<Shados>
pie_: Most products now do basically suck. There's some really great stuff out there, but such things are lost among the ocean of crap made for planned obsolesence...
drakonis has joined #nixos-chat
drakonis_ has quit [Ping timeout: 248 seconds]
<colemickens>
gchristensen: are you using stable firefox from nixpkgs or firefox nightly? If you're on Stable, is it usable on Wayland these days?
<colemickens>
I'm at my wits end with Nightly and I finally was impacted by their ongoing shitshow and am going to take some time and sort it out. Figured I'd go with stable if its tolerable
<pie_>
colemickens, this worked for me if you mean the addon stuff v
<{^_^}>
#60916 (by deliciouslytyped, 9 hours ago, open): Firefox 1548973 Hotfix (addons disabled due to expired intermediate certificate)
<colemickens>
Oh ouch, I hadn't actually realized stable builds still needed hand holding.
<pie_>
idk
<pie_>
i went to bed at 7 am, "meh not impacted yet"
<colemickens>
I'm more just trying to decide if I should start a new profile with nightly or stable. I was mostly on Nightly for Wayland support. But that was moons ago. I think Stable is there.
<colemickens>
Once it finally happened to me, it removed any doubt about whether this will be disasterous for Mozilla.
<pie_>
came back at ?? oclock, "oh there go all my addons" "meh i need to enable user studies? fine." "5 minute shave passed why isnt this doing anything? *pie checks for an update timer* *pie decreases timer to 20 seconds* *nothing happens* *pie turns off user studies,pokes through user studies api - oh whats THIS XPI link?...W00T werkz 4 me*"
<colemickens>
I don't think you're alone.
<pie_>
tbh i feel like they should have just made this a supported option, but then if you break something again you have nonpowerusers installing some kind of half broken work around that you have to fix later so ehhhh
<pie_>
but maybe thats just me trying to justify what theyre doing :P
<pie_>
i would have preferred to just have the xpi to begin with
<pie_>
i wonder if you can see a sudden surge in ad revenue :D
<colemickens>
I'm more curious how many Tor Browsers users were exposed as a result of this.
<colemickens>
I guess it would have to be a combination of TBB users, landed in the slice of users affected at that moment, browsing a site with malicious javascript, combined with some sort of vuln/zero-day..... but still. It must be startling to see NoScript disappear and JS start executing on the next page load as a Tor user...
drakonis has quit [Ping timeout: 264 seconds]
<pie_>
colemickens, can you guess what my first thought was when i heard of this?
<pie_>
"oh no what about the tor users...."
<colemickens>
I mean, I didn't notice they were gone, relogged into Facebook.
<pie_>
now im feeling kind of funny how ages ago when i used tor i wondered about disabling javascript in the browser as well, not just noscript
<colemickens>
Now that the FB Container extension is reinstalled, I'm pretty sure my original FB cookies are still in the un-isolated store. So I'm potentially even leaking some info I wasn't before. Christ, I haven't even gotten all of them reinstalled yet because I don't have a list. Just, wow.
<colemickens>
pie_: yes! in retrospect, I hope they're now taking a look at failing shut.
<pie_>
i guess the lesson to learn is that you shouldnt depend on optional components for security? critical component failure should cause fail-fast of functionality
<pie_>
at least thats my kne-ejerk reaction, not sure if its the smartest thing that one could think of
<colemickens>
idk, not a big user and I don't know that much about browser security models, but it seems like disabling JS would be prudent for TBB
<pie_>
colemickens, problem is i guess if you dont need to be super secure you might want to enable javascript in some places
drakonis has joined #nixos-chat
<colemickens>
I mean, they have been making an effort to upstream TBB bits into Firefox and some of its already landed. I'd love to see a Firefox-mainline "toggle" that will go hardcode - disable any non-origin cookies, disable WebGL/WebRTC, disable JS, disable SVG, resistFingerprinting=true, etc.
<pie_>
yeah
<colemickens>
pie_: I think we're too "power user" but I'd still rather have FF default no-JS and allow me to selectively enable. Which is the opposite of the current situation.
<pie_>
put some substance behind the privacy posturing :p
<colemickens>
or at least have a mode where I can opt-into such strenuous security/privacy.
<pie_>
yeah, think is im probably going to miss a setting here or there if i have to set it up myself
<colemickens>
I mean, if they *really* cared, they'd kill the Pocket integration, release the Pocket server code, apologize for having a "save" button that uploaded screenshots to their servers, stop running experiments with advertising companies and agencies, etc.
<colemickens>
But, now I'm just crying into my coffee.
<pie_>
well i dont know how to run a project so idk about organizing features like this
<pie_>
privacy is a moving target if you dont have something like a capability system (random conjecture)
<pie_>
well i guess addons have a capability system, but thats not the browser innards
<pie_>
:c same <colemickens> But, now I'm just crying into my coffee.
<pie_>
on the other hand you have to fund a company, but shit
* colemickens
nods
<pie_>
theres probably tons of great work in firefox, theres just some...weird stuff too that really leaves me going "ehhhhhhhhh"
<pie_>
"please at least you guys..." :P
* colemickens
wishes he knew a Mozilla employee to get the inside pulse
<colemickens>
lol, yes, 100% agree with that too!
<pie_>
now im getting really depressed lol
<colemickens>
so, so, so many UX warts and inconsistencies and that's not even to mention Linux specific UI problems
<colemickens>
I fear what will happen when Chromium works well on Wayland. I must stay strong.
<pie_>
eh i havent had too much ux pain, but i havent used any other browsers
<pie_>
i much preferred native extensions ux-wise though
<pie_>
also i still dont understand the addons.mozilla.org overhaul
<pie_>
i hate it so so much and it feels crippled
<pie_>
by dont understand i mean i never spent the time to find out why they did tht
<colemickens>
I've recently discovered that the only way to really pop a persistent popup from an extention (which is necessary in a number of scenarios) is to inject into the page DOM... which, as you can imagine, works well 90% of the time and is a disaster the other 10%.
<pie_>
anyway im trying to figure out how to get a demotool to connect my microphone and i have no idea...i dont like poking around in C code :c
<pie_>
colemickens, nyeh.
<colemickens>
sounds like possibly less fun than musing about Firefox. heh, good luck!
<colemickens>
Many users won't get it... because they disabled that "mechanism"...
<colemickens>
because Mozilla used it to install an extension in users' browsers as part of a campaign with the TV show Mr Robot.
<colemickens>
Couldn't make this stuff up.
<pie_>
the uhh
<pie_>
n somethingorother?
<pie_>
normandy mechanism
<pie_>
not even sure what it is, something about pushing updates to subsets of users for testing i think? not even sure
drakonis_ has quit [Ping timeout: 252 seconds]
<pie_>
there seems to be like 20 mechanisms for this kind of stuff
<pie_>
i was like wtf is this
<pie_>
smh, browsers
<pie_>
burn the marketing department
<jasongrossman>
colemickens: Wow.
<jasongrossman>
This makes me hope palemoon takes off. (Unlikely, I know.)
<pie_>
colemickens, i wouldnt care, id disable it, but like, yeah i could understand how some people would flip their s** over that, and now here we are
<pie_>
colemickens, theres an issue on bugzilla with people complaining about not getting the update while getting others, and generally the fix not working for some reason
<pie_>
dont know why yet though
<colemickens>
lol the behavior in browsers seems to differ too
<colemickens>
LMAO
<colemickens>
I think Firefox Sync kicked in and synced the "1 extension installed" over the list of previously installed extension, so shits just getting clobbered everywhere now regardless of patch or not.
<colemickens>
Sorta makes me want to donate to Mozilla though, tbh. Or send a 6 pack over tonight :(
<pie_>
hm yeah
<pie_>
anyone using nixos near mozilla hq? :P nixos sends beers :P
<pie_>
man IDK how to c++ ...
<pie_>
maybe i should just try inspecting these values in gdb
jasongrossman has quit [Ping timeout: 252 seconds]
jasongrossman has joined #nixos-chat
jasongrossman has quit [Client Quit]
jasongrossman has joined #nixos-chat
<colemickens>
Is there any way to reuse flatpak's containerization/isolation, but without giving up Nix / nixpkgs?
<colemickens>
I guess I don't even care about Flatpak that much though, chromium OS is doing more interesting isolationy things.
<aanderse>
i used firejail for a while
<aanderse>
very configurable
drakonis_ has joined #nixos-chat
drakonis has quit [Ping timeout: 248 seconds]
<pie_>
colemickens, whats chromeos doing
<colemickens>
Linux apps run in a container inside a VM
<colemickens>
Or vice versa or both, can't remember
<colemickens>
But VM and container isolation, hardware acclerated graphics work, they have a special virtual Wayland device...
<pie_>
damn
<colemickens>
I can type out more thoughts on a real keyboard later.
<pie_>
thats sounds pretty good
<pie_>
damn google having all those resources :P
<colemickens>
I think that often
drakonis_ has quit [Ping timeout: 246 seconds]
drakonis_ has joined #nixos-chat
drakonis_ has quit [Ping timeout: 268 seconds]
drakonis_ has joined #nixos-chat
drakonis has joined #nixos-chat
drakonis_ has quit [Ping timeout: 258 seconds]
drakonis_ has joined #nixos-chat
drakonis has quit [Quit: Leaving]
drakonis has joined #nixos-chat
drakonis_ has quit [Ping timeout: 258 seconds]
drakonis1 has joined #nixos-chat
drakonis_ has joined #nixos-chat
drakonis has quit [Ping timeout: 264 seconds]
drakonis1 has quit [Quit: WeeChat 2.4]
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.2)]
endformationage has quit [Ping timeout: 252 seconds]
<MichaelRaskin>
Should we just switch to shipping Developer Edition by default?
<MichaelRaskin>
I wonder if not shipping the official branding would allow us to switch off signature verification
jasongrossman has joined #nixos-chat
<joepie91>
[02:48] <pie_> not even sure what it is, something about pushing updates to subsets of users for testing i think? not even sure
<joepie91>
yes
<joepie91>
basically, it's a mechanism that allows Mozilla to run arbitrary code without pushing a full FF update, to a select part (eg. percentage) of the userbase
<joepie91>
_on paper_ that is meant for A/B-test-type stuff
<joepie91>
in practice, it seems to be getting used for marketing
<joepie91>
MichaelRaskin: you already can, in stable
<joepie91>
for some reason the stable Firefox that we ship lets you toggle the signing flag
<MichaelRaskin>
Then why are people discussing the complicated details of making Normandy work in Nixpkgs issue tracker, I wonder
<joepie91>
MichaelRaskin: because it doesn't require disabling signature verification for all extensions :)
<MichaelRaskin>
I would say that installing new extensions is unwise right now anyway…
<joepie91>
MichaelRaskin: this also affects automatic updates of extensions
<joepie91>
the risk is admittedly small, but bigger than when using the normandy xpi
<MichaelRaskin>
Ahahah
<MichaelRaskin>
I tried running valgrind bash
<MichaelRaskin>
Aborted
<MichaelRaskin>
It probably is Nixpkgs-specific…
__monty__ has joined #nixos-chat
<manveru>
hmm, my firefox has never stopped working, afaict...
<MichaelRaskin>
Firefox doesn't stop working, that's kind of the problem — extensions do
<manveru>
they do?
<MichaelRaskin>
Unless you already turned off signature checks
<manveru>
not that i can think of...
<manveru>
guess i'm lucky then :)
<__monty__>
Talking about yesterday's fuckup?
<MichaelRaskin>
Well, my setup is not affected at all, it is all kill-all-UI per-page Firefox instances
<MichaelRaskin>
With domain-blocking in Squid
<MichaelRaskin>
And no network access except Squid and DNS
<__monty__>
Are adblocking proxies the future?
<MichaelRaskin>
Well, I should configure HTTPS MITM in Squid, but I don't know a truly good way of certificate error handling of that
<MichaelRaskin>
Majority of web content I just read in Vim
<etu>
__monty__: DNS ad-blocking has been around for a long while, but I usually prefer an inline adblocker because that can hide things the DNS-based ones can't. Like ads embedded in videos. etc.
<__monty__>
etu: I know, but the APIs you need for those seem to be constantly under attack.
<MichaelRaskin>
JS off by default helps, but dumping to text without going through a true browser helps even more
<etu>
MichaelRaskin: And pass it through email as well ;)
<MichaelRaskin>
Why bother?
<__monty__>
I think there's a currently open issue by ublock origin's author on mozilla's tracker going in depth about some seemingly innocuous changes.
<MichaelRaskin>
Next level: WebDriver-based ad blocking.
<etu>
MichaelRaskin: Just to be like RMS :p
<MichaelRaskin>
I use a laptop with WiFi…
<etu>
MichaelRaskin: Just kidding :)
<MichaelRaskin>
We should try adding valgrind tests — Abort in bash is kind of surprising
jD91mZM2 has quit [Remote host closed the connection]
jD91mZM2 has joined #nixos-chat
<aanderse>
i met rms once
<aanderse>
it was glorious
<aanderse>
i got to shake his hand *after* he gave himself a foot massage with some sort of ointment
<aanderse>
ha ha ha
<pie_>
as you said, glorious
<aanderse>
mhm mhm
<pie_>
too many things run on linux these days lol
<pie_>
im having trouble thinking of things to package for my wine packages
<MichaelRaskin>
Orbiter!
<pie_>
ive also been kind of lazy to make some more useful infrastructure though :/
<pie_>
MichaelRaskin, oh thats an ide
<pie_>
a
<pie_>
i still need to add some kind of management for mutable profiles so i stop deleting my saves every time i run something different
<pie_>
coding library functions is tiring because i actually have to think xD
aanderse has quit []
__monty__ has quit [Quit: leaving]
<gchristensen>
colemickens: stable works, yeah
<gchristensen>
let's base stdenv off powershell
<tilpner>
gchristensen--
<tilpner>
D:
aanderse has joined #nixos-chat
<manveru>
pie_: irfanview :)
<manveru>
hmm, automating updates for ruby gems should be pretty straightforward...
<gchristensen>
manveru: yeah? might be worth collaborating with ryantm on that -- or I run a number of regular nixos crons and could add it to that list :P
<gchristensen>
sounds great at any case :)
<manveru>
yeah, i've done just about enough that i'm tired of doing it manually now :)
<manveru>
and some of them were vastly out of date
<manveru>
but i think there's a simple way to do it for about 80% of packages
<pie_>
hey, thats a big chunk *shrug* :D
<manveru>
just have to understand the code of nixpkgs-update :)
<MichaelRaskin>
Hm. I cannot find out how gcc reacts to valgrind, because bash fails first
<pie_>
MichaelRaskin, doh x)
Synthetica has joined #nixos-chat
<MichaelRaskin>
Oh nice, Debian valgrind on Debian bash on a Debian system still gives «Aborted»
<joepie91>
sounds like you found a bug :D
<MichaelRaskin>
It looks like Bash punts the bug towards Valgrind
<MichaelRaskin>
Ouch ouch ouch. They build their own malloc/free and sometimes these have separate names and sometimes free is called free
<joepie91>
who, bash or valgrind?
<MichaelRaskin>
Bash
<MichaelRaskin>
Well, Valgrind, _also_ have their own malloc/free
<MichaelRaskin>
But they consistently try to replace glibc malloc and free
<pie_>
did you know you can paste images (screenshots) into github
<pie_>
issues.
<pie_>
if they added this recently idk who is on the microsoft github team but they are doing a good job impressing me
__monty__ has joined #nixos-chat
<manveru>
it's been there for as long as i can remember
<gchristensen>
it was a fun goal -- absolute minimal terminal and delegate the rest to screen / tmux, and fun watching it slide away from that
<__monty__>
Why do people like alacritty? I haven't been able to figure it out. If GPU terminal is what you want kitty seemed nicer to me overally.
<__monty__>
st still takes that approach I think?
<samueldr>
in my case, when I switched, alacritty was the only terminal that worked for my criterias, I don't recally what the issue was with kitty though
<samueldr>
but I wasn't seeking a gpu-accelerated terminal, just _a_ terminal that felt right
<__monty__>
Main issue with kitty is the maintainer : ) He's pretty abrasive.
<samueldr>
next round of "let's fine the least worst terminal" I'll maybe try and take notes :)
<samueldr>
let's find*
<MichaelRaskin>
I mostly just use rxvt-unicode
<__monty__>
Tbh, gnome-terminal seemed hard to beat in my testing.
<MichaelRaskin>
Sometimes mlterm
<__monty__>
I prefer terminology and kitty for reasons though.
<infinisil>
Yeah I also tried out a couple terminals, including konsole and sakura, but only alacritty seemed to work decently well (out of the ones I've tried)
<samueldr>
gnome-terminal was good, but they started removing all options, and I don't remember which at the time, a specific option I used was removed :/
<gchristensen>
I've been fairly happy with sakura, but I don't like that it fairly regularly segfaults
<samueldr>
segfaults with sakura was an issue here too
<__monty__>
Why do you stubbornly stick with it, gchristensen?
<samueldr>
"least worst" :)
<gchristensen>
it seems to have been the least worst of the other ones I've tried
<__monty__>
Anyone given plain xterm a serious spin?
<gchristensen>
yeah
<samueldr>
though alacritty, because I'm always backing the terminals with tmux, wasn't missing features; don't need scrollback nor search when tmux handles it
<MichaelRaskin>
I remember xterm being harder to configure for more than one font than rxvt-unicode
<gchristensen>
xterm was a bit annoying, as I'd hit control and be dumped in this wild menu of things to do
<__monty__>
urxvt doesn't handle fonts well though. Can't remember what it was but it was some spacing issue.
<MichaelRaskin>
I sometimes use screen scrollback and sometimes urxvt scrollback
<samueldr>
riiight, one big issue is how they handle fallback fonts, at the time I had "funny" unicode characters (you know, just characters?) and some would render really badly in some terminals
<MichaelRaskin>
Yeah, urxvt doesn't like full-width stuff
<joepie91>
Matrix is pretty interesting and has a reasonably solid design, but I can't help but feel a bit iffy about the fact that "X violates spec / spec is unclear" comments now make up like 10% of my comments by volume
<joepie91>
it definitely needs more work
<__monty__>
You implementing a matrix client?
<joepie91>
__monty__: homeserver
<joepie91>
judging from the issues I've been running into and my conversations with Matrix people, seemingly the first 'cleanroom' implementation
<joepie91>
(ie. implemented from spec, not from Synapse source)
<joepie91>
so this project has involuntarily sorta changed into a spec review :)
<MichaelRaskin>
I think there were a few cleanroom-ish attempts?
<tilpner>
joepie91: Ruma tries to implement by-spec too
<MichaelRaskin>
Ruma?
<MichaelRaskin>
Yes
<joepie91>
makes me wonder how they haven't run into these issues then
<tilpner>
They've had a long list of things blocking progress
<joepie91>
yeah but like, I've literally only started on login/registration and I've already filed like 7 different issues
<tilpner>
Some spec, some Rust, and they've just been waiting for those to go away
<tilpner>
So no real progress
<joepie91>
one of them being Riot violating the spec
<joepie91>
the other 6 being spec ambiguities / missing info / contradictions / etc.
<joepie91>
oh, sorry, other 5, not other 6
<tilpner>
They have opened issues, but I don't have any numbers
<joepie91>
hm, right
<tilpner>
Any other goals than "from spec"?
<joepie91>
for my project? not particularly, I mostly want to see how difficult it is
<joepie91>
though I wouldn't mind a hackable, non-resource-hoggy homeserver implementation :)
<tilpner>
I know Synapse is resource-hoggy, but I don't know exactly why. Have you looked into this in more detail than "it's Python and Python is very dynamic"?
<joepie91>
not really; I tend to avoid Python
<tilpner>
My intuition fails me when trying to estimate how node compares to python in resources-hoggy-ness
<joepie91>
I've had enough troubles with its dependency management and ecosystem documentation to last me a lifetime
<tilpner>
They seem about equally dynamic, but I don't know what magic v8 can do wrt. memory savings
<tilpner>
IIRC pypy makes the slots trick of cpython unnecessary, so node probably does at least that too
<joepie91>
tilpner: JS in Node.js is considerably faster than Python (in the standard runtime); async I/O support is much much better; and I believe that the memory usage for similar tasks is a bit lower for reasons that never quite became clear to me
<joepie91>
that having been said, V8 is a bit aggressive in its heap allocations
<joepie91>
so sometimes needs to be... told... not to do that
<joepie91>
:)
<tilpner>
Faster, sure, no argument from me, it's got JIT compilation, CPython doesn't
* tilpner
was just unsure about the memory aspect
<joepie91>
I'm a bit fuzzy on that as well
<joepie91>
but memory usage is not my main concern
<joepie91>
I'm more concerned about slowness under load
<tilpner>
You'll probably want some sort of integration testing setup at some point
<__monty__>
Implementing in rust?
<tilpner>
__monty__: joepie91 seems to be using node
<joepie91>
correct
<tilpner>
And if you have that setup, you can probably simulate a lot of activity fairly easily (though not as easily federation with huge histories)
<joepie91>
tilpner: also, sytest exists :)
<__monty__>
Oh.
<tilpner>
Oh right, I knew there was something, just forgot the name, and didn't know if you could steal that
<joepie91>
I've inquired, and yep
<joepie91>
previously Synapse-specific
<joepie91>
now supposedly usable with any HS impl
<tilpner>
(Is it still clean-room if you use Synapses integration tests?)
<tilpner>
(Not like that matters as much here)
<tilpner>
... and of course it's perl
<joepie91>
eh, this isn't strictly cleanroom anyway
<joepie91>
given that I'm following somebody elses spec
<joepie91>
:)
<joepie91>
at least, as I understand the term
<gchristensen>
would you have to, what, capture pcaps for it to be clean-room?
<tilpner>
True, there's probably no copyright infringement you have to worry about
<sphalerite>
AFAIU the idea of cleanroom is to have one person or team observe the behvaiour of the original system and write a spec from that, which is given to a second person/team to implement
<joepie91>
right
<sphalerite>
so that the implementers never had any contact iwth the original system
<sphalerite>
in this case it's not so much wanting to avoid the original code for legal reasons as wanting to use the spec because of unwillingness to touch the original software :p
<samueldr>
though looks like the clean room wikipedia entry says "aka chinese wall"
<sphalerite>
(or to test the spec?)
<joepie91>
sphalerite: mostly the latter
<joepie91>
I don't like importing technical debt
<joepie91>
so I'd rather help get the spec into shape :)
<gchristensen>
hrm. maybe alacritty is not quite the one for me. it doesn't seem to copy-paste, although it is really good at inserting "Error: target STRING not available"
<samueldr>
ah, copy/pasting in wayland I guess, as otherwise it works in X11
<gchristensen>
yeah. skimming the issue tracker it either does or did just exec xclip
<samueldr>
hm, elegantly inefficient way to handle that agnostically if it just execs
<gchristensen>
agnostic?
<samueldr>
if it's configurable
<gchristensen>
oh heh I don't think it does
<samueldr>
(and even if it isn't yet...) it means that there's no deep X11 logic for handling copy/pasting
<gchristensen>
I think it uses an external crate for copy/paste support
<gchristensen>
good thing I excel at dealing with recursive yak shaves
<sphalerite>
if I had a quotes file I'd stick that in
<samueldr>
oh, I think I remember a thing that peeved me off from some terms: I need the windows to *not* snap to cols/rows, and instead resize fluidly, it annoys me when resizing those kind of windows in awesome
<samueldr>
gchristensen: let me guess, you're looking at alacritty... and it's somehow related to... launching a rocket?
<gchristensen>
"Fix empty clipboard freezing Alacritty on X11" awesome :D
<gchristensen>
haha
<samueldr>
*thinking* would it be possible to do use an xclip compatible tool, that works with whatever wayland protocol for clipboards?
<samueldr>
it wouldn't be useful only for alacritty, I guess a bunch of instructions rely on xclip being available
<MichaelRaskin>
I know that Matrix misses a few clearly necessary things from the spec, but I want to wait until joupie91 fixes what's already there
<sphalerite>
samueldr: I'd expect that you can get awesome to ignore the increments stuff?
<samueldr>
thinking, among others, about git[hub|lab] instructions for ssh keys
<samueldr>
sphalerite: maybe
<sphalerite>
i3 seems to ignore them
<sphalerite>
or I've been using the right terminals
<samueldr>
but from the behaviour it exhibited I think it would leave unsightly gaps
<samueldr>
:0, I broke my graphical session when using it, but I got frecon started and logged into my account using it
<samueldr>
frecon being chromiumos' vt replacment, which draws to the framebuffer so it can handle hidpi better
<sphalerite>
lol
Church- has joined #nixos-chat
<Church->
Heya folks
<gchristensen>
cool, alacritty from master does copy/paste
<Church->
Heya gchristensen
<gchristensen>
hey Church-
<Church->
Hmm hey question how would I check if NixOS has a certain driver? Not at my laptop but I want to check that my asus zenscreen mb16ac works on Nix
<sphalerite>
Church-: you can have a look at the kernel config using `nix build nixpkgs.linux.configfile`
<sphalerite>
(and looking at the contents of `result`)
<Church->
Nod nod
<Church->
Thanks sphalerite
<sphalerite>
(that's an appropriate question for #nixos FWIW though ;) )
<sphalerite>
Also, are you the same Church as in that quote on bash.org about functional programming and Church and state? :p
<Church->
No?
<Church->
Link?
<Church->
I've got one on there about IOT cock-rings though
<Church->
Although that might be under aleph-
<Church->
sphalerite: After checking, yes I am.
<sphalerite>
:D
<manveru>
zimbatm: what do you think is a good place for a `make-bundler-app` kinda executable?
<manveru>
i'd like to add it to bundix somehow, but not sure...
<Church->
Heya manveru
<manveru>
Church-: yo
<manveru>
zimbatm: basically, i got a little tool now that can download a gemspec and generate the default.nix/Gemfile from it :)
* Church-
is munging data and setting up a media server
<manveru>
zimbatm: including all the executable names that are in the gemspec
<manveru>
unfortunately, for some reason the rubygems API doesn't included that information, so i actually have to download the gem to do it :(
<qyliss>
manveru: oh that's awesome
<manveru>
qyliss: wanna try? :)
<qyliss>
please!
<qyliss>
I think bundix would be a good place for it, fwiw
<manveru>
ideally the ERB part will then be replaced with nixer
<manveru>
well, mostly... :)
hedning_ has quit [Quit: hedning_]
<qyliss>
I've been looking into adding a devdoc output to Ruby today
<qyliss>
because I'm getting sick of having to build my own ruby to get ri working
<manveru>
hmm
<manveru>
the docSupport thing?
<qyliss>
yeha
<qyliss>
it's not simple though
<manveru>
i can imagine... nobody's used it for ages :)
<manveru>
but would be nice if we could have it... and it's not too large
<qyliss>
Yeah
<qyliss>
it's 50MB, so not worth including by default
<qyliss>
but I'd love to be able to opt-in without building my own ruby
<manveru>
i think i removed it originally?
<qyliss>
i haven't checked
<manveru>
because it made docker containers even larger :|
<manveru>
no, the flag has existed longer than that, it was just not used right
hedning_ has joined #nixos-chat
<manveru>
hmh, strange failure when i try to enable it
<qyliss>
really?
<qyliss>
it's worked perfectly for me forever
<samueldr>
oh, TIL about Nixer in bundix
<manveru>
qyliss: must've been some strange lorri failure, i think it builds it right now
<manveru>
qyliss: so... what's missing?
<manveru>
at least it includes core and stdlib
<qyliss>
nothing's missing, I'd just like it to be in its own output so hydra could build in
<manveru>
aye... that sounds tricky :|
<qyliss>
I think I can make it work
<qyliss>
And it's a significant pain point for me so I'm willing to put in th etmie
<manveru>
cool :)
<manveru>
i wouldn't mind it for my development either... sucks having to find the matching versions online every time
<qyliss>
yeah
<qyliss>
I think I can do something with a setup hook
__monty__ has quit [Quit: leaving]
<zimbatm>
manveru: it looks like a `bundix init` command
<zimbatm>
manveru: `bundix --init` should really be called `bundix --shell`
<zimbatm>
`bundix --type app --init <gem-name>`
<zimbatm>
maybe merge the shell.nix with the default.nix?
<manveru>
hmm
<manveru>
how would you merge them?
<zimbatm>
I'd like to establish some terminology
<zimbatm>
there is the in-project and out-project use of bundix
<zimbatm>
in-project is when you have a rails app and you want to manage the dependencies with nix
<zimbatm>
out-project is for example in nixpkgs, pulling existing projects and packaging them
<zimbatm>
then there are the lib and app modes for an app
<zimbatm>
the shell.nix is mainly useful in-project isn't it?
<manveru>
yeah
<zimbatm>
it's used to provide dev dependencies
<manveru>
well, it doesn't really differ much
<zimbatm>
why couldn't a bundlerApp not provide those
<zimbatm>
if that's the case then the default.nix is enough
<zimbatm>
maybe with an extra shellHook that setups some environment variables
<manveru>
i guess it could, since it has the env :)
<manveru>
anw, what i meant is that i rarely need a "dev" bundlerEnv anyway, since all the executables are wrapped by it, so having all those env vars in my shell doesn't do anything
<manveru>
so, unless you want to run some unrelated ruby version with gems for another ruby... i'm not sure how useful bundlerEnv is
<manveru>
it's still convenient, but i don't find it clean :)
<zimbatm>
yeah it's been bugging me too but haven't been doing enough ruby to properly think about it
<zimbatm>
at least in the context of nixpkgs, bundlerApp is really the only one that's need
<zimbatm>
needed
<manveru>
i think so, yeah
<manveru>
even for rails apps
<manveru>
because in the end you still run it via `rails` or `rake` or `thin` or whatever...
<zimbatm>
well during dev I want all gem bins to be one the PATH
<zimbatm>
s/one/on
<ajs124>
nix just gave me "error: stack overflow (possible infinite recursion)". I think I'll give up for today…
<manveru>
zimbatm: good point, bundlerApp doesn't do that :|
<zimbatm>
so maybe `bundix --mode dev` is a bit different
<zimbatm>
also the `default.nix` output might be not the same
<zimbatm>
in nixpkgs all that's needed is a function that takes the arguments
<zimbatm>
but in a project, the default.nix also need to import nixpkgs
<manveru>
true
<manveru>
generally i find it hard to combine much in shell.nix/default.nix
<manveru>
without having some third file
<manveru>
if it should be a traditional default.nix with only one derivation
<qyliss>
Could bundlerApp have an env attribute like bundlerEnv?
<qyliss>
I never use bundlerEnv directly, but I do use its env attribute for local development
<manveru>
hmm
<manveru>
all .env gives you is a modified irbrc?
<manveru>
can't say i've used it
<qyliss>
IIRC it gives you a wrapped Ruby that can load all your bundled gems?
<manveru>
yes
<manveru>
i use gems.wrappedRuby directly
<qyliss>
Ah
<qyliss>
well that works too
<manveru>
i don't think you can put gems.env in buildInputs of a mkShell?
<qyliss>
I don't know
<manveru>
so, doing a nix-shell on `bundlerEnv.env` works, and you can add buildInputs in bundlerEnv
<manveru>
env can't be in buildInputs though, so i generally don't do this
<manveru>
irb works without the modified irbrc
<qyliss>
I usually just overrideAttrs on bundlerEnv.env and add buildInputs in there
<qyliss>
But it's a bit annoying
<qyliss>
and it wouldn't work if anything else took bundlerEnv.env's approach
<manveru>
as usually, there's more than one way to do things :)
<manveru>
i think zimbatm is responsible for all those shell hooks
<manveru>
anw, i really want these three tasks to be streamlined: generate shell.nix and default.nix for new projects, generate a default.nix for a given gem, and update gemset.nix
<manveru>
i think adding subcommands for that to bundix might be the best way
<zimbatm>
maybe it's time to fork bundix, not worry about b/c
<zimbatm>
subcommands work too
<manveru>
i'd rather avoid forking... people are confused enough already :(
<qyliss>
I've been thinking about forking, or at least rewriting
<manveru>
bundix is already a twice rewritten fork of a fork :P
<qyliss>
I think there are better ways of doing lots of the things it does
<qyliss>
And I'm not sure those can be bolted on as new features without it being confusing
<manveru>
i'd love to know your ideas
<qyliss>
Once I've got Ruby 2.6 sorted in Nixpkgs I think I'll start looking at Bundix
<zimbatm>
it's been so hard to nail the developer experience
<zimbatm>
qyliss: if you have some fresh ideas, don't hold back!
<qyliss>
I do, but I think I really need to write them up rather than describing ad-hoc over IRC to really get it across
<manveru>
do we need a bundix channel? :)
<manveru>
or #nixos-ruby
<manveru>
to make it future-proof
<qyliss>
I'm up for that
<qyliss>
I think it would be really valuable, in fact
<qyliss>
gchristensen: channel request? ^^
<zimbatm>
is there anyone else than the 3 of us? (not sure I even count)
<manveru>
well, i registered it for now :)
<manveru>
maybe should've done it on matrix
<qyliss>
I'm not sure, put having a channel just for the three of us would still be useful
<qyliss>
And there are other people who are at least interested
<simpson>
To give a bystander's thoughts: I am interested in direct proportion to how much Ruby I use on a daily basis. Right now, very little; a couple years ago, a moderate amount.
<manveru>
hehe, yeah
<zimbatm>
same, I used to do a lot of ruby, but not so much
<manveru>
i don't use ruby that much myself anymore, but i still work at a company that has almost everything on it
<manveru>
plus i still like the language, don't like it being a second-class citizen in nixpkgs
<qyliss>
I don't use Ruby too much myself at this point either, but my job is streamlining the development environments for a company of 60 Ruby developers
<qyliss>
And I'd love to be able to get them on Nix
<qyliss>
Because I'd never have to deal with another fucking dynamic linking issue when Homebrew updates readline or whatever
<manveru>
:D
<manveru>
pretty much same here
<manveru>
just a hard sell if the tools are a pain to use
<manveru>
though at this point i don't notice it anymore, so i'd love some new perspective
ivan has quit [Quit: lp0 on fire]
<qyliss>
Yeah exactly
<samueldr>
I use ruby as my first go to language to get things done, I thank you for the work you're doing :)
<qyliss>
Me too, when it's not bash or execline
ivan has joined #nixos-chat
MichaelRaskin has quit [Quit: MichaelRaskin]
<Church->
God I hate spreadsheets so damn much
<Church->
Heya samueldr
<Church->
How goes it
<jasongrossman>
manveru++
<{^_^}>
manveru's karma got increased to 13
<jasongrossman>
I'm using ruby at the moment. I've decided to bypass the ecosystem, but it makes me appreciate manveru's work.
<Church->
Yeah manveru did some top work there
<Church->
And I really, really need to upstream this service at some point
<jasongrossman>
qyliss: SIXTY Ruby programmers?!
<qyliss>
approximately
<Church->
Okay think I've figured out this api almost
<qyliss>
I mean, you have to be reasonably big before it makes sense to hire someone like me who doesn't do any actual product work, but spends all day working on the experience of the other developers
jasongrossman has quit [Ping timeout: 246 seconds]