
<sophiag>
actually i should ask on one of the stack exchange sites
<sophiag>
thanks for the idea
RchrdB has quit [(Ping timeout: 260 seconds)]
<sophiag>
but anyway, as for what i'm trying to do now...do you think adding "options = [ "rw" ];" would even do anything?
<dweller>
hmm, did anyone had problems with building gnutls 3.4.17?
<goibhniu>
cool, good luck!
jgertm has quit [(Quit: WeeChat 1.6)]
<goibhniu>
well, I'd play with umounting and then mounting with rw options first, before worrying about the nixos config
eacameron has quit [(Ping timeout: 260 seconds)]
<goibhniu>
FWIW the last time I tried mounting an hfs+ drive (3 years ago?) I couldn't get it to work
<goibhniu>
I could create an hfs+ partition, and mount that rw ... but I couldn't use one created on a mac ... but that's ancient history, and anecdotal
jgertm has joined #nixos
rmeis has quit [(Ping timeout: 264 seconds)]
eacameron has joined #nixos
<sophiag>
i was concerned it was corrupted given the source, but the actual problem was it wasn't actually a dmg. i forgot the format, but 7z automatically detected it and switched over whereas dmg2img failed
<sophiag>
and i checked everything is there too
<sophiag>
i tried cp -a to a new dir but it failed maybe 10% of the way through
<sophiag>
i might just try cp -a on the user directory and see if that works
<sophiag>
regardless i guess it's safe to shred the backup drive i was using
<sophiag>
anyway, maybe you're onto something in that i didn't try unmounting it before remounting
<sophiag>
i should be able to then mount not as loopback since it's in configurations
<goibhniu>
rysnc might be a nice option, if it's a bit flaky ... it can resume and also do a checksum on the contents
eacameron has quit [(Ping timeout: 240 seconds)]
<sophiag>
ok, so "umoutn MacintoshHD" and then "mount -o,rw MacintoshHD.hfsx MacintoshHD" still doesn't make it read-write
<sophiag>
what about rsync now? pretty sure it's not a corruption issue if that's what you're thinking
<goibhniu>
well, if it keeps failing when you use cp, using rsync instead of cp might help ... but yeah, I was thinking it might be corrupt
<sophiag>
oh i see
<sophiag>
or fsck?
<goibhniu>
I'd prefer to use the mac tools for that, unless you have it all safely backed up
<sophiag>
i mean, this whole mounting and umounting business seems a huge waste just to copy it to a directory
<sophiag>
can you advise on how to use rsync?
<goibhniu>
why do you need it rw if you're just copying data from it?
<sophiag>
which means you're probably right, yeah?
<sophiag>
minor enough corruption is would unzip and mount, but some files deep in there are
<goibhniu>
I think rsync with -car would do the trick
<IITaudio>
goibhniu: do you have suspend issues too? My nixos on 2 different machines wont go to sleep half of the times. It just freezes and wakes up after one minute
wizeman has joined #nixos
<goibhniu>
IITaudio: nah, not any more
<sophiag>
IlTaudio: that was me, except the issue (apparently quite common) is it not coming *out* of suspend
<sophiag>
i'm told it's hardware specific
<IITaudio>
sophiag: the same (not waking up) happened on my asus (i7 4400U), got almost resolved by a kernel upgrade I think
<jophish>
morning all
* goibhniu
decided to get a laptop with official Linux support this time to avoid all that stuff: http://www.tuxedocomputers.com
<goibhniu>
sophiag: it would totally be worth trying the latest kernel available, and even nixos-unstable ... you can always roll back
<goibhniu>
I think nixos-unstable has a much more recent nvidia module
<MichaelRaskin>
For some cards nouveau is more or less fine. On the other hand, on my ThinkPad I had Optimus configured in some way that worked reliably until a kernel upgrade, but now I cannot fully switch off the nvidia card after using it without a reboot.
<MichaelRaskin>
(The laptop LCD is controlled by an Intel card, so if I reboot and don't enable nvidia I can get battery life as expected from the really impressive optional battery I have)
<sophiag>
does nixos-unstable contain the latest kernel?
<goibhniu>
ouch
Hunner has joined #nixos
Hunner has quit [(Changing host)]
Hunner has joined #nixos
<goibhniu>
sophiag: not by default, you'd still need to select it
<sophiag>
ah ok
<goibhniu>
that's a great idea though, to try nouveau or just intel
<NixOS_GitHub>
[nixpkgs] hrdinka opened pull request #22621: cdi2iso: init at 0.1 (master...add/cdi2iso) https://git.io/vD2gU
<IITaudio>
sophiag: That clip made the history :)
<sophiag>
well, i only hope it actually changed things!
bennofs has quit [(Ping timeout: 258 seconds)]
<NixOS_GitHub>
[nixpkgs] hrdinka opened pull request #22622: attract-mode: init at 2.2.0 (master...add/attract-mode) https://git.io/vD22k
<sophiag>
so before posting to unix stack exchange i found some similar issues, even on asus, but all slightly different and none exactly the same. however, the common thread between all is, as you've all mentioned, they were solved by kernel updates (in one case a custom kernel, but that was ubuntu and i'd rather not think about the prospect)
<sophiag>
so given that, i'm wondering how i check what my kernel version is?
<goibhniu>
sophiag: `uname -a` will tell you
<goibhniu>
sounds promising :D
<sophiag>
well, it's far from universal. i have 4.4.45 and another Asus user said 4.4.8 didn't make a difference
<sophiag>
it seems 4.9.9 is the latest stable version
<IITaudio>
sophiag: for me the change from "suspension does not work 90% of the times" and "sometimes it doesn't wake up" was like from 4.2 to 4.4
<sophiag>
llTaudio: good to know
<sophiag>
(sorry, keep typing your nick wrong)
<sophiag>
can i upgrade to 4.9.9 while staying on the stable branch of nixos? i'm hesitant to switch to unstable given i have literally one week of experience and the learning curve was quite steep at first
<MichaelRaskin>
What do you all people do to make suspend not work on an Asus, I wonder…
<IITaudio>
MichaelRaskin: for me it was just: install ubuntu. Suspension does not work. Upgrade kernel. Kernel panic. Switch to nixos. does not work. Upgrade kernel. Does not work. Repeatrd until kernel upgrade fixed
<sophiag>
goibhniu: thanks. i'll put that in my queue of bullshit :p
jensens has quit [(Ping timeout: 240 seconds)]
<MichaelRaskin>
Hm. Is it about suspend-to-disk or suspend-to-RAM or suspend-to-both?
<sophiag>
everything :(
<MichaelRaskin>
How do you initiate the suspend, though?
<MichaelRaskin>
echo mem >/sys/power/state ?
<sophiag>
goibhniu: ugh, speaking of bullshit....rsync to a long time to run and i still get an error trying to cp the same file as before
<sophiag>
MichaelRaskin: i've tried suspend and hibernate from gnome and freeze using that method
<sophiag>
goibhniu: do you know when cp -a halts on one file if i can get it to just skip it and copy the rest instead of killing the process? that might be sufficient
bennofs has joined #nixos
<sophiag>
oh wait...maybe the answer was simply "patience" lol
<goibhniu>
sophiag: I bet you can get it to ignore that one file, it's handy that it will resume too
<sophiag>
(it's not "good morning for me" should be "good night" hence the attitude)
<MichaelRaskin>
Well, the old standard troubleshooting guide recommended trying to login via a non-X console and run the suspend initiation command from there
<MichaelRaskin>
I remember at some time my suspend script was switching to tty1 and switching back after resume
<MichaelRaskin>
But it has been a long time since such tricks were actually needed on any of my laptops
<sophiag>
MichaelRaskin: oh thanks for the tip
<MichaelRaskin>
Also
<MichaelRaskin>
Hm, what else do I see in my scripts' comments…
<MichaelRaskin>
unmounting stuff (removable devices and FUSE)
<MichaelRaskin>
Switching off WiFi (I think there was a device where switching WiFi off and removing the driver module work)
<MichaelRaskin>
Stopping everything that could do sound output
<sophiag>
that's...a little much for me. although i do appreciate it
<sophiag>
i'll certainly test on tty as well as trying the kernel upgrade tho
<MichaelRaskin>
I mean, that's the union of _all_ workarounds I have ever tried
<sophiag>
ha. more making the point of how much of an ass pain it is?
<MichaelRaskin>
More like how much tricks you accumulate over five devices and twelve years
<sophiag>
almost makes more sense to have the whole thing continuously back up to dropbox where all your major data is stored and ditch the cloud entirely
m0rphism1 has quit [(Quit: WeeChat 1.4)]
<MichaelRaskin>
It is like the first search result
<sophiag>
my hangup about cloud pricing is it doesn't include a huge amount of storage so i'd really not dropbox (or whatever replaces it if they go bankrupt) as well and then it's just a mess
<MichaelRaskin>
And yes, it's a fit drive, i.e. I do throw laptops with such USB drives inserted into bags without second thoughts
m0rphism has joined #nixos
rmeis has joined #nixos
<sophiag>
goibhniu: looks like i'm good with cp -a! could have save a lot of time mucking around with fileSystem :p
<sophiag>
just testing to feel comfortable shredding the shitty sata backup
<goibhniu>
\o/
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<LnL>
clever: did you look at setuid for the systemd runner script yet? otherwise I will
<mindB>
Okay. Still working on weechat. To me it looks as if the weechat default.nix has a special postInstall variable that makes weechat aware of any python package included as a extraBuildInput. So if I've got a lua plugin that has external dependencies, I need to do the same thing for lua.
ZoomZoomZoom has quit [(Read error: Connection reset by peer)]
Zoom__ has joined #nixos
<clever>
LnL: not yet
mizu_no_oto has joined #nixos
<LnL>
allright, I've been busy with other stuff so I didn't look at it any further either
<clever>
LnL: same, and i need to start keeping a todo list, so i cant forget stuff like this
<LnL>
I also want to look at the user definitions, would be pretty cool if I could get passwd and such from the modules
msd has joined #nixos
<msd>
Hey, anyone here using Riak?
<msd>
**the Riak package?
<msd>
And getting odd NixOS-stemming errors concerning the filesystem?
iyzsong has quit [(Quit: bye)]
<LnL>
I played with it once, but don't actively use it
<gchristensen>
nixos-stemming?
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
mudri has quit [(Ping timeout: 245 seconds)]
<Unode>
clever: have you recently used nix in a non-standard location? I'm getting some strange systemd/stdenv errors when trying to install vmTools: http://dpaste.com/1GH37CS
Shou has joined #nixos
<clever>
Unode: havent tried doing that, i always have root, so i just set it up right
<Baughn>
(Also KDE5 doesn't start. No, I don't know why.)
Hunner has joined #nixos
Hunner has quit [(Changing host)]
Hunner has joined #nixos
<Baughn>
((Well, I suspect an interaction with g-sync.))
mizu_no_oto has joined #nixos
<gchristensen>
kde4 hardly even gets _critical_ security patches these days, and was last released over 2 years ago
defel has joined #nixos
* Baughn
shrugs. What security patches does a WM need?
<clever>
Baughn: what if the <title> in a webpage enters the title bar of a window, and causes a buffer overflow in the WM?
Fare has joined #nixos
<Baughn>
Okay, point. I think that's an unlikely attack, though.
<Baughn>
It remains true that KDE5 doesn't work. Maybe I'll try FVWM.
<MichaelRaskin>
Quite a likely attack; it is not that likely that a WM is not resistant to it
<clever>
Baughn: as unlikely as javascript being able to flip bits in the paging tables without having to use any buffer overflow bugs?
<clever>
(row hammer)
<Baughn>
I don't think any KDE patch is going to protect against a hardware bug..
<MichaelRaskin>
I would expect that title attack to need soem Unicode corner cases to crash the rendere
<MichaelRaskin>
renderer
<clever>
MichaelRaskin: yeah, but i have seen worse, windows handles font rendering in the kernel
reinzelmann has quit [(Read error: Connection reset by peer)]
reinzelmann has joined #nixos
<bennofs>
Where is the code to generate the .narinfos? Only in nix-push?
<clever>
bennofs: its also present in nix-serve
<bennofs>
and hydra as well...
<gchristensen>
fwiw KDE has _loads_ of vulnerabilities.
<gchristensen>
a title bar bug is just a trivial one
* Fare
just went back from KDE to stumpwm
cpennington has joined #nixos
<pie__>
0o
phreedom has joined #nixos
dramforever has joined #nixos
* pie__
feels the suddenurge to bail from kde
<gchristensen>
I mean, all software does
<pie__>
well ok
<pie__>
i guess
<gchristensen>
and kde happens to be fairly big and complicated
<LnL>
not sure how I feel about using an unmaintained display manager either
<pie__>
i still kinda miss the crunchbang setup, havent gotten around to trying to replicate it thogh
<pie__>
(openbox)
<jophish>
gchristensen: Did anyone ever get around to getting install metrics from the cache, it would be interesting (if nothing else) to know how many people are still using kde4
<gchristensen>
but please, do, upgrade from kdm to sddm and kde4 to kde5
<gchristensen>
jophish: I don't think we can practically do that :/
<jophish>
fwiw the de on the virtualbox images is still kde4
<clever>
bennofs: this is the command that hydra internaly uses to eval a channel, it will spit out json listing every job in the channel, and you can run it without configuring hydra
<clever>
bennofs: but beware, it will flood /nix/store with about 30,000 .drv files, and that has crippled my btrfs in the past
dramforever_ has joined #nixos
dramforever has quit [(Disconnected by services)]
* Baughn
once again suggests ZFS. :P
dramforever_ is now known as dramforever
<clever>
Baughn: i have since switched to zfs
<Baughn>
Ah.
<Baughn>
Shouldn't be a problem, then.
<bennofs>
clever: ah, cool. I'm just looking at generate-programs-index.cc, and it seems that it also builds a cachedb that has information about *all* paths, not just programs? Is there a way to get access to this db? :)
<gchristensen>
bennofs: it is probably at /nix/var/nix/profiles/per-user/root/channels/nixos/programs.sqlite
<bennofs>
gchristensen: programs.sqlite only contains executables
<Baughn>
Is that what provides the "Install this package" hint for 404s in bash?
<clever>
bennofs: so we need to find out what runs this
<gchristensen>
ah
<bennofs>
clever: mirror-nixos-branch runs it
<Unode>
guys, is running nix on a non-standard location with non-root access supported? I.e. should I file bug reports if something doesn't work in such setup?
<Unode>
Well my setup is "tricky" to say the least. /share/nix on NFS with inter-node functionality by having a nix-daemon on a single node and all other nodes having an socat piping to to the nix-daemon via SSH tunnel.
<manveru>
it's not valid json though :|
eacameron has joined #nixos
dramforever has joined #nixos
<Unode>
everything works... unless packages fail for some reason.
<bennofs>
clever: the idea is to build a big locatedb for all store paths in a evalution, to help with questions like "which package has this header/.so/..."
fritschy has quit [(Quit: Leaving.)]
msd_ has joined #nixos
msd has quit [(Read error: Connection reset by peer)]
<clever>
bennofs: yeah, that does look like it would be of use
<manveru>
bennofs: thanks
msd_ has quit [(Remote host closed the connection)]
<Unode>
is it worth reporting problems that affect the stable channel but not the unstable one?
<gchristensen>
yeah
dramforever has joined #nixos
phreedom has quit [(Quit: No Ping reply in 180 seconds.)]
uralbash has quit [(Quit: Leaving.)]
jgertm has quit [(Quit: WeeChat 1.6)]
phreedom has joined #nixos
jgertm has joined #nixos
<rly>
Is there any tool which can do directory diffs on NixOS that works? The ones I usually use do not on NixOS.
<rly>
I don't really get why people package things in a broken state in the first place.
<ToxicFrog>
rly: "diff -r"? "git diff"?
<rly>
ToxicFrog: I am thinking of something like meld.
<ToxicFrog>
Oh. No idea.
<rly>
ToxicFrog: (which is broken)
<rly>
ToxicFrog: it's mostly a merge tool.
<ToxicFrog>
(presumably they worked when packaged and dependency updates have since broken them; nix doesn't formally do versioned packages so this happens with noticeable frequency)
<MichaelRaskin>
What's broken about meld?
msd has joined #nixos
<MichaelRaskin>
I mean, I have just launched it from NixPkgs master and it seems to do directory comparisons fine
<rly>
MichaelRaskin: try to modify your configuration.
Fare has quit [(Ping timeout: 258 seconds)]
<rly>
MichaelRaskin: dconf doesn't work.
<rly>
MichaelRaskin: so, e.g. try to ignore some files like foobar.
dramforever has quit [(Ping timeout: 240 seconds)]
rmeis_ has joined #nixos
<rly>
ToxicFrog: no, I doubt it has ever worked.
<rly>
ToxicFrog: it might have worked in one particular environment, but that's it.
<MichaelRaskin>
Oh, that's more than I have tried, that's true
<MichaelRaskin>
Yep, it is not that hard to ensure you have DConf accessible to your DBus session
<MichaelRaskin>
So maybe some people randomly have it and then DConf worcs
<MichaelRaskin>
works
<rly>
MichaelRaskin: packaging becomes worthless, if it's not reliable.
<rly>
"It works in my environment" is exactly what Nix was supposed to solve.
<rly>
Specifically on NixOS this holds even stronger.
rmeis has quit [(Ping timeout: 258 seconds)]
<gchristensen>
well
<rly>
"Ship it, it compiles!" <- is what these packages communicate.
<gchristensen>
let's all go back to debian, then
<MichaelRaskin>
Well, I don't run NixOS mainline anyway and specifically separate DBus sessions when I do care about them
<clever>
rly: part of it, is the difference between running foo in a script (and assuming the user has it in $PATH), vs running ${pkgs.foo}/bin/foo and always working
<rly>
clever: every occurence of that is a bug.
<rly>
clever: if a package doesn't handle all those cases, it should not be considered "packaged".
<clever>
rly: some packages also use propagated-user-env-packages, which basicaly forces nix-env to install X automaticaly when Y gets installed
<gchristensen>
I'd love to see your PRs, rly, but until now I'm finding you to be abusive and would like you to stop.
<clever>
rly: but i avoid that, because different X's can now conflict, and Y cant be tested via ./result/bin/y
<rly>
gchristensen: abusive? I am just saying that the process known as NixOS packaging is broken.
<rly>
gchristensen: do you prefer alternative facts instead?
<MichaelRaskin>
There is no package source that has no bugs
<gchristensen>
rly: I have asked you to stp.
<rly>
gchristensen: I am also asking you to stop.
mkoenig has joined #nixos
<rly>
gchristensen: sad.
rly has left #nixos [requested by gchristensen (rly)]
rly has joined #nixos
<rly>
I think it's extremely sad to abuse power in this way.
<rly>
Sad, sad, sad.
<MichaelRaskin>
I agree that it's a bug, but you make a judgement on relative importance of different issues
<gchristensen>
if you find issues in Nix/OS, we would welcome your contributions or productive and constructive feedback in the bug tracker.
<rly>
What I am saying is *true* and you hope that the truth goes away when you kick someone.
<rly>
It's just sad and you just lost a massive amount of respect.
<Baughn>
Okay, rly is being deliberately disruptive. Why don't you just ban him?
<Baughn>
(And no, no he didn't.)
eacameron has joined #nixos
rly has left #nixos [requested by gchristensen (rly)]
<MichaelRaskin>
I wonder how often that latter kind of ban should be cleaned up.
<MichaelRaskin>
Because I guess one-year old IP bans for residential IPs, even if nominally static, are out of date
<gchristensen>
probably so
* gchristensen
looks up ban formats
<Baughn>
There's no way to auto-expire them.. apparently.
<clever>
Baughn: thats usualy a feature of chanserv on most networks
<MichaelRaskin>
Is there a way to look up _old_ bans, while we are discussing this anyway?
<Baughn>
clever: That's what I thought, but..
<gchristensen>
MichaelRaskin: yeah
<clever>
MichaelRaskin: old, or current?
<clever>
MichaelRaskin: "/mode #nixos +b" lists the current ones
<Baughn>
I found a nine-year-old blog stating that it isn't a feature on freenode (and to track your bans), and nothing newer to obsolete it.
eacameron has quit [(Ping timeout: 240 seconds)]
dramforever has joined #nixos
mudri has joined #nixos
panostimos has quit [(Ping timeout: 255 seconds)]
<NixOS_GitHub>
[nixpkgs] pSub pushed 2 new commits to master: https://git.io/vD2Su
<NixOS_GitHub>
nixpkgs/master de9720b Christoph Hrdinka: aqemu: init at 0.9.2
msd has quit [(Remote host closed the connection)]
msd_ has joined #nixos
rmeis_ has quit [(Ping timeout: 240 seconds)]
<fre>
Hello. How does NixOS manage /etc/alternatives?
Hunner has quit [(Ping timeout: 255 seconds)]
<Baughn>
fre: It doesn't. No such thing exists.
<fre>
ugh you're absolutely right
<Baughn>
To answer a little more generally...
<fre>
I confused machines my bad. ;-)
<fre>
but I'm interested.
<Baughn>
/etc/alternatives is for distributions where you can't, for each package / user context, specify what packages should be available.
<fre>
a sensible default
<Baughn>
It's one way to manage name clashes in a global environment.
<Baughn>
With NixOS, we prefer to eliminate the global environment.
msd has joined #nixos
<fre>
I see. thanks!
<Baughn>
I.e: If there are two packages providing the same binary, then install *neither*, and specify dependencies when needed using nix-shell or similar mechanisms.
<gchristensen>
ack, domenkozar, I've been trying to wait until you're the guinea pig on the XPS 15 but my computer might not make it! :P
<Baughn>
...though you can also just tell Nix to prioritize one package over the other. The mechanism does exist, it just isn't preferred.)
<Baughn>
fre: But, for a simple example...
<fre>
is there a menagerie of nix expressions for various languages available?
<Baughn>
Yes.
<domenkozar>
gchristensen: it will take another 1-2 weeks they told me today
dramforever has quit [(Ping timeout: 240 seconds)]
rmeis has joined #nixos
<fre>
Yeah I do similar things Baughn , I'm wondering about a larger collection of such things
<NixOS_GitHub>
[nixpkgs] hrdinka pushed 1 new commit to master: https://git.io/vD2HV
<NixOS_GitHub>
nixpkgs/master 3047bb2 Christoph Hrdinka: nsd: 4.1.13 -> 4.1.14...
MercurialAlchemi has quit [(Ping timeout: 240 seconds)]
<NixOS_GitHub>
[nixpkgs] hrdinka pushed 1 new commit to release-16.09: https://git.io/vD2HM
<NixOS_GitHub>
nixpkgs/release-16.09 498b5c8 Christoph Hrdinka: nsd: 4.1.13 -> 4.1.14...
<Baughn>
fre: Have you looked at nixpkgs?
ebzzry has joined #nixos
<fre>
hihi I understand that I can use any package's build environment, but I'd like to know if there are some full fledged dev environments with linters, static analyzers, editors ect
<Baughn>
Ah, like so... I'm not aware of any such repository. Would be nice to have.
msd has quit [()]
<fre>
yes :_)
danharaj has quit [(Ping timeout: 252 seconds)]
<fre>
on another note, does this channel provide IRC logs?
mkoenig has quit [(Remote host closed the connection)]
dramforever has joined #nixos
<gchristensen>
banning someone always makes the day seem a little crappier.
<goibhniu>
FWIW you did the right thing IMO ... the last time rly returned after being kicked or banned for a bit, they were surprisingly well behaved and helpful
<goibhniu>
it's just stressful for everyone to sit here when someone is ranting abusively
<goibhniu>
so, thank you!
<gchristensen>
you're welcome :P heh, I remember that last time. I won't tolerate people being abusive or needlessly critical. it still sucks, though :)
joshie has quit [(Ping timeout: 245 seconds)]
dramforever has quit [(Ping timeout: 240 seconds)]
katyucha has quit [(Quit: Lost terminal)]
mkoenig has joined #nixos
katyucha has joined #nixos
panostimos has joined #nixos
mkoenig has quit [(Ping timeout: 260 seconds)]
mkoenig_ has joined #nixos
alx741 has joined #nixos
ebzzry has quit [(Ping timeout: 245 seconds)]
tvon has quit [(Quit: System is sleeping...)]
civodul has joined #nixos
phreedom has quit [(Ping timeout: 255 seconds)]
mkoenig_ has quit [(Read error: Connection reset by peer)]
<gchristensen>
copumpkin: one thing the github wiki might do is provide spam protection beyond just auth
<jophish>
Can the github wiki be made pull request based? IMHO I think that's a much better way to curate that kind of content than people editing willy-nilly
phreedom has joined #nixos
<gchristensen>
you should comment on the issue probably :)
erasmas has joined #nixos
<gchristensen>
("well right back atchya, gchristensen")
<Biappi>
copumpkin: " I get that misleading documentation is bad, but there are pros and cons to both sides." i stopped trying to use nixos because of bad documentation, the con being "i'm not sure what's supported or not". the manual, albeit confusing in its own rights, proviedes a "seal of approval" kind of feeling, that this is the way this system should be used
dbmikus has joined #nixos
mudri has quit [(Ping timeout: 258 seconds)]
tvon has joined #nixos
Rotaerk has quit [(Read error: Connection reset by peer)]
<pie__>
docstrings might be nice
<pie__>
do we do that?
<gchristensen>
for options? or
t7 has quit [(Remote host closed the connection)]
<pie__>
for stuff in nixpkgs
t7 has joined #nixos
<pie__>
and stuff in the languge
<pie__>
or runtime or whatever it is that provides stuff like overrides
<gchristensen>
codedmart: well (1) you need to probably pass an argument telling it to write elsewhere, (2) it is likely reading a default config telling it to run as the user arangodb
<gchristensen>
which you likely need to override via an option as well
<codedmart>
gchristensen: Would those be options that arango provides? Or standard options?
<gchristensen>
those would be arango-provided options
<codedmart>
OK I will look into it. Thanks!
<jophish>
gchristensen: commented
<gchristensen>
thank you!
<gchristensen>
so cool, seeing like 5 comments in 10 minutes
<gchristensen>
!m copumpkin for having the gall to bring this up :P
<[0__0]>
You're doing good work, copumpkin for having the gall to bring this up :P!
drakonis has joined #nixos
<FRidh>
guess its been at least a month so it was about time to have it brought up again :)
<gchristensen>
FRidh: usually only grumblings
<gchristensen>
not an actual issue
eacameron has joined #nixos
davidak has joined #nixos
tvon has quit [(Quit: System is sleeping...)]
mkoenig has joined #nixos
<goibhniu>
so exciting!
<jophish>
I wonder if there's a "best of both worlds" solution, where PR based editing and wiki based editing are supported
tvon_ has joined #nixos
<FRidh>
jophish: you can use the edit button on github, and it will automatically create a PR for you
<gchristensen>
garbas: ping
ryanartecona has joined #nixos
<jophish>
FRidh: personally I think just PRs would be great, I don't know if everyone else feels the same way, a mix might be the best way to get a PR supporting wiki in place
<jophish>
I'm expipiplus1 on github btw
<garbas>
gchristensen: pong
<gchristensen>
garbas: a couple things: it seems mozilla's nixpkgs isn't updating automatically? second: how do you recommend people actually use the mozilla nixpkgs? (I'm replicating the idea over here at work.)
<garbas>
gchristensen: yeah, on my todo list. waiting to be hooked with our ci system (taskcluster).
<gchristensen>
ah ok
<garbas>
gchristensen: to many things todo so i have to pick "battles" that make the most impact
<gchristensen>
I know what you mean :)
derjohn_mob has quit [(Ping timeout: 240 seconds)]
<jophish>
I wonder if nixpkgs@home could ever work securely and efficiently. Help take the load off hydra
<gchristensen>
I've recently changed my opinion of "Hydra has limited capacity" to "hydra is doing fine"
<goibhniu>
jophish: I think we need bit perfect deterministic
drakonis has quit [(Ping timeout: 240 seconds)]
<goibhniu>
... builds for that
<jophish>
well, there are lots of things that can be done with a greater capacity hydra, gchristensen
<eacameron>
My simple PHP app on NixOS can't send mail (it's wordpress)
<jophish>
ghcjs packages, building more versions of haskell packages
<eacameron>
Do I need to install something?
<gchristensen>
gotcha, jophish
<gchristensen>
garbas: do you have people put nixpkgs-mozilla in to their NIX_PATH and import <nixpkgs-mozilla>
phreedom has quit [(Quit: No Ping reply in 180 seconds.)]
<garbas>
gchristensen: personally i put it packageOverrides and maintain fetching nixpkgs-mozilla with fetchFromGitHub
rmeis has joined #nixos
phreedom has joined #nixos
<garbas>
gchristensen: but you can also give advice to just point to the latest NIX_PATH=nixpkgs-mozilla=https://github.com/.../master.tar.gz
bennofs1 has joined #nixos
<gchristensen>
cool, thank you
bennofs has quit [(Ping timeout: 240 seconds)]
<pie__>
def need deterministic builds for hydra"home
<gchristensen>
and then you need to identify who to trust
<MichaelRaskin>
More precisely, whom to trust not to collude
<MichaelRaskin>
The question is, what is the correct hash of the compiled LibreOffice
<jophish>
I know that I'd be happy to lend the cpu-cycles in our office to building nixpkgs when nobody is in here, perhaps there are enough people in similar situations around the globe to get a good coverage
<ekleog>
but compilation is the bigger issue
<pie__>
i need to look into ipfs, it keeps coming up
<MichaelRaskin>
Just because everyone is assumed to remember about Bittorrent and Tahoe-LAFS
takle has quit [(Remote host closed the connection)]
<jophish>
It's a shame that compilation with nixpkgs isn't some nice NP problem which can be verified easily :(
Jackneill has quit [(Read error: Connection reset by peer)]
<ekleog>
with hydra starting to check reproducibility of builds one could envision tagging reproducible builds as such, and dispatching them to nixpkgs@home
<pie__>
ekleog, but what good would that do?
reinzelmann has quit [(Quit: Leaving)]
<ekleog>
well, that still questions trust
<ekleog>
like, how to know we're not all some agency colluding to get our backdoor into nixos
<pie__>
on the other hand, it wouldnt decrease the amount of work, but it could make it easier for people to verify builds
<gchristensen>
that is a good thing
<ekleog>
indeed
<pie__>
but then agian cant you already do that with the hashes?
<gchristensen>
you can nix-build --check, yes
<ekleog>
well, you can for reproducible builds, which isn't the case of all builds
<pie__>
right
<pie__>
but repro builds are orthogonal but necessary for this
<pie__>
also i think i read hashes are sha256 truncated, why truncate?
<gchristensen>
they're not truncated
<pie__>
ok
<pie__>
nevermind then
<gchristensen>
they're just encoded differently
<gchristensen>
(is why they're shorter)
<ekleog>
anyway, right now the easiest way for some agency to get things into nixpkgs is most likely the github https cert during channel updates, as it's checked through the usual CA model
<eacameron>
In configuring nginx I often see a setting for dhparams but I don't know how to set that in NixOS? Anyone know how?
<gchristensen>
ekleog: or pay a single employee to contribute security updates on a weekly basis for a few months and become a trusted entity, all for just the price of one employee, and they don't even have to break crypto.
<eacameron>
ekleog: Ah right. But how does one build this dhparams.pem file?
<pie__>
we already have systemd
<ekleog>
I wasn't thinking about breaking crypto, a number of govs are tightly bound to CAs
<pie__>
what moer do you need
<ekleog>
but it's another nice attack
<ekleog>
eacameron: just like in any other distro I guess, I don't know of any nixos-specific tool ;)
<pie__>
also reproducible buils means reproducible exploits ;P
<pie__>
death by homogeniety \o/ ":D
<MichaelRaskin>
gchristensen: it's better to have three employees, so that we can have a nice cover-up discussion of unrelated parts of the commit that introduces a backdoor, and nobody suspects that went in unreviewed
Jackneill has joined #nixos
<gchristensen>
MichaelRaskin: they'd never suspect us! >.< :D
<pie__>
and those that did would be crackpots! :D
<ekleog>
pie__: well, anyway as soon as the build is on hydra exploits are reproducible, you'd need to build packages yourself for non-exploit-reproducibility
<MichaelRaskin>
No, no, worse than crackpots. People criticizing the KDE packaging in NixPkgs without contributing to it.
* gchristensen
isn't a nefarious actor, just to be clear :)
<pie__>
* gchristensen isn't a nefarious actor, just to be clear :)
<eacameron>
ekleog: Ah ak. I guess I thought there was more magic behind it. So for anyone using something like nixops to deploy lots of servers, it seems way too manually to generate a dhparams.pem file by hand for every server. Is there a better way?
<pie__>
the most useless line said by man
* gchristensen
shrugs
<pie__>
:P
<gchristensen>
I didn't want it to go unsaid :)
<pie__>
of course of course
<pie__>
i concede
<MichaelRaskin>
gchristensen: _now_ I start being suspicious of randomly guessing something…
<ekleog>
eacameron: well, you could generate a dhparams by adding the relevant command in system.activationScript ;) (and could even contribute that back to nixos as a module, I guess)
<gchristensen>
eacameron: hmmm! it might be good to make the nginx module do it
<ekleog>
activationScripts*
<MichaelRaskin>
Remember this «No, of course no» line by Linus, said while _conspiciously_ nodding
<ekleog>
gchristensen: it's more generic than just nginx, openvpn etc. also need it
<gchristensen>
yeah, but you probably want unique dhparams per
<ekleog>
having a separate module used as a backend by all these would be best, wouldn't it?
<ekleog>
like, [something].genDhParams = true, and boom, you have your local dhparams
takle has quit [(Remote host closed the connection)]
<gchristensen>
no, I think you want to generate unique dhparams per service
<ekleog>
h
<pie__>
ekleog, i cant remember what the question was
<pie__>
* MichaelRaskin
<ekleog>
hmm, afaik the only known attacks on dhparams are ones that are due to everyone having the same ones, but I guess it doesn't cost anything but first activation time
phreedom has quit [(Remote host closed the connection)]
<NixOS_GitHub>
[nixpkgs] FRidh pushed 2 new commits to master: https://git.io/vDakH
<NixOS_GitHub>
nixpkgs/master 742db3f Herwig Hochleitner: pypy: enable dbm support in db dependency...
<NixOS_GitHub>
nixpkgs/master a323b84 Frederik Rietdijk: Merge pull request #22611 from bendlas/fix-pypy...
<ekleog>
pie__: question was having dhparams generated by a nixos module, if that's the same I was talking about
<pie__>
i mean the linus one
<MichaelRaskin>
pie__: «Has NSA ever asked you explicitly to insert a vulnerability into Linux», more or less
<pie__>
ah
phreedom has joined #nixos
<ekleog>
oh ^^
<MichaelRaskin>
«Have any of you been approached by the US for a backdoor?» in a context of Linux development Q&A
<eacameron>
gchristensen: ekleog: I'll create an issue to track this. I'm not quite savvy enough with nixos modules to write it myself. But system.activationScript looks awesome.
<mrkgnao>
there is a shell.nix for use with Stack, but I don't exactly know where to go from there
<MichaelRaskin>
Ow, it doesn't just select a window, it gets the whole rectangle
<MichaelRaskin>
Nice
<manveru>
i use it all the time
<manveru>
but might be hard to know where you already selected
mkoenig has quit [(Ping timeout: 240 seconds)]
<manveru>
so cutting in gimp and exporting in xclip might be better
<manveru>
if you rename it to sha1sum of the out.png or something
<jophish>
There is something that nixpkgs@home could do which requires less trust than building the packages
<manveru>
then you could run it in a loop every second
<jophish>
running the checkPhase of packages is something which could usefully be done by untrusted users
<jophish>
there is a certain amount of trust involved, but the consequences of bad actors are reduced
takle has joined #nixos
<Dezgeg>
that won't really work since checkPhase runs in the same derivation as the other stuff
<jophish>
this assumes that the checkPhase can be split from the build/install phases, and doesn't require modifying $out
<Dezgeg>
but nix@home could be used to test pull requests or reproducible builds, for instance
<MichaelRaskin>
Which may be something to change
<Dezgeg>
I'm not sure it's feasible to split checkPhase - as it needs stuff from the build directory
<jophish>
It would be a substantial change, for sure
ertes has quit [(Ping timeout: 258 seconds)]
<NixOS_GitHub>
[nixpkgs] FRidh pushed 3 new commits to master: https://git.io/vDaYb
<NixOS_GitHub>
nixpkgs/master b2fb95a Evan Danaher: neovim-remote: move from python-packages to /neovim/neovim-remote.
<NixOS_GitHub>
nixpkgs/master ee4f732 Evan Danaher: neovim-remote: add meta section....
<NixOS_GitHub>
nixpkgs/master c9f4d71 Frederik Rietdijk: Merge pull request #22417 from edanaher/move-neovim-remote...
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
IITaudio has quit [(Quit: Leaving)]
mkoenig has joined #nixos
tvon_ has quit [(Quit: System is sleeping...)]
stepcut has quit [(Remote host closed the connection)]
<eacameron>
What do you all recommend for setting up sendmail?
<eacameron>
Nix-env tells me I can use busybox, exim, postfix, or ssmtp
<eacameron>
But I've not really used anything for SMTP on a server before.
<eacameron>
(I've always used REST based services)
<orbekk>
I use postfix on nixos
phreedom has quit [(Ping timeout: 240 seconds)]
<eacameron>
orbekk: Cool. One recommendation is good enough for me. :D
<orbekk>
services.postfix has everything I need for my mail server
tvon has joined #nixos
<MichaelRaskin>
Hm, the final closing brace is still attributed to the quick and small postfix job I have written in 2008.
<MichaelRaskin>
Not much else has survived for that long, though
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vDaZi
<NixOS_GitHub>
nixpkgs/master 66fe4af Vladimír Čunát: Merge recent staging...
<mbrock>
eacameron: I'm pretty happy with opensmtpd
<gchristensen>
MichaelRaskin: 9 years! nice!
<eacameron>
mbrock: Any reason I should use that over postfix?
civodul has quit [(Remote host closed the connection)]
<mbrock>
I'd say if you're a fan of OpenBSD-style simplicity (and I guess security track record), or something like that
<pie__>
i get the feeling obsd security tack record is somewhat moving goal posts
jgertm has quit [(Ping timeout: 252 seconds)]
<gchristensen>
oh?
<MichaelRaskin>
Well, after an exploit in OpenSSH they also disabled it y default so that the next one wouldn't count as an exploit against a default installation
<pie__>
i mean they probably do have a bettor track record, but i get the very slight feeling theres a bit of marketing in there too :P
<mbrock>
I have nothing against postfix for the record, I just chose opensmtpd after asking around for a mail daemon with a simple and understandable configuration format, and it's been humming along nicely ever since
goibhniu has quit [(Ping timeout: 240 seconds)]
ttuegel has joined #nixos
danielrf has quit [(Ping timeout: 240 seconds)]
<gchristensen>
ttuegel: re, your comment about our deprecation policy: I'm not inclined to let an unofficial policy stop me from doing the right thing :|
<ttuegel>
gchristensen: I don't disagree with you: we should remove KDE 4. I'm just frustrated because I implemented a deprecation mechanism for Nixpkgs years ago which was rejected, yet no other solution has materialized.
tvon has quit [(Quit: System is sleeping...)]
ambro718 has joined #nixos
tvon has joined #nixos
<MichaelRaskin>
I wonder if it is possible to keep stuff that is less likely to be affected by vulnerabilities, like Kig and Kdiff3…
simendsjo has joined #nixos
<ttuegel>
MichaelRaskin: Sure. We will keep kdelibs4 at least until upstream drops it, which is scheduled for December 2017.
<gchristensen>
ttuegel: oh I see, that is most unfortunate
<ttuegel>
If somebody wants to keep kdelibs4 even longer, I have no problem with that, but at that point upstream will stop releasing it as part of KDE 5 and it won't be part of my KDE 5 package set anymore.
<Profpatsch>
damnit, fpletz
<MichaelRaskin>
Maybe I should learn to package KDE5 stuff — Kig seems to have some tagged commits that should be KDE5-based releases
<Profpatsch>
I had the changes for searx 0.11 in my working dir right here.
<gchristensen>
Profpatsch: fpletz is good :)
<Profpatsch>
Now I got the biggest merge conflict in all of history. :P
<gchristensen>
aww
<Profpatsch>
It’s been out for a while.
<Profpatsch>
But updating 200 python packages is a horror.
<gchristensen>
a
<gchristensen>
real nightmare
pi3r has joined #nixos
<Profpatsch>
I made the mistake of wanting to run the searx tests.
<Profpatsch>
They depend on one rather high-level package of zope.
<gchristensen>
just by-default it is generated on the box
<eacameron>
fpletz: gchristensen: ekleog: Yes I think if it is configurable then that's ideal. The current status quo means people (like me) probably forget to even have dhparams.
<eacameron>
That's not exactly better.
<fpletz>
eacameron: +1
davidak has quit [(Quit: Leaving.)]
layus_ has joined #nixos
<eacameron>
ekleog: No wonder we didn't see the other module. The PR is from yesterday!
<eacameron>
Oh wait
<fpletz>
gchristensen: ok, agreedm byt we should ensure every tls-enabled service does this… there are even services like dovecot which generate dhparams themselves on first start and put them somewhere in /var/lib :)
<eacameron>
Haha...last year
<fpletz>
*agreed, but
<gchristensen>
sounds great :)
katyucha has quit [(Quit: leaving)]
layus__ has quit [(Ping timeout: 240 seconds)]
<musicmatze>
can someone help me overriding CFLAGS for a single package?
<musicmatze>
I cannot find any documentation on how to do this...
<FRidh>
Interesting to read how other distro's handle Python's bytecode and updates. The recent Python 3.5.3 maintenance update changed the 'magic' number, and therefore bytecode of 3.5.3 is different from 3.5.0 - 3.5.2. Some distro's weren't too happy about it, since they don't do any rebuilds of dependents after such maintainence updates (of Python).
<NixOS_GitHub>
[nixpkgs] Mounium opened pull request #22635: libreoffice-still: Added support for Hungarian (master...patch-3) https://git.io/vDa0G
digitus has joined #nixos
ilyaigpetrov has joined #nixos
<fpletz>
eacameron: yeah, every service actually :) but we also need an interface how to configure the number of bits for every service, probably makes sense to do it in the service? and to be able to override the dhparams with a custom file it should also take types.path value
<eacameron>
fpletz: I see. Yah that'd be nice. It seems that ekleog's PR would actually work fine for other services by just configuring a new dhparams attribute.
<eacameron>
In other words, services could use his "global" dhparams module to build the params file.
<NixOS_GitHub>
[nixpkgs] 7c6f434c closed pull request #22635: libreoffice-still: Added support for Hungarian (master...patch-3) https://git.io/vDa0G
bennofs1 has quit [(Ping timeout: 252 seconds)]
<NixOS_GitHub>
[nixpkgs] 7c6f434c closed pull request #22535: libreoffice: Adding support for Hungarian language (master...patch-2) https://git.io/vDBCz
<eacameron>
Ok so "sendmail" is weird. It doesn't seem to exist in a package per se. It's configured by a service. How do I tell PHP to use the one installed by the service?
jgertm has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vDauf
<NixOS_GitHub>
nixpkgs/master 8c6f811 Tuomas Tynkkynen: libvpx: 1.5.0 -> 1.6.1...
pombreda has joined #nixos
<teto>
nixos noob here, I try overriding a derivation to compile the master of neovim but when running $ nix-shell, I have "run/current-system/sw/bin/nix-shell: a single derivation is required" with https://transfer.sh/cEGv2/default.nix (6 lines), any idea why please ? (and possible to add it https://nixos.org/wiki/Error_Messages ?)
<pem_>
teto: pkgs: rec {… is a function
<pem_>
taking "pkgs" as argument.
<pem_>
You called it packageOverrides
Sonarpulse has joined #nixos
ttuegel has quit [(Quit: No Ping reply in 180 seconds.)]
<pem_>
Now you need to called it, probably with the "real" pkgs.
ttuegel has joined #nixos
<pem_>
(which you can get using "with import <nixpkgs> {};")
<musicmatze>
gchristensen: I cannot get it working -.-'
pombreda has quit [(Ping timeout: 252 seconds)]
bennofs1 has joined #nixos
<musicmatze>
I'm currently trying this in nix-repl: :b pkgs.lib.overrideDerivation pkgs.dmenu (o: { buildPhase = ''echo CFLAGS: $CFLAGS; export CFLAGS=DOPOPOPOPOPOPOPOPOPOPOPOP''; })
<teto>
pem_: thanks for the explanation, I can't say I understood much since nix doesn't look like any of the language I know so I will start by having a closer look at it
alx741_ has quit [(Quit: alx741_)]
alx741 has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 5 new commits to master: https://git.io/vDaVX
<NixOS_GitHub>
nixpkgs/master 6a48087 Tuomas Tynkkynen: jade: Cleanup a bit...
<NixOS_GitHub>
nixpkgs/master 6bf0a98 Tuomas Tynkkynen: ding-libs: Some cleanups...
<NixOS_GitHub>
nixpkgs/master 1862a9a Tuomas Tynkkynen: perlPackages.TextWrapI18N: Fix glibc path...
alx741 has quit [(Client Quit)]
alx741 has joined #nixos
pi3r has joined #nixos
<maurer>
Anyone have optimus working under nixos?
<maurer>
I've tried a number of configurations, but cannot manage to get optirun or chromium to work
<MichaelRaskin>
optirun part is actually easy to get working most of the time…
<MichaelRaskin>
Just enable the bumblebeed
tvon has quit [(Quit: System is sleeping...)]
<maurer>
So, optirun gives me a gl version error and dies
<maurer>
I'm turning bumblebee back on now to try debugging this, so it'll be a quick reboot to get the exact error
<MichaelRaskin>
I don't think you need to reboot to launch bumblebee…
<maurer>
MichaelRaskin: It's probably not strictly required, but this whole thing has been so finicky I didn't want to trust rebuild switch to dtrt
xwvvvvwx has joined #nixos
<MichaelRaskin>
Are you using nouveau or nvidia drivers?
georges-duperon has quit [(Ping timeout: 252 seconds)]
tvon has joined #nixos
bfrog has joined #nixos
<NixOS_GitHub>
[nixpkgs] ttuegel pushed 5 new commits to master: https://git.io/vDa6x
<NixOS_GitHub>
nixpkgs/master 2ead5df Thomas Tuegel: elpa-packages 2017-02-10
<NixOS_GitHub>
nixpkgs/master 9fbb76c Thomas Tuegel: melpa-packages 2017-02-10
<NixOS_GitHub>
nixpkgs/master cc79f18 Thomas Tuegel: melpa-stable-packages 2017-02-10
<ttuegel>
It's unbelievable to me that, in this day and age, anyone is even *willing* to download software that isn't GPG signed.
mudri has joined #nixos
<gchristensen>
right?
<MichaelRaskin>
Nobody trusts GPG keys to be correct anyway
<simpson>
Trusted releases are hard.
<ttuegel>
I don't even mean in the "I know what the full legal name of the developer is."
<savanni>
I didn't even bother to install gpg...
<ttuegel>
I mean in the "This is the same thing the developer uploaded."
<gchristensen>
I mean
<gchristensen>
literally _any_ way to verify it
s33se has joined #nixos
<simpson>
gchristensen: Presumably something easier than the developer confirming git hashes over IRC?
<ttuegel>
I sign my GitHub commits and yeah, my GPG key isn't signed, so you have no idea who I *really* am, but you can be confident that I am at least the same unknown entity I have been all along.
<gchristensen>
simpson: they're binary artifacts, not even git shas
drakonis has joined #nixos
<simpson>
gchristensen: Oh, trusting *binaries*. Yeah, that's totally different; I thought that we were talking about trusting source tarballs.
<drakonis>
huh, there's a nixos newsletter now
<gchristensen>
yeah I mean
<gchristensen>
I trust the creators
<drakonis>
and i looked at the pull request that added overlays to nixos and i'm surprised that mozilla has a nixpkgs overlay now
<maurer>
MichaelRaskin: Sorry, got a phone call. The error optirun gives is:
<maurer>
[ 1966.435820] [ERROR]Cannot access secondary GPU - error: [XORG] (EE) /dev/dri/card0: failed to set DRM interface version 1.4: Permission denied
<gchristensen>
anyone happen to have a config where you can log in over ssh as root with a password? I can't manage to make it work, I get authentication denied errors from ssh and PAM in the logs.
<maurer>
MichaelRaskin: the error persists as root
<maurer>
MichaelRaskin: I'm using nvidia drivers, the nouveau drivers are blacklisted
<drakonis>
gchristensen: what is this impending branch off in the nixos weekly issue comment you posted
<LnL>
gchristensen: password login is disabled for root by default
<gchristensen>
LnL: I set permitRootLogin = "yes"
<MichaelRaskin>
maurer: Hm, I use nouveau with nvidia not installed…
<MichaelRaskin>
_what_ do you launch as root?
<gchristensen>
drakonis: we're branching off master to create release-17.03! :)
<drakonis>
oh!
<drakonis>
when's that? march?
<gchristensen>
like 20 days
<maurer>
MichaelRaskin: I tried "optirun chromium" as normal user, and "sudo optirun chromium" and they gave the same error
<drakonis>
oh, march it is then
<drakonis>
the versioning is very much like ubuntu's :v
<MichaelRaskin>
I think bumblebeed runs as bumblebee, if card0 is in for video group and bumblebee is not in it, that's the problem
<drakonis>
year and month
<maurer>
MichaelRaskin: The other issue I have, which I suspect is related, is that chromium has no video acceleration when switchable graphics are enabled
<maurer>
(it works fine with discrete)
<MichaelRaskin>
Maybe Chrome doesn't want to enable it on Intel, or on some kinds of Intel
<gchristensen>
LnL: is there something else I need to change there, to make it work?
<LnL>
gchristensen: strange, why do you want it?
avn has joined #nixos
<maurer>
MichaelRaskin: Chrome works fine, amusingly enough
<maurer>
it's just chromium that doesn't
<maurer>
root 4460 0.0 0.0 34856 2252 ? Ss 14:00 0:00 /nix/store/bpr1y6xwy7wzk4f46hz3j2vm1fjysazs-bumblebee-3.2.1/bin/bumblebeed --use-syslog -g wheel --driver nvidia --pm-method auto
<maurer>
indicates that bumblebee is running as root
<maurer>
not bumblebee
<maurer>
(in fact, there's no bumblebee user on my system it seems)
<MichaelRaskin>
It would be fun if it only works for me because I don't run mainline NixOS
<maurer>
I'm currently on nixos-unstable
<maurer>
what are you on? Something custom?
<drakonis>
hahaha welp, i'm catching up on the mailing list and nix-bundle has solved flatpak already
<MichaelRaskin>
Well, I just generate my bootscripts with hand-written Nix expressions just for that, and then export a few NixOS services into runner scripts
<MichaelRaskin>
So I don't have any real init
<maurer>
I see
<maurer>
Also potentially relevant, when I run glxinfo, I get no GLX extension
DarcsAurelius has joined #nixos
<maurer>
if I disable the nvidia card, I get a GLX based on intel
<maurer>
if I boot in discrete only mode, I get GLX based on nvidia
<maurer>
but if I leave optimus on, I get no GLX extensions
<MichaelRaskin>
Hm
<MichaelRaskin>
Wait what
<LnL>
gchristensen: did you set a password? I'm pretty sure empty passwords are not allowed by default either
<MichaelRaskin>
In the optimus mode, what does glxinfo say?
<gchristensen>
LnL: yeah, I tried that too
<MichaelRaskin>
In the third line
<maurer>
MichaelRaskin: Xlib: extension "GLX" missing on display ":0".
katyucha has joined #nixos
<maurer>
Almost every line is that
<MichaelRaskin>
Wow
<MichaelRaskin>
That's interesting…
<maurer>
If I use optiumsDisable = true; it goes away and flips to an intel provided GLX
<maurer>
If I flip to discrete in the bios, it also works
<maurer>
*optimus
volhovm has joined #nixos
<MichaelRaskin>
I wonder what will happen if you enable optimus mode in BIOS, disable optimus in NixOS, and enable only Intel drivers
<MichaelRaskin>
(worst case: X cannot start…)
<maurer>
MichaelRaskin: I've done that, this works fine for most programs, but not for chromium
<MichaelRaskin>
It may be that nvidia drivers just don't support reasonable versions of something
<MichaelRaskin>
At least bumblebee bug tracker looks like that
<MichaelRaskin>
(and we have libdrm that is too new for it)
<maurer>
Do you want me to try swapping to nouveau?
<MichaelRaskin>
I would at least try
<gchristensen>
LnL: well son of a gun, now it works!
<MichaelRaskin>
(I do have Optimus which I run on Intel + Nouveau — or Intel-only when trying to optimise for battery life)
bennofs1 has quit [(Ping timeout: 240 seconds)]
<maurer>
I mean, my end hope here is to be able to run primarily intel, and switch on the nvidia card to drive external monitors
<maurer>
and have no tearing in chromium while doing this
volhovm has left #nixos []
<MichaelRaskin>
Well, external monitors in such a setup is another cna of worms (I do have this somewhat working, but it was tricky; and it works less reliably than with an older kernel right now)
<maurer>
MichaelRaskin: With nouveau, chromium is accelerated, glxinfo reports well, optirun still has the same error
<MichaelRaskin>
How interesting
sigmundv has joined #nixos
<maurer>
I don't _really_ care about optirun working except to the extent that it might be related to whether I can flip on the nvidia card for driving monitors
ttuegel has quit [(Remote host closed the connection)]
<MichaelRaskin>
Oh well
<MichaelRaskin>
If you want to go the intel-virtual-output way, then optirun is needed
<MichaelRaskin>
If you want to configure Prime, then it may be unrelated…
<maurer>
I don't really care, just as long as I can drive my external monitors
<MichaelRaskin>
Where I use the lack of proper escaping in the NixOS X.org configuration generator to get the exact config I need…
ryanartecona has joined #nixos
<gchristensen>
hehe yeah that can be helpful sometimes :)
<MichaelRaskin>
The problem is: I don't run NixOS but I use its X.org module, I cannot easily test the patch I would want on NixOS in the case where it is supposed to matter
<DarcsAurelius>
I'm trying to install intero-0.1.20, which the emacs intero package uses by default. Only 0.1.18 and 19 are in the nixos 16.09 channels, so I looked at hydra, but none of the builds seem available anymore. Would the recommended option be to override the intero package with a custom src/version, or is there something easier I can do?
<MichaelRaskin>
So I cannot easily add a more reasonable way of doing the things
<maurer>
re: prime, is there a way to get xrandr to list providers?
<maurer>
nevermind, I'm dumb
<maurer>
--listproviders
<maurer>
It only lists the intel card though, not the nvidia one
<maurer>
I'll try your adjustment to xorg I guess
<MichaelRaskin>
Yep, I was checking whether the hyphen is needed
<MichaelRaskin>
For prime — surely yes
<MichaelRaskin>
What kernel do you run?
<maurer>
Linux durandal 4.9.8 #1-NixOS SMP Sat Feb 4 08:47:29 UTC 2017 x86_64 GNU/Linux
<maurer>
Not heavily attached to it though, I can upgrade/downgrade
<MichaelRaskin>
No-no
<MichaelRaskin>
The point is if it works fine for you, I may try switching from bumblebee-based setup to prime-based setup again
<maurer>
Ah, prime is not working for me
<maurer>
:P
<maurer>
I'm looking into whether I can make it work
<maurer>
I'm gonna disable bumblebee while I investigate
<MichaelRaskin>
I know it is not, but You didn't have proper xorg.conf whan you tried, right?
<MichaelRaskin>
I think having bumblebee and not using it should still be fine
<maurer>
I did not, no
<MichaelRaskin>
But yes, I understand taking no chances
stepcut has joined #nixos
<MichaelRaskin>
The point is: I have a mildly broken situation now, and I expect that The Bright Future is prime; and previously my setup with bumblebee worked better. For some strange reason your setup doesn't work with bumblebee at all, so you _will_ actually try prime.
<MichaelRaskin>
This can be a validation that the prime time has come
<binaryphile>
was wondering if there is a way to make a ~/.nixpkgs/config.nix that works for an alternate (private) package repo the way nixpkgs grabs config.nix automatically
<binaryphile>
not sure how nixpkgs is making it automatically get passed in as the config var
<drakonis>
wait a little longer for overlays
<binaryphile>
wonder if that's hardcoded into nix special for nixpkgs
<drakonis>
then you can have as many as you want :v
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/vDaHA
<NixOS_GitHub>
nixpkgs/master 3809938 Dan Peebles: ecs-agent module: remove debug print...
<maurer>
MichaelRaskin: we're evidently at the latest nvidia driver already
<maurer>
so I dunno
mkoenig has joined #nixos
<MichaelRaskin>
The problem is that the latest nvidia driver is normally lagging behind the X.org versions of interfaces
danharaj has quit [(Ping timeout: 240 seconds)]
<niksnut>
gchristensen: I thought it already is the default?
<niksnut>
on the installation CD
<NixOS_GitHub>
[nixpkgs] nicknovitski opened pull request #22639: bash-completion: call readlink without errors on Darwin (master...darwin-readlink-bashcompletion) https://git.io/vDapz
<drakonis>
why is kde4 the default DE on the iso
<niksnut>
it isn't
<niksnut>
KDE 5 is the default
<niksnut>
(on master)
<drakonis>
hm
<drakonis>
ncie.
<drakonis>
nice.
<gchristensen>
niksnut: on master? I don't know, maybe, but the generated config includes defaults for kdm / kde4
<gchristensen>
niksnut: the PR I mean is #22610
<NixOS_GitHub>
[nixpkgs] edolstra closed pull request #22610: nixos: update default cases from KDM/KDE4 to SDDM/KDE5 (master...switch-to-kde5-by-default) https://git.io/vDgb5
<niksnut>
right, thanks
<gchristensen>
yay! I hate to see Baughn having issues with it, but I think we need to keep pushing forward ...
<gchristensen>
niksnut: is there a way to warn when enabling kde4? I read around some code / the manual, but didn't find a way.
<niksnut>
the "warnings" option
<gchristensen>
ooh! I must have missed that. I'll get a PR in this evening. thank you!
<viric>
gchristensen: do you use kde5?
<gchristensen>
no
<gchristensen>
I did for a bit yesterday when I was playing around, it was pretty good!
Itkovian has joined #nixos
<gchristensen>
I liked it _far_ more than I expected
mk-fg has quit [(Ping timeout: 240 seconds)]
<obadz>
Is the chromium failure in release-16.09 real?
sellout- has quit [(Quit: Leaving.)]
<viric>
gchristensen: I only used kdenlive
<viric>
gchristensen: and I could not use it with all icons and so on
<viric>
gchristensen: I don't know, yet, how to make a kde5 program work fine out of a kde5 desktop
stepcut has quit [(Remote host closed the connection)]
derjohn_mob has quit [(Remote host closed the connection)]