<abathur>
nix#4925 sounds to me like there's been a latent issue in Nix (at least 2.3.x) for a while, and that some shift (probably in nixpkgs in the past several days?) has suddenly exposed it and caused a number of people with single-user installs on Linux to present
<lukegb>
Yeah, I was trying to work out if there's a particular filesystem structure or something that would cause it
<abathur>
and https://github.com/NixOS/nix/issues/4295#issuecomment-738504227 feels like the first comment that has the issue drilled down far enough that someone with sufficient nix-fu may have a sense of what's actually going on here, and whether the role Nixpkgs may be playing is a problem in its own right
<lukegb>
My gut feeling is fetchurl or related though
andi- has joined #nixos-dev
<abathur>
ah, hmm, yeah; I guess I can imagine that
<abathur>
I've been wondering if something changed in a builder or hook, though that comment says it's pretty global
<abathur>
so fetch would make more sense in that light
<V>
I don't think the regex catches URLs with fragments?
<abathur>
V I did the short nix#id fmt first
<V>
nix#4295
<abathur>
I assumed it was dead, but maybe it doesn't notice first-word
<{^_^}>
https://github.com/NixOS/nix/issues/4295 (by fzakaria, 3 days ago, open): error: moving build output '/nix/store/...-source' from the sandbox to the Nix store: Permission denied
<V>
nah it's just favouritism
<abathur>
or maybe it tracks who mentions it and doesn't give you another?
<abathur>
seems doubtful
<V>
(or possibly because you posted a lot of links recently)
<abathur>
(but I did mention it this morning)
<abathur>
nah
<abathur>
I've had a sock in it alllll day :D
<abathur>
nix#4295
<{^_^}>
https://github.com/NixOS/nix/issues/4295 (by fzakaria, 3 days ago, open): error: moving build output '/nix/store/...-source' from the sandbox to the Nix store: Permission denied
<abathur>
bog
<abathur>
:)
<abathur>
maybe it was actually disconnected briefly
<lukegb>
you said 4925 instead of 4295
<lukegb>
:^)
<abathur>
oh
<abathur>
yeesh
<abathur>
<3 lukegb
<{^_^}>
lukegb's karma got increased to 9
supersandro2000 has quit [Read error: Connection reset by peer]
supersandro2000 has joined #nixos-dev
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-dev
<lukegb>
abathur: yeah, figured it out
<lukegb>
rename requires you to have write permissions on the directory you're renaming, because it updates ".."
<V>
oho
<abathur>
lulz
<abathur>
good errata to know
<abathur>
makes sense, but I can see how it's easy to miss
<{^_^}>
#105885 (by cmm, 15 minutes ago, open): firefox removed all my add-ons
<eyJhb>
Pinging Mic92 :)
<supersandro2000>
I'd rather merge the bugfix than reverting things now if we can fix it in a day or two
<Mic92>
supersandro2000: makes sense
<supersandro2000>
but I do not use Firefox with nix so I can't test things
<eyJhb>
It makes sense, but at the same time it should maybe have been reverted before it could actually do some damagae, but atm. it is just poluting :) But yes, it needs to be merged, if it is ready. Also, I can't really test it, as I run a weird setup of FF
<eyJhb>
So mine resets whenever anyways
<siraben>
Will we be able to specify Firefox flags in configuration.nix/home.nix the future as well?
<siraben>
s/the/in the
<supersandro2000>
eyJhb: it is on unstable. Please don't rush anything
<andi->
Is there a way to disable the in-browser "apply suggested changes" for the repository? It really messes up the commit history and less git savvy users are getting confused about fetching again and rebasing..
<qyliss>
I'd be very surprised
<qyliss>
I think committers should just do the rebase in lots of cases
<andi->
Asking people to do things is not working unless you enforce it..
<andi->
I've looked at https://jorisroovers.com/gitlint/ this morning to configure in a way that at least enforces proper commit messages.
<qyliss>
interesting
<eyJhb>
siraben: you can already do so in home-manager :)
<eyJhb>
supersandro2000: I see where that comes from, and I know it is unstable, but a browser is fundamental :)
<siraben>
eyJhb: ooh
<eyJhb>
siraben: now it is just a matter of using this new PR instead maybe :)
<ryantm>
<jtojnar "ryantm: opened https://github.co"> Cool! Though I hope we can eradicate Docbook before we can practically use that.
mmlb22 has joined #nixos-dev
<FRidh>
ryantm: there is a desire from some contributors, in varying degrees, to have additional features on top of commonmark
<ryantm>
FRidh: Yeah, I think the way we are doing heading anchors is non-standard (not totally sure).
<ryantm>
The RFC accounts for allowing additional features.
<FRidh>
ryantm: because there is no standard way
<ryantm>
They need to be available in multiple implementations
<ryantm>
It would be good to start an Issue with a list of non-standard features that we want to use and document which ones are approved to use.
<FRidh>
sounds like stuff for an rfc
<ryantm>
I think the RFC gives us leeway to use non-standard features as long as we vet they are available in multiple implementations; I'm hoping that is enough of a guideline to move forward without another RFC.
<{^_^}>
#105845 (by lukegb, 13 hours ago, open): fetchzip: get write permission on unpacked directory
<lukegb>
there's a corresponding Nix patch that's in 2.3.10 but imo it's probably best to workaround for people currently on 2.3.9, especially since 2.3.9 is current at latest master :o)
* lukegb
steps off soapbox
alp has quit [Ping timeout: 256 seconds]
cole-h has quit [Ping timeout: 240 seconds]
rajivr has quit [Quit: Connection closed for inactivity]
thibm has quit [Quit: WeeChat 2.6]
<ekleog>
Did anyone ever look at whether stalebot was actually useful, by eg. looking at the ratio between things correctly marked as stale and things that people have to unmark as stale?
<jtojnar>
ryantm: that format only seems to define new in-document alias – it does not look like it gets stored in the generated docbook file
<jtojnar>
so we cannot use it unless we change the way we generate the manual
ris has joined #nixos-dev
<jtojnar>
ryantm: regarding the admonitions, we can just add the patch to nixpkgs untill pandoc is updated
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
<jtojnar>
we will want to have admonitions regardless of the final format so it does not make sense to me to convert them to blockquotes
<ryantm>
Yeah sounds good. I merged some changesets that converted them to blockquotes because I didn't think it looked too bad, but I can go back and fix those.