<ekleog>
huh… anyone managing to comment on https://github.com/NixOS/nixpkgs/pull/11886 ? I'd like to comment “(triage) @spinus, do you still plan to rebase this PR and have a look at whether the issue with systemd user services still happens? :)”, but get a “You can't comment at this time.” at every attempt…
<{^_^}>
#11886 (by spinus, 2 years ago, open): refactor pam-u2f: add keysPath, verbose, fix docs about u2f_keys path
<ekleog>
oh… weird, somehow it did work for other PR's… thanks! :)
<samueldr>
not that it's because of that... but github seems to be having issues right now :)
<ekleog>
huh and it looks like github just sent me one mail per attempt at sending this comment :D
pie___ is now known as pie_
Cale has quit [Ping timeout: 260 seconds]
lassulus_ has joined #nixos-dev
lassulus has quit [Ping timeout: 252 seconds]
lassulus_ is now known as lassulus
orivej has quit [Ping timeout: 240 seconds]
drakonis has quit [Quit: WeeChat 2.2]
jtojnar has joined #nixos-dev
jtojnar has quit [Ping timeout: 244 seconds]
drakonis has joined #nixos-dev
<ekleog>
oh just had an idea
<ekleog>
one issue with derivations `cp`'ing files is that it needs time and then later will be hardlinked by the optimize-store
<ekleog>
which churns the hard disk badly
<ekleog>
now, it's not possible to `ln`, because nix uses `mount --bind` and you can't hardlink across filesystems
<ekleog>
what about a FUSE FS that'd restrict the store paths visible by the guest the same way the usual wave of `mount --bind` does, and would forward hardlink requests to the underlying FS?
<ekleog>
it'd require the store to be a single FS and not split between several FS's, but I guess that's a reasonable expectation
<ekleog>
and then derivations could directly `ln` from their inputs to their outputs
<ekleog>
(use case: generating VM's store, I gave up generating them in the store because it took too long cp'ing everything to just rm it after the optimize-store passed, and just rsync'd to /var, which is both impure and ugly)
<ekleog>
(and using mount --bind with all the VM's store path just makes things awfully slow, somehow linux doesn't like having hundreds of thousands of bind-mounts (a))
drakonis has quit [Ping timeout: 264 seconds]
drakonis has joined #nixos-dev
FRidh has joined #nixos-dev
<srhb>
Unstable is blocked on https://hydra.nixos.org/build/83153504 -- this looks like a very rare case of nondeterminism in the tests, can someone restart to verify?
<ekleog>
does that sound good to you? it's less deprecation time than would be ideal, but it's for master anyway, and 19.03 can have release notes…
<ekleog>
(btw, long live alt+prtscr+f, sounds like I don't have enough RAM for building the tarball :D)
<srhb>
ekleog: It sounds good to me for now at least.
<srhb>
But I've been known to be fast-and-loose with attribute removal, since I'm OK with build time errors. :-)
<ekleog>
meh, same here
<ekleog>
but well, there's really no other choice anyway…
<ekleog>
(currently at least)
<srhb>
If someone comes around with a better solution, we can implement that. For now, it sounds like a fine way to unblock. :)
<ekleog>
hmm, any way to add release notes to nixpkgs? I'm just thinking this should go into nixpkgs release notes, but it looks like doc/release-notes.xml has been untouched since 2012 :/
<srhb>
ekleog: I think we only use the NixOS ones, but they tend to serve both purposes.
<ekleog>
yeah, I'm a bit sad because that means a human must then go and split nixos from nixpkgs when reading release notes that only interest them from nixpkgs :(
<ekleog>
thank *you* for the heads'up! (was doing a format-patch in case you couldn't get from the other url either, but looks like http://ix.io/1pKm will be useless ^^)
<srhb>
ekleog: I did git pr 48810 (alias for fetching the pr ref) and cherryed it over.
<ekleog>
nice :) so it was really just the webui being down
<ekleog>
btw I'm curious, how did you alias it? afaiu you need `git fetch origin pull/[id]/head:pr-[id]`, which is not alias-able per se?
<ekleog>
guess I can imagine the script, though :)
<srhb>
Maybe it'll work later :P
<ekleog>
yeah ^^
__Sander__ has joined #nixos-dev
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos-dev
orivej has joined #nixos-dev
FRidh has quit [Quit: Konversation terminated!]
sir_guy_carleton has joined #nixos-dev
pie_ has quit [Ping timeout: 264 seconds]
<gchristensen>
mac2 is alive again! for a "long time" like half the macs have been broken. mac3 recently started working, mac2 _just_ started working, and the rest will be healthy again soon too
<srhb>
gchristensen: Nice :)
<domenkozar>
is coretemp on IRC? I'm losing my patience with the "criticism".
<vcunat>
gchristensen: do you think all our mac HW supports VT-x
<vcunat>
(I wonder about performance penalties.)
<gchristensen>
vcunat: I did a performance comparison and found there was a minor hit, but nothing extreme. I haven't checked all 9 of them but I would be surprised if any of them didn't
<LnL>
you'd have noticed if that was the case
<gchristensen>
I've only converted 2 macs now, mac2 and mac3
<srhb>
Rerun: Unstable is blocked on https://hydra.nixos.org/build/83153504 -- this looks like a very rare case of nondeterminism in the tests, can someone restart to verify?
<srhb>
... either someone already did or someone's very quick on the trigger. Thanks!
<gchristensen>
they're all i5-3210M or i5-4278U and both of those have VT-x for sure, good question vcunat. can you tell me yours? sysctl -n machdep.cpu.brand_string
<vcunat>
it's core2 IIRC
<vcunat>
I tried to ssh during the weekend but it failed :-)
<vcunat>
Some core2 have it and some don't from what I've read on Wiki.
<vcunat>
A single machine won't do a large difference in any case.
orivej has quit [Ping timeout: 245 seconds]
<vcunat>
I'll try to remember, power it on in the evening and check for the feature. (I thas no DHCP lease ATM.)
<gchristensen>
ok cool
<gchristensen>
and fingers crossed the virtualisation method is more effective
<vcunat>
Yes, the queue length is still diverging to infinity so far.
<simpson>
Well, you can consider me to be the typical person who will have to leave if the typical CoC is adopted.
<simpson>
But AFAICT almost all folks are reasonable, to the point where you *could* ban specific people on an ad-hoc basis and that would be acceptable.
<domenkozar>
I want to have rules based on which w judge that
<simpson>
Of course, given *that*, we should talk about who you want to ban! This `coretemp` person sounds like an ass, but whatever, right?
<domenkozar>
if you want anarchy, that's fine, it doesn't work in big groups :)
<simpson>
I don't want anarchy, I want thicker skins.
<domenkozar>
guix is a reasonable small not to have this problems
drakonis has joined #nixos-dev
<domenkozar>
as was Nix 5 years ago
<domenkozar>
I understand some people will leave
<domenkozar>
but I don't have a problem if people leave due to idelogoy while we're solving problems here
<domenkozar>
even though it's sad
<simpson>
It's not sad. It's just people.
<gchristensen>
we can have guidelines and ban on guidelines, I think rules that try to be too specific are not too useful
<simpson>
(We aren't effective altruists here, but folks who want a better UNIX userland experience.)
<gchristensen>
more important than a coc, to me, is coming up with guidelines on how we want to treat each other and then a way to discuss potential problems and come to solutions
<domenkozar>
should have said gudielines, not rules
<domenkozar>
it's not binding
<gchristensen>
I don't care if we call it a coc or not, and I don't care if we adopt a well known coc or not
<domenkozar>
we all want a better world, disagreement comes which paths to pick :)
<gchristensen>
I just want us to treat each other with care
<gchristensen>
importantly, I want us to have a group of people so the labor of deciding how to handle a disruptive person in shared and not loaded on to one person
<gchristensen>
(or even if a person is being disruptive, which is possibly more tiring)
<gchristensen>
I'm heading to an airport in about 1hr, and I won't be around for many hours
<ekleog>
simpson: I'm curious, why would you have to leave if the typical CoC is adopted? (feel free to ignore me if it's for personal reasons)
<simpson>
ekleog: Because most CoCs ask behavior of me that is not going to happen if I want to participate in the community for an extended period of time.
<gchristensen>
because it is divisive
<simpson>
I am the kind of person that respects safe spaces by /part'ing them.
<samueldr>
it all depends what you define as a "safe space"
<samueldr>
one can ask of others to respect the person, even if you don't respect their opinion
<ekleog>
^ that's basically what I read in most CoC's
<samueldr>
and in that situation respecting most time is "don't say anything"
<simpson>
samueldr: Don't tell me how to act, thanks~
<samueldr>
so don't say X is a big fat poopy pants, but X's idea is horrible
<samueldr>
sorry simpson
<simpson>
Whatever. Y'all want *somebody* to tell this coretemp person off, you just don't want to get your hands dirty.
<gchristensen>
I'm happy to do it
<gchristensen>
it just tiring
<gchristensen>
very, very draining
<samueldr>
simpson: but as of right now, your actions have proven to me you're more likely to do the right thing :)
<globin>
I'd prefer we talk about this in london over the next few days?
<samueldr>
(actions on nixos channels here, maybe I'm missing things)
<gchristensen>
yes I agree, globin
<globin>
especially in person
<gchristensen>
IRC is not a good place to discuss
xeji has joined #nixos-dev
<domenkozar>
ye
<domenkozar>
simpson: I will say this, I'm a fan of empirical evidence, no matter how good or bad the theory
<domenkozar>
CoC turned out to be a good thing for hundreds of communities
<domenkozar>
now, the intent behind writing such document has to do more than the document itself
<gchristensen>
(somebody has to be the first to stop replying in order for this to not happen on IRC)
<ekleog>
gchristensen: time to op yourself and kickban everyone :D
<domenkozar>
and it turns out in practice, it works well
<ekleog>
that's the alternative method
<domenkozar>
well I assume simpson won't be at nixcon
<domenkozar>
yeah tat'sa good example of trolling without good intentions
<gchristensen>
yes
<domenkozar>
to devalue coc and "prove a point"
<gchristensen>
I really do not like that I started this conversation hours before I am offline for many hours. I very very much did not want to do it this way.
<ekleog>
well, it's trolling but it does fulfill the same role as a CoC :)
<simpson>
domenkozar: IYSS, I guess. I've seen lots of folks lose their jobs/etc. for no good reason due to this sort of thing, and I'd rather stay at home than risk unemployment.
andi- has quit [Remote host closed the connection]
* ekleog
just tried to lighten the atmosphere and subtly move the center of attention
<gchristensen>
ekleog: it isn't working
<sphalerite>
+1 for discussing it at nixcon
<adisbladis>
"Do not love much or boisterous laughter"
<adisbladis>
No fun allowed
<domenkozar>
ekleog: yes lots of people will leave sqlite development because a person is not mature
<domenkozar>
I'd call that progress nevertheless
<simpson>
adisbladis, domenkozar: I am sad that you think that a "non-traditional" (extremely traditional!) CoC is "trolling". It sounds like you're not willing to tolerate folks with different views of how communities should function.
<adisbladis>
simpson: No I'm very critical of most CoCs.
<simpson>
domenkozar: Nope. Variety of reasons, biggest being that I don't have commit bit and this is clearly a committer's summit.
<domenkozar>
where did you get that impression?
<simpson>
Although you can put down "trapped in a Stage 4 fascist police state and unwilling to cross the border" as a close second.
<domenkozar>
that I can believe in :)
<domenkozar>
maybe we should say that mostly people that do want to get more involved are welcome
<domenkozar>
those that already are, probably know most of the talks already
<zimbatm>
simpson: also, NixCon has a CoC :D
<simpson>
zimbatm: Hope nobody gets fired, then.
<zimbatm>
next year we'll go some place more friendly
<zimbatm>
only fired up
makefu has quit [Ping timeout: 245 seconds]
<domenkozar>
simpson: good thing noone is employed at nixcon :P
<domenkozar>
so there's noone to fire
drakonis1 has quit [Quit: WeeChat 2.2]
<adisbladis>
Lets employ someone so we can fire them
<Taneb>
adisbladis: feels like a lot of paperwork
<infinisil>
domenkozar: I didn't get the impression of it only being a commitbit summit
<infinisil>
I was also there last year when I didn't have the bit at least :)
<domenkozar>
glad to hear that :)
<adisbladis>
"NixCon is a community-oriented conference for contributors and users of Nix and NixOS."
<adisbladis>
First sentence on the page
<domenkozar>
I never doubted zimbatm for a second, just wanted to see based on what information that statement was made
<gchristensen>
I think I know where simpson got the idea, though. last year's tickets were given out in order of how many contributions you make.
<gchristensen>
since the size was so limited
<domenkozar>
ah yes, we do have scaling issues :)
<simpson>
domenkozar: By analogy with how other large open-source/Free Software groups run their conventions. The size limits are similar to Linux's developer summit, for example.
rsa has joined #nixos-dev
<domenkozar>
I hope we can eventually have a limit that is not reached, but usually places become expensive once you go above 100 people
<simpson>
ekleog: Yes, I know. And I weigh CoCs against other factors when deciding whether to attend events.
<ekleog>
oh sorry you meant your sentence for “for whom is NixCon” ._.
<ekleog>
I thought you meant NixCon didn't need a CoC because the linux developer summits don't need CoCs
<thoughtpolice>
domenkozar: Yeah, most people seem to bankroll it purely through employer relations/sponsorships, realistically. 100 people is a pretty tight limit though. That said I get prioritizing tickets, though I find it a bit sad picking and choosing since having new interested people is always enjoyable.
<gchristensen>
agreed
<gchristensen>
with more sponsors we can get a bigger place
<simpson>
Ah, sorry. No, nothing like that. My position is near the polar opposite and contrapositive: *Because* there are typically CoCs, and because there is a small-but-non-negligible chance that I could lose my job merely by attending a conference with a CoC, therefore I should weigh that possibility against whatever motivation I have for attending.
<domenkozar>
I think we can solve most of such issues by picking a city that's afforedable
<gchristensen>
let's all come to Berkshire County, MA! NixCon: Graham's House
<adisbladis>
gchristensen: Is that an invite? :)
<domenkozar>
since we're flying everyone in, but that's hard to do since intersection between those willing to organize and those living in a cheap city so far is empty set
<gchristensen>
sure
<ekleog>
there's the lottery solution for distributing tickets, but it requires some more complexity by not actually charging the tickets but only having a payment authorization and triggering payment only after lottery is done, I guess
sir_guy_carleton has joined #nixos-dev
<ekleog>
there's the “rush and first-come-first-serve” strategy too
<domenkozar>
so, who's organizing nixcon 2019 in a cheap place? :)
<gchristensen>
ok bye, leaving for real now.
<ekleog>
see you :)
* domenkozar
waves
<adisbladis>
gchristensen: \o
<adisbladis>
Cya
<globin>
o/
<domenkozar>
you might also say that london at 25 eur is so affordable that everyone wanted to come just for london
<domenkozar>
:D
<ekleog>
domenkozar: it's all a convenience * quality / price ration, I guess
<infinisil>
ekleog: i want a lottery solution weighted by commit count, earliness and subjective disappointment of not being chosen, with a good balance between those
<infinisil>
For some definition of good..
<ekleog>
the later one being self-declared? that sounds interesting :)
<adisbladis>
Considering we didn't run out of tickets for quite a while ticket lottery wouldn't have helped much. If you really wanted to go there was ample time.
<samueldr>
earliness... I thought this had to do with ears for a moment :) like "I am a good listener"
<ekleog>
like, you could take a ticket but still say “meh it's not a problem if I don't get picked”
<ekleog>
adisbladis: oh I assumed there were size constraints
<infinisil>
ekleog: yeah, sounds like a good thing to have
* ekleog
couldn't get there for technical reasons anyway
<ekleog>
next year hopefully :)
<domenkozar>
hmmz :)
<domenkozar>
I highly doubt there's a fair system
<adisbladis>
Battle Royale
<domenkozar>
better to strive for more sponsors+cheaper venue
<samueldr>
in a field, in the middle of nowhere :)
<sphalerite>
with tents!
<domenkozar>
could go with tesla tent
<adisbladis>
Nixcamp :3
<domenkozar>
oh actually there's a burningman competition in spain called nowehre
<samueldr>
(though, I have to say technological events can happen and be successful in tents!)
<domenkozar>
could just run our own track
<Taneb>
samueldr: there's something I've been meaning to go to for a while that's like that, called ElectoMagnetic Field
<domenkozar>
aso goes well with NL translation of Nix
drakonis1 has joined #nixos-dev
andi- has joined #nixos-dev
vcunat has quit [Ping timeout: 260 seconds]
<adisbladis>
Taneb: EMF seems cool. Planning to go next summer.
<Taneb>
adisbladis: I thought it was every two years?
<Taneb>
But I'm planning on going next time, regardless
<adisbladis>
Taneb: I thought it was yearly? Anyway.. Next time
<adisbladis>
It was too far this time
<zimbatm>
we had a 2 year break at one point but it wasn't on purpose
<adisbladis>
It would be awesome if someone has an in at a university for next nixcon
<sphalerite>
NixCon at uu.nl :D
<sphalerite>
although they're not even hosting our mailing list anymore D:
<domenkozar>
zimbatm: I guess it feels great after spending months on this to hear people take 5min to think :D
<zimbatm>
haha I don't mind
<domenkozar>
anyway, that should encourage people to apply for organizing 2019
<zimbatm>
better have NixCon in London than no NixCon at all
<ekleog>
sphalerite: optical fiber is like 0.66€/m, if we pick a field not too far from a village it could do :D
<domenkozar>
exactly :)
<domenkozar>
conferences are like engineering
<domenkozar>
ots of corners you ahve to cut
<zimbatm>
2019 should be a breeze, we have almost everything templated
<domenkozar>
unless you have Jonas that templates everything
<adisbladis>
The Jonas templating engine is great
<zimbatm>
lol
<zimbatm>
I think everyone should organise a conference once in a life
<zimbatm>
I definitely have a different perspective of what it takes to do it
<zimbatm>
it doesn't take that long, it's just a lot of tiny little pieces expanded over the whole year
<domenkozar>
the problem becomes once one year sets standards really high
<domenkozar>
then noone dares to meet them :)
<domenkozar>
seens it happen a few times
<domenkozar>
seen*
FRidh has joined #nixos-dev
sir_guy_carleton has quit [Quit: WeeChat 2.2]
__Sander__ has quit [Quit: Konversation terminated!]
Lisanna has joined #nixos-dev
<ekleog>
hmm, any reason why emails to meta.maintainers are not turned back on?
<ekleog>
ISTR they were disabled because introduction of a new jobset triggered lots of mail-sending, but…
<ekleog>
(also, I was thinking about making it actually more generic and making it possible for maintainers to get a mail at each rebuild of their packages, because at some point a package I was maintaining has been re-taken by someone else without me noticing, thinking I had to do some packaging work again and not finding time for it while someone else had actually done it)
xeji has quit [Ping timeout: 264 seconds]
jtojnar has joined #nixos-dev
FRidh has quit [Quit: Konversation terminated!]
cransom has quit [Quit: WeeChat 2.1]
cransom has joined #nixos-dev
makefu has joined #nixos-dev
Synthetica has joined #nixos-dev
sir_guy_carleton has joined #nixos-dev
drakonis1 has quit [Quit: WeeChat 2.2]
drakonis_ has joined #nixos-dev
drakonis_ has quit [Read error: Connection reset by peer]
drakonis1 has joined #nixos-dev
drakonis has quit [Ping timeout: 252 seconds]
drakonis has joined #nixos-dev
drakonis_ has joined #nixos-dev
drakonis1 has quit [Ping timeout: 264 seconds]
* gchristensen
is feeling proud of the community today