<aanderse>
gchristensen: if i test this is there any chance I'm going to have a bad Saturday? ;-)
<gchristensen>
wellllll
<aanderse>
i don't mind having a bad Monday
<aanderse>
but a bad Saturday....
<gchristensen>
I feel pretty confident that is a "no", but I am never going to want you to have a bad Saturday, so do wait until Monday :P
<aanderse>
lol ok
<aanderse>
putting in calendar now. Monday @ 7:30
<gchristensen>
thanks
malSet has quit [Ping timeout: 260 seconds]
<gchristensen>
for comparison, I have never seen this branch not work correctly. that is why I feel good about it :P
<aanderse>
sounds good
malSet has joined #nixos-chat
malSet has quit [Ping timeout: 255 seconds]
malSet has joined #nixos-chat
malSet has quit [Ping timeout: 260 seconds]
malSet has joined #nixos-chat
malSet has quit [Ping timeout: 255 seconds]
malSet has joined #nixos-chat
malSet has quit [Ping timeout: 260 seconds]
malSet has joined #nixos-chat
drakonis has quit [Quit: WeeChat 2.7.1]
eyJhb has quit [Quit: Clever message]
eyJhb has joined #nixos-chat
cole-h has joined #nixos-chat
maxdevjs has quit [Ping timeout: 272 seconds]
waleee-cl has joined #nixos-chat
maxdevjs has joined #nixos-chat
abathur has joined #nixos-chat
MichaelRaskin has joined #nixos-chat
<__monty__>
Contributing to RFC discussions: "Yeah, I'm not gonna read all that here's what I want to say." Makes you feel like people are really interested in other people's opinions...
<emily>
in this case I was specifically asked to comment and I've spent sufficient amount of time with lightweight markup languages to offer my 2¢ same as everyone else is doing ¯\_(ツ)_/¯
<emily>
otherwise, if you'd like to pay me for the hours it'd take me to hit expand on the "68 hidden items" and read every single one...
<samueldr>
that's a good way to gatekeep partitipcation only for the select few that found the RFC early on
<samueldr>
s/partitipcation/participation/ # I swear I know how to spell... not sure about typing
<emily>
(and to be more blunt, given that the brief skim showed that people were suggesting things like LaTeX and org-mode I'm pretty sure the comments are just an unproductive trash fire.)
<samueldr>
not having the time, wherewithal, or else to engage in full shouldn't diminish the value one wants to bring to the discussion
<emily>
(I don't know how much time anyone suggesting LaTeX has spent trying to convert it to good HTML, but I'm pretty sure I've lost more to it than them)
<gchristensen>
emily: ouch
yegortimoshenko has joined #nixos-chat
<joepie91>
I'm surprised how few RFC-esque processes involve a maintainer summarizing the discussion every N posts
<joepie91>
seems like a relatively easy and cheap way to me to make the discussion more structured
<gchristensen>
+1
<joepie91>
and avoid the "not gonna read all that" problem which invariably results in endless repeats of the same arguments and counterpoints
<cole-h>
You should file an RFC and make that part of the RFC process
<cole-h>
:P
<joepie91>
my days only have 24 hours :D
<samueldr>
perfect, one hour per RFC
<joepie91>
I'm happy to make the suggestion, and even help think about some issues that might come up and how to address them... but championing an RFC or other proposal with a commitment to continue discussing and defending it, is something I can rarely afford :P
<emily>
joepie91: the good thing about the same few points coming up is that it helps cut off the fringe bikesheds the discussions gradually drift into >.>
<emily>
admittedly "consensus by drowning out" is not the ideal solution
<__monty__>
emily: I wasn't specifically targetting you, you were just the drop. Did you even read the RFC you're commenting on? ASciidoc is already included in the formats under consideration. So unlike Abathur you didn't even add anything.
<MichaelRaskin>
So instead we keep up with the basic truth that all formats are _obvuiysly_ unacceptable
<joepie91>
emily: dunno, I think it can have the opposite effect; drowning out legitimate nuanced criticisms, leaving room for pointless bikeshedding where it's easy to fire off an argument
<MichaelRaskin>
No fringe nikeshedding on details
<joepie91>
emily: (speaking as someone who regularly tries to make a nuanced criticism and gets drowned in kneejerk "counterpoints" from people who haven't read carefully, then having to spend 3 back-and-forths re-explaining the point to them, only to then have someone else *do the exact same thing*)
<{^_^}>
rfcs#64 (by Infinisil, 7 weeks ago, open): [RFC 0064] New Documentation Format for nixpkgs and NixOS
<samueldr>
emily: even if you did only rehash things other said, having one more voice is a net positive in the end
<cole-h>
yegortimoshenko: Thanks for reading between the lines :P
<__monty__>
I'm not mad. I made at least one comment on that thread I think's worth at least consideration though. But what's the point if people are just not gonna read the thread?
<emily>
(for bonus salty, I'll point out that endless bikeshedding and gatekeeping and stalling like this is why NixOS rarely significantly changes for the better and everyone just spends all their time defending the status quo)
<srk>
asciidoctor has some nasty dep chain leading to pango/mesa/gl stuff
<samueldr>
srk: that's a solvable implementation detail
<srk>
sure
<__monty__>
I would never ask anyone to delete a comment. Apparently I'm terrible person for sharing my frustration though so feel free to not read this.
<srk>
hm, if pandoc can do asciidoc
<emily>
srk: yeah, and pandoc's isn't great either... but probably solvable, and I'm not sure end-user machines need to be building the docs from source most of the time?
<srk>
do you even need asciidoctor?
<srk>
:)
<emily>
I think pandoc can only do asciidoc output rather than input
<emily>
sadly
<srk>
like pandoc isn't any better when it comes to deps
<emily>
fwiw my recollection dealing with the docbook xml toolchain is that it isn't exactly the most pleasant thing either
<emily>
but that was years ago
<samueldr>
emily: end-user machines build manual each generation
<emily>
since I've steadfastly avoided it since :V
<srk>
yeah, only output
<yegortimoshenko>
is it ruby deps that blow up closure size or is it something else?
<emily>
samueldr: hm, it really doesn't seem like they should have to
<samueldr>
man pages
<srk>
anything xml based/related isn't pleasant to work with
<emily>
samueldr: why can't they be built on hydra? they don't change with user config
<samueldr>
yegortimoshenko: there were some relatively trivial gains early on that q*liss slayed easily
<srk>
pandoc could be nice as you could support all inputs for docs and let ppl choose :D
<samueldr>
yegortimoshenko: not sure about the rest
<emily>
no, making nixos documentation be a polyglot of ten formatting languages that nobody knows how to contribute to each file of would not actually be an improvement :p
<MichaelRaskin>
srk: that's exactly what I proposed. _Of course_ some people actively did not like it
<samueldr>
emily: it does since you can add modules to the system, man configuration.nix lists new options
<samueldr>
or uh, I thought it did, just checked and it's missing my added options
<emily>
samueldr: oh hm
<emily>
well I have literally never used that and I'm surprised to hear it exists, for what it's worth
<samueldr>
at any rate, the build time closure is to be considered because of how it's used on end-user systems
<MichaelRaskin>
emily: a proposal to start by discussing what documentation we should have and how it should be structured and accessed, so that requirements can be discussed in context, did not gain traction
<srk>
emily: well you can force people to use one language and some people won't contribute because they don't want to learn yet another markup language. choosing the right one is hard as you can see :)
<joepie91>
surely a local build would only be necessary if there is actually custom modules necessitating as such?
<MichaelRaskin>
Existence of some of the documentation outputs is not too discoverable, indeed
<MichaelRaskin>
joepie91: there is definitely at least an option to provide the manual as a single properly crosslinked object
<srk>
btw is discoverability gonna get worse with flakes? like no-longer monolithic 'man configuration.nix'
<MichaelRaskin>
So the toolchain is neeeded
<emily>
yeah I think there are more problems with the current docs than just format
<emily>
I think nixpkgs + nixos documents should definitely be merged
<emily>
and nix manual should probably be transcluded in it too
<__monty__>
srk: nixpkgs isn't moving away from the monorepo even if flakes become a thing. (Simply repeating what I've seen repeated many times.)
<MichaelRaskin>
There are some optimistic ideas that we somehow keep entertaining
<srk>
cool, I guess many people like nixpkgs being a monorepo
<MichaelRaskin>
We can afford not to do Turing-complete-by-requirements stuff written in Nix
<MichaelRaskin>
We can split Nixpkgs monorepo in a usable way
<MichaelRaskin>
I guess some more
<emily>
I don't think there's much benefit to splitting it up
<joepie91>
srk: there's always going to be some people turned off by whatever you decide; the goal is to minimize that amount of people who you *want* to be involved :)
<__monty__>
emily: If you'd read the thread you could've answered that you disagree with the people who said they think the three projects should each choose their own documentation format and toolchain.
<joepie91>
in that sense, accessibility is IMO important
<MichaelRaskin>
Oh, right, there is a format for writing documentation that will help things even without a discussion how this documentation is to be used
<srk>
joepie91: :)
<yegortimoshenko>
monty: oh come on
<emily>
__monty__: I have no idea what to say to anyone who thinks nixpkgs and nixos should use different document formats... please leave me alone.
<joepie91>
(a highly technical and specialized format means that the people willing to contribute to docs are gonna be mostly people who are very experienced and not in touch with the issues that beginners run into)
<emily>
I think a grand project to rewrite and restructure the entire documentation from scratch would never work
<srk>
is there a way to quickly render standalone dockbook page?
<emily>
so the fact that asciidoc could be slotted in "relatively" simply and by definition can convert everything in the current docs is a big advantage
<srk>
.. pandoc
<MichaelRaskin>
Rewriting definitely won't work
<emily>
srk: pandoc --standalone --out html or something?
<__monty__>
MichaelRaskin: The post about 4 different kinds of documentation and their different goals and requirements has been brought up multiple times in the RFC discussion. People are not neglecting your concerns. The scope of the RFC is limited to just the format because that's been identified as a painpoint and someone stepped up to try to get that part of the problem resolved. Thanks to infinisil for
<MichaelRaskin>
Autoconversion and moving around can
<__monty__>
doing this.
<MichaelRaskin>
I am not even concerned anymore…
<srk>
<meta markup="markdown"># woo</meta>
<MichaelRaskin>
I feel like Nix manual is safe, and the rest is not getting any large changes anyway, whatever format it gets converted to
<srk>
joepie91: latex!
<yegortimoshenko>
srk: well an option sure, but in that case markdown is just an opaque html blob
<srk>
indeed
<srk>
yegortimoshenko: but you can pandoc markdown -> docbook :D
<srk>
needs more isomorphism
<yegortimoshenko>
haha
<yegortimoshenko>
well, i guess the fundamental problem is that markdown doesn't feature docbook semantics, and it won't try to infer those, either
<__monty__>
srk: Have you worked with pandoc a lot? As has been raised in the discussion before, problem with pandoc is that it only supports a subset of most formats. reStructuredText's custom custom directives for example.
<yegortimoshenko>
s/and it/and pandoc/
<__monty__>
This is why there's been requests for demos btw. People can go on at length about what they like but without some sort of demonstration others won't be convinced.
<srk>
__monty__: not a lot, I do use it frequently to render stuff to html/pdf. sure you can't convert all stuff perfectly, thus the lack isomorphisms between formats
<emily>
ok, I read all the comments. I wasted a bunch of time and learned nothing. my least favourite was the one where someone bikeshedded over whether an "average machine" was a MacBook Air or a ThinkPad T420.
<srk>
tomorrow I'll be able to man configuration.nix \o/
<__monty__>
Imo it's not redundant to consider what hardware the docs should build on. (I don't personally agree with the time they appear to have settled on.) People talk a lot about being inclusive but it seems like nobody cares about all the people in the world that are stuck on old hardware.
<srk>
I'm switching from x86 desktop to armv7 laptop, lots of fun
<srk>
still need x86 for compiling haskell and browser
<MichaelRaskin>
ARMv7, not even 64-bit?
<MichaelRaskin>
Indeed, !!FUN!!
<srk>
TIL you can x11spice & nix-shell -p spice-gtk --run spicy
<srk>
my armv7l browser
<srk>
(meanwhile learning w3m)
<MichaelRaskin>
Not links?
<srk>
no, w3m is way better
<srk>
it can even do images provided the planets are aligned correctly
<emily>
links can do images too, it has a graphical mode, fwiw
<MichaelRaskin>
links -g can do images without special planet alignment
<emily>
(I don't use text mode browsing enough to say which is really best)
<srk>
nice
<emily>
links2 != elinks
<emily>
(dunno which of those is best, either)
<MichaelRaskin>
My use of text-mode browsing is with a custom HTML-to-text convertor and Vim
<srk>
maybe if I disable transparency in my terms it would work
<srk>
but I like them transparent
<emily>
srk: also, not even aarch64? don't die
<MichaelRaskin>
links2 is probably more alive; elinks has _some_ nicer features than links2
<emily>
srk: btw, is ivory tower your project? really cool if so
<gchristensen>
oh cool
<gchristensen>
emily: does that use any of drew devault's software?
drakonis has quit [Ping timeout: 240 seconds]
<srk>
emily: not really, was created by Galois (and kind-of abandoned). I'm just adding hardware support and general usability fixes/stuff/docs/drivers
<srk>
I've tried sway but switched to i3 :D would prefer xmonad/waymonad but no ghc /o\
<gchristensen>
on aarch64?
<srk>
armv7
<gchristensen>
ah
<emily>
srk: hmm, galois dropping releases of things and then abandoning them sure seems like a pattern x.x
<emily>
gchristensen: does what?
<srk>
I will try to fix it at some point but it takes way too long time
<emily>
sway is ddevault's
<emily>
paperwm isn't
<emily>
I would ask why it's a criterion, but... ngl "not maintained by ddevault" was a distinct pro in favour of switching
<gchristensen>
emily: yeah, but like wlroots etc
<yegortimoshenko>
gnome doesn't use wlroots afaik
<srk>
emily: I don't mind really. the tech is pretty cool and not that hard to maintain, I now routinely port it to recent GHCs, it's way easier with nix now
<gchristensen>
cool
<yegortimoshenko>
paperwm is just a tiny gnome shell addon (i also use gnome + paperwm and swear by it)
<viric>
what is good in an armv7 laptop?
<gchristensen>
nice
<viric>
^ srk
<emily>
yeah, gnome is an independent compositor implementation
<gchristensen>
yeah, just making sure I guess
<emily>
that some people (e.g. mpv devs) think is shitty, so take that for what you will
<srk>
viric: what do you mean? why do I use it?
<emily>
but i've had less problems than with sway in practice
<yorick>
gchristensen: are you looking for somethign that doesn't use DDV software? :D
<yorick>
(had another sway crash today :/)
<viric>
srk: I mean I can understand the complications of an armv7 laptop. What are the benefits? :)
<gchristensen>
yorick: not exactly, but I do somewhat begrudginlgy use sway
<gchristensen>
yorick: did you use kexec, and did Sway's devs intend for it to crash? :P
<srk>
viric: it's blob free laptop. also native environment for embedded arm development
<yorick>
gchristensen: no, it's a use-after-free this time
<viric>
srk: good enough.
<gchristensen>
I have a ticket which has been stuck in my craw for almost a year now
<srk>
viric: some fun features like spartan 6 connected directly to imx6q, hd ips display, usb hub control, 4 uarts..
<viric>
a spartan6 free to use?
<srk>
well with proprietary toolchain
<gchristensen>
yorick: surely I've linked you to it already
<yorick>
gchristensen: the 'exec' thing?
<srk>
viric: with its own dedicated sdram
<gchristensen>
yea
<gchristensen>
lol
<viric>
srk: I mean available for whatever the user wants
<srk>
yes
<emily>
gchristensen: also, if you're on a laptop, imo once you get used to the three-finger workspace/window scrolling gestures you'll never go back
drakonis has joined #nixos-chat
<gchristensen>
oooh
<emily>
first time i've actively felt like using touchpad gestures since i moved off macs
<gchristensen>
emily: doesn't seem like paperwm is packaged, though?
<__monty__>
Why are people either religious adherents or not-a-fan of Drew Devault?
<yorick>
the 3-finger gestures work well with sway+gebaard
<emily>
gchristensen: no, but it's easy to add yourself. I just use it from git as a niv dependency
<gchristensen>
cool
<yorick>
__monty__: he's very polarizing
<srk>
viric: it has high speed connector available for extension boards, there's gpio, rom emulator, SDR board. they even built an oscilloscope for it
<gchristensen>
__monty__: well, your "religious adherent" bit seems to capture my problem with im
<gchristensen>
__monty__: well, your "religious adherent" bit seems to capture my problem with him*
<viric>
srk: I was building the nixpkgs cross compiler for armv7; I'm checking the assembly produced by the atomic builtins and so on
<emily>
gchristensen: (the patch isn't important, just some personal tweaks that I should probably upstream)
<gchristensen>
nice
<emily>
I believe the paperwm dev uses nixos, so :)
<gchristensen>
nice
<srk>
viric: like comparing to native toolchain?
<__monty__>
yorick: That's what I gather but I wonder why? SourceHut seems like a cool take on forges. And I kinda wish the best to anyone making a living off of foss work without being beholden to a company, failure would just give the foss detractors more ammo.
<gchristensen>
emily: what is your Twitter handle?
<viric>
srk: no no; I'm curious how that's different from armv6
<emily>
gchristensen: @lambdakitten
<viric>
maybe there are not armv6 multicore.
<gchristensen>
thanks :) I tried @lambda... and @kitten... but it didn't come up
<srk>
I see. I'm no expert when it comes to arm assembly
<viric>
srk: neither me. But ARM requires proper use of the acquire/release atomic semantics or will fail (unlike x869
<viric>
(x86)
<srk>
uuh
<yorick>
__monty__: I desperately want the software to succeed, but my personal problem is that he's very vicious in convincing people to join his cause, and I don't think using or contributing to his software is fun for many of the people doing it
<viric>
srk: that's for writing lock-free code or similar. mutex work everywhere. :)
<yorick>
__monty__: I'm mostly using sway because if I don't move to wayland then wayland will move on without the things I use and in a few years I'll be forced to gnome
<yorick>
but after 1.5 years of using it it still feels more like beta-testing, with weekly undebuggable crashes
<yorick>
and even more "I wonder what will break with this update / if this software will work at all" with every update than just regular nixos or linux
<srk>
viric: interesting. I'm only doing single core stuff on small embedded ICs. was adding few __dsbs recently to overcome some cache issues
<yorick>
xwayland still isn't usable for me and nobody capable is interested in fixing it, and I keep getting the feeling it is because nobody will port anything to wayland if xwayland works well
<emily>
he also likes WONTFIX-closing legitimate issues
<emily>
so I have very little faith that it will ever stop feeling like beta software, if he's uninterested in actually tracking the problems
<srk>
I kind-of like being stuck with modesetting X with no browser or video player. ideal distraction-free desktop with only terms
<emily>
srk: careful, or you'll become one of those people who uses the linux console for everything
<yorick>
mpv supports that
<samueldr>
srk: I wonder if netsurf builds readily for armv7
<srk>
emily: I'm already there for like 90% of stuff
<nvd>
Heh, I remember I went about a year not being able to do anything graphical on my laptop at all (misconfigured firmware)
<srk>
samueldr: will try, deps look fine
<samueldr>
it's a project I'm on the lookout for as they have been working for years, and are showing good continuing develpoment
<emily>
srk: the thing with the linux console is that it's not even a good implementation of a terminal :p
<emily>
garbage unicode support, garbage ansi sequence support
<emily>
bad performance
<srk>
well performance is the reason I'm stuck with urxvt
<srk>
unicode is fine, I don't need all the emojis and weird chars in the galaxy anyway
<samueldr>
though it's literally limited to 256 codepages
<samueldr>
which is not fine to any stretch
<samueldr>
256 chars codepages*
<samueldr>
[citation needed], working from flawed memory
<srk>
samueldr: is it nix-shell -p netsurf.browser?
<samueldr>
no, by default it builds for framebuffer use only
<samueldr>
it's something I wanted to review, all of the netsurf stuff to make it more ergonomic and having good defaults
<samueldr>
not saying "gtk as a default", but maybt "netsurf-gtk" as an attribute
<samueldr>
since right now netsurf with gtk isn't built by hydra
<sphalerite>
if I want to run a UDP service that's reachable via the internet on my laptop, but have privacy extensions enabled — does anyone know if I can get it to reply to incoming connections with the right (non-privacy-extension) source address, and how?
<__monty__>
srk: Emoji are a good litmus test for wide characters though. Which close to half the world kinda needs afaict?
<srk>
__monty__: not sure about that, would love to have ligatures for haskell tho :)
<gchristensen>
is there something like which mushes tmux + tail? tails a bunch of files, but automatically panes them?
<samueldr>
there's no one to tell me my ideas are bad when they are
<worldofpeace>
yegortimoshenko++
<{^_^}>
yegortimoshenko's karma got increased to 1
* yegortimoshenko
plays "price is right losing horn"
<cole-h>
yegortimoshenko++ LOL
<{^_^}>
yegortimoshenko's karma got increased to 2
<yegortimoshenko>
haha thank you :p
<worldofpeace>
samueldr: I bet it can be. But, what really is a bad idea? A result you can produce will always be different than what someone else could come up with. But maybe things in a earlier stage can be like this.
<samueldr>
worldofpeace: depends
<samueldr>
but you're also right
<samueldr>
the joke is: that way I could have told them "no" :)
<worldofpeace>
🤣 I believe it is almost always easier to be that person samueldr
abathur has joined #nixos-chat
abathur has quit [Read error: Connection reset by peer]