<gchristensen>
I wish I'd learned Make better than I do
<abathur>
https://github.com/uasi/git-anything - slightly cleaned up from google translation: It's kind of useful to turn every command into a subcommand of git. Just pass the bin path and you can do whatever you want with git: git echo hoge, git ruby fuga.rb, or git shutdown -h now.
<gchristensen>
lol!
<abathur>
but you appear to know expect
<gchristensen>
me?
<gchristensen>
yeah
<abathur>
I'm not sure which dark art I'd choose...
<samueldr>
I had to learn COBOL, not sure if it force filled that "dark art" niche
<samueldr>
it was weird how to use the compiler for the class, dosbox was the better option
<samueldr>
for the HOW OLD ARE YOU questions: that was back in 2008 or 2009
<energizer>
anybody know what's the best tool for pulling structured tables out of PDFs?
<abathur>
a dentist?
<abathur>
have you tried anything already, or just generally fishing?
* abathur
creates a branch with a single quote in the name
<gchristensen>
lol
<energizer>
i used Tabula a few years ago, it was alright but needed a lot of postprocessing
<energizer>
whoa gchristensen that's interesting
<ajs124>
abathur: I started a habit of putting slashes in branch names. That was fun recently, when I tried pushing to test/bla when test already existed on the remote.
<gchristensen>
a fellow rural MA resident works on it
<abathur>
ajs124: I've wondered about that; I like it when I see it but assume I'd break it and stick to underscores :]
<ajs124>
🤷♂️ it worked well until it didn't
<samueldr>
abathur: try mixing and matching macOS, linux and é
<samueldr>
in a branch name
<samueldr>
not that *I* did it, I knew better
viric has quit [Ping timeout: 246 seconds]
<abathur>
I don't always make all of my branch names valid shell expansions if not properly escaped...
<samueldr>
git checkout -B '`echo hi`'
<abathur>
hmmmmmmmmmmm
<abathur>
can probably do some weird stuff to people's aliases and completions
<abathur>
wonder if anyone evals any git output in their prompt script
<samueldr>
*evals*... whew that's scary
<samueldr>
I don't like that idea
viric has joined #nixos-chat
rajivr has joined #nixos-chat
<abathur>
so, there are people
<abathur>
who do things like
<gchristensen>
I bet there are people who do that :(
<abathur>
in their defense
<abathur>
I did go check to see if I do it
<abathur>
and even if I don't, this fancypants bash events library I use does, so I can't swear I don't
<abathur>
but I haven't broken my prompt, yet
<gchristensen>
ack
<gchristensen>
git rebase -x is so good
<samueldr>
for what good use case?
<gchristensen>
like running a formatter on every commit, or verifying tests pass on every commit, since a parent branch
<abathur>
nice
<samueldr>
I don't know how the formatter thing would work; would it cause "as few" conflicts as possible since it's done before a commit is done?
<gchristensen>
nah it is actually extremely painful for that usecase :P
<samueldr>
right, as I thought it would be :)
<gchristensen>
I have a script for that though ...
<abathur>
luckily that's just basically part of an alias
<drakonis>
i'm starting to feel like i should pave over this server and just run nixos in it
<drakonis>
i'm hating ubuntu with all my soul
<samueldr>
TIL about .git/info/exclude
<samueldr>
basically a local variant of .gitignore
<drakonis>
i hate the house of cards that other distros are
<drakonis>
i spent 2 hours troubleshooting ubuntu's mysql package
<drakonis>
i could've been productive
<elvishjerricco>
Aw. NixOS can't boot from a degraded ZFS pool. Like if you've got raidz2 and you pluck a disk, it won't boot because the `poolReady` bash function in `tasks/filesystems/zfs.nix` only looks for the ONLINE state and doesn't consider DEGRADED.
<drakonis>
i think i'm going to ask my teacher whether i can raze ubuntu and set up nixos because holy shit this is improductive
<samueldr>
elvishjerricco: sounds like you know how to make the fix at least :)
<drakonis>
thank goodness for lustrate being a thing
<drakonis>
as i can only reach it while sshed
<elvishjerricco>
samueldr: It does raise the question of if the user should be prompted about continuing. Some people don't like running a degraded pool.
<samueldr>
I wouldn't be able to answer that question
<gchristensen>
no
<gchristensen>
I should be able to recover my still healthy enough machine remotely without going in to emergency
<drakonis>
i remember it coming up on 2018's roadmap talk
endformationage has quit [Quit: WeeChat 2.9]
<samueldr>
>> I just informed Greg KH privately about it. He was not made aware of these issues prior to now. What's up with Intel and Google's security disclosure process ?
<ldlework>
hmm yeah i tried that. `npm i @hqjs/hq` no workie on nixos
<ldlework>
something something mozjpeg
spudly has joined #nixos-chat
<drakonis>
i'm excited about those nixcon talks
<drakonis>
plus i'm going to deploy nixos on a server
<drakonis>
another one, that is.
spudly has quit [Ping timeout: 256 seconds]
<drakonis>
config recommendation system talk
<drakonis>
experimental modules system talk
<drakonis>
ho ho ho
<drakonis>
then mainstream nix talk
<drakonis>
now this one's going to be something
spudly has joined #nixos-chat
spudly has quit [Ping timeout: 260 seconds]
spudly has joined #nixos-chat
<sphalerite>
Why do my phones never survive me moving? Moved into a new flat today, promptly dropped my phone on the kerb.
<sphalerite>
Relatedly: can anyone recommend a smartphone? Especially interested in vendors that are good in terms of their kernelness, hence ping samueldr ;)
spudly has quit [Ping timeout: 258 seconds]
spudly has joined #nixos-chat
<sphalerite>
ldlework: probably use nix-shell instead, and also ask in #nixos :p
<sphalerite>
on that topic, does anyone have good tricks for finding disassembly videos/tutorials? I tried searching for ylxd01yl disassembly on ddg and youtube, but haven't found anything for the exact model…
<gchristensen>
I always start with ifixit's website, and then I'll go to the FCC database which usually has disassembly pictures, and sometimes there are clues
<sphalerite>
gchristensen: where do I find the fcc database? :D
aleph- has joined #nixos-chat
<sphalerite>
hm actually maybe this was never tested by the FCC :/
<sphalerite>
it doesn't have an FCC logo on it in any case
xd1le has quit [Read error: Connection reset by peer]
spudly has joined #nixos-chat
spudly has quit [Client Quit]
xd1le has joined #nixos-chat
spudly has joined #nixos-chat
spudly has quit [Ping timeout: 264 seconds]
<samueldr>
sphalerite: better option for availability of everything: google
<samueldr>
I had good experiences with ASUS, but YMMV
<samueldr>
oneplus are apparently doing like Xiaomi now, and forcing a wait time before an unlock for newer devices [citation needed]
<samueldr>
lenovorola I don't know about their kernel release, though it appeared that Lenovo (not motorola) branded devices might sometimes not have their kernel released promptly
<samueldr>
(or at all?)
<samueldr>
xiaomi still releases quickly enough, but the whole thing of linking with an account and waiting 2 weeks is not good
<drakonis>
the wait time is like rolling a dice, sometimes it can take longer than two weeks
<samueldr>
that's when you do something they deem suspcious
<samueldr>
e.g. trying to go around the limits :)
<samueldr>
though with my limited 2-devices experience it never jumped past
endformationage has joined #nixos-chat
<samueldr>
samsung to me is still a big unknown
<samueldr>
I would hazard a guess that kernels are made available in a timely manner
<samueldr>
though bootloader unlock is not a given
<samueldr>
LG similar story, but worse, AFAIK all their north-american devices can't be unlocked [booooooooooooooo]
rajivr has quit [Quit: Connection closed for inactivity]
<samueldr>
trying to think about other OEMs
<samueldr>
sony are good open source citizens
<samueldr>
but pricey (meh) and uh, sometimes hard to get
xd1le has quit [Read error: Connection reset by peer]
xd1le has joined #nixos-chat
<samueldr>
oh, fairphone are probably a fine choice for Europeans
<samueldr>
though if you intend to frequently rely on North American bands, YMMV
cole-h has joined #nixos-chat
<sphalerite>
my dad had a terrible experience with a sony phone, so I think I'll pass on that one
<samueldr>
everyone can find someone that has had a terrible experience with everything ;)
<samueldr>
but I do concur, that it leaves a bad taste
<samueldr>
though anyway their phones are quite expensive
<sphalerite>
I think fairphone might be the right thing for me, at least if I'm buying new. I bought my last two phones used, and feel a lot better about the environmental impact doing that :D
<samueldr>
the big problem with buying a used phone, or even a last-gen phone, is that you basically buy into not having security updates sooner :|
<samueldr>
can we please apply the wintel paradigm to phones?
<sphalerite>
well I'm not buying one willy-nilly, I won't go for anything that isn't currently supported by lineageos
<sphalerite>
and the whole kernel thing is about sustainability for me as well (as in I can sustain use thanks to downstreams like lineageos and postmarketos)
<samueldr>
you do get more security updates, but even then it's bound to stop with LineageOS :(
<samueldr>
yeah
<sphalerite>
I mean, also mobile-nixos, but mostly sustainability :D
<samueldr>
yeah, in my mind if you talk about any alternative software possibility Mobile NixOS is a possibility
<sphalerite>
:D
<samueldr>
proven by asus-x018d, no custom rom available, even the only TWRP available is built using the binary kernel from the OEM
<sphalerite>
oooooh
<samueldr>
but Mobile NixOS with a kernel built from the OEM sources exists
<sphalerite>
yeah maybe I'll just go for the fairphone…
<sphalerite>
how much of the hardware works? ;)
<samueldr>
x018d?
<sphalerite>
yeah
<samueldr>
as much as expected for the current state of things
<samueldr>
mediatek-based
<samueldr>
wifi not explored, modem not explored
<samueldr>
but everything else seems to work just fine
<sphalerite>
ah ok
<samueldr>
basically same state as the popular qualcomm devices
<samueldr>
if the fairphone was available here, I surely would have gotten them
<samueldr>
(available also mean useful connectivity-wise)
<sphalerite>
unrelated: how might I be able to recover data from a failed (kernel can see it but I get messages like "ata4: softreset failed (device not ready)") SSD?
<sphalerite>
because _of course_ the one thing I have no backups of is the boot SSD of my backup server…
<samueldr>
ironic
<samueldr>
I would hazard a guess that nothing can be done
<pie_>
samueldr: kinda orthogonal but you miught try asking people in #openfgpa this kinda stuff
<samueldr>
pie_: about SSDs? that's sphalerite
<pie_>
no about teardowns and idk
<sphalerite>
that was also me x)
<pie_>
oh
<sphalerite>
ah well, guess I'll have to redo my home assistant setup
<sphalerite>
not the end of the world I guess, I wanted to move to docker for it anyway
cole-h has quit [Quit: Goodbye]
<drakonis>
hrrrm i need to expand my linode vps
<drakonis>
as i need more memory
<colemickens>
Is everyone just lurking in IRC during NixCon?
<eyJhb>
Wait, is there Nixcon?
* colemickens
wishes he was in Munich
<colemickens>
eyJhb: :) I'm glad I said something, it starts tomorrow! It really crept up on me.
<drakonis>
roadrunner uses the same binary namespace as rr
<drakonis>
sadly
<eyJhb>
But it would be cool!
<drakonis>
its in python no?
<eyJhb>
Yeah it is
<drakonis>
i'll check it later
<drakonis>
hmm, that module talk... i cant wait to not have to finagle with overriding for enabling package features
<eyJhb>
`calonloupanseajzkrfls@kekasihhatikusiapayangmau.com` legit Apple Email
<gchristensen>
abathur: is there a way to add this syntax to bash: ((foo))<>((bar)) where foo and bar are two programs, and the pipes are connected as: foo | bar, and bar | foo ?
<drakonis>
the great dictionary nightmare is over
<cole-h>
That's super weird and now I want it
<gchristensen>
it is possible to do, and I've done it, but if there is a terrible way I could trick bash in to giving me that syntax for it, I'd prefer it :P
<eyJhb>
gchristensen ! Do you know of any libraries that uses NixOS? As in, the places with the books
<gchristensen>
sure
<gchristensen>
why?
<eyJhb>
Wondering if there is any success in providing NixOS to the users, like, the computers that can freely be used to browse the WWW, write documents, etc. at the library
<eyJhb>
Not the backend systems as such, because then we have srhb :D
<gchristensen>
ah :)
<gchristensen>
not sure
<eyJhb>
Damn... Would be cool, however... Maybe not that old people friendly, if they want to teach them how to use a Linux system at home
<lovesegfault>
nice, nixos-unstable advanced 🎉
<eyJhb>
Once more?
<eyJhb>
Aww, you had my hopes up for another one lovesegfault
<lovesegfault>
eyJhb: one a week is enough to celebrate!
<eyJhb>
unstable is so ofte killed atm. :(
spudly has joined #nixos-chat
<samueldr>
gchristensen: back and forth?
<gchristensen>
yeah, the back and forth operator
<gchristensen>
:)
<samueldr>
forever?
spudly has quit [Ping timeout: 256 seconds]
<gchristensen>
forever, or until one of the writes gets a SIGPIPE I guess :P
<samueldr>
,escape'' ''
<{^_^}>
Escape this in '' strings with: ''' '''
<{^_^}>
Escape this in " strings with: '' ''
<samueldr>
I uh... that seemds like a weird answer
<samueldr>
but it's right, but it seems to be answering to ,escape(.*)
spudly has joined #nixos-chat
<drakonis>
nice!
spudly has quit [Ping timeout: 260 seconds]
<abathur>
gchristensen: not sure if there's anything terse, but I haven't really tried/considered that one before; what's the other way you've done it?
<gchristensen>
have to create a 3rd pippe and <&3 ... >&3
<abathur>
hmm
<abathur>
there's a particular bit of arcana here that I rarely see used, I wonder a bit...; I don't feel like I have a great handle on what all it can do yet, though I do remember trying to do some weird stuff with it and did manage to establish that there are some things that seem like valid syntax that it won't do
<abathur>
I'll go look for the doc, but basically function definitions can have redirections on the closing }
<abathur>
so like whee(){ echo hehe; } >2
<abathur>
and then bare invocations of whee will respect it
<infinisil>
samueldr: Yeah I should probably fix that
<infinisil>
Or make it a bit more intuitive
Jackneill has joined #nixos-chat
<energizer>
abathur: i would love a keyboard-driven gui based on that
cole-h has joined #nixos-chat
<abathur>
on which?
<abathur>
pipexec?
xd1le has quit [Remote host closed the connection]
__Sander__ has quit [Quit: Konversation terminated!]
<pie_>
domething that grinds my gears: why dont any major operating systems have a rhobust low-memory environment you can access when you hang your system
<pie_>
by resource exhaustion
<pie_>
any if they do why isnt it available by default
<gchristensen>
all major operating systems pretend to be a turing machine
<gchristensen>
which is to say, they all pretend memory is infinite
<gchristensen>
Rust, a language famous for making you deal with problems doesn't even acknowledge the idea that a memory allocation can fail
<pie_>
i cant be the only person thats wanted this in the past 80 years
<gchristensen>
almost certainly no significant number of C/C++programs can do anything reasonable but fail if it fails to allocate more memory
<pie_>
im saying the OS could acknowledge this even is apps dont
<pie_>
the problem is it doesnt fail
<pie_>
the problem is the machine slows to a crawl
<pie_>
so probably swapping but idk
<gchristensen>
yeabh
<gchristensen>
that is why I don't have swap anymore
<gchristensen>
and why facebook made oomd (now systemd-oomd I think?)
<pie_>
i still brefer that to everything crashing suddenly for some reason
<gchristensen>
slowness?
<pie_>
i think os should reserve a small amount of ram for some sort of supervisor that remains responsive and resident
<gchristensen>
AH
<gchristensen>
ah*
<pie_>
yes the slowness
<pie_>
and the supervisor would be a method to selectively unslow :p
<samueldr>
already, having all processes sharing from the same bowl, without some kind of priority or preference is not great
<gchristensen>
I much prefer the rapid death, personally
<pie_>
well i guess vms or containers or whatever are the solution to this
<gchristensen>
away from CEO computing and onwards to robust-first computing!
<samueldr>
your firefox process can decide it takes it all, so now there's none left for your WM
<samueldr>
or the system processes
<pie_>
gchristensen: id probably like that more if i didnt work with tons of state and most things being garbage at orthogonal persistence
<gchristensen>
yea
<pie_>
if things were good at restoring themselves, yeah sure have fun with process mass murder
<pie_>
i still need to look at criu
<samueldr>
since I started limiting the firefox process, I haven't had any problem
<pie_>
+containers
<samueldr>
though tbf I also changed my usage pattern around the same time :/
<pie_>
idk even with firefox limited to 8g i stll have no trouble fillign up most of my 16gb of ram
<pie_>
and i get a slow ff as a bonus but at least the rest of the sys kinda works
<samueldr>
yeah, most of the "real work" happens on another computer
<samueldr>
(or happened)
<pie_>
samueldr: im kinda doing that too right now but i have to give my mom her laptop back now xp
<pie_>
also whaat triggered this discussion is i got windows to hang
<pie_>
gchristensen: hm guess ill watch that later
<gchristensen>
it is a neat one
<pie_>
something is garbage with either win10 or the filesystem or firefox
<pie_>
or all 3
<pie_>
every time i hard reset it loses the entire session
<abathur>
maybe a little too OT, but does anyone use a ram-backed TMPDIR for Nix builds? Curious about performance, and how much RAM you have to throw at it
<gchristensen>
I don't have good answers other than "great" and the fact that I used "all of it"
<abathur>
hmm
<andi->
abathur: if you have >=16GB of RAM it should be okay most of the time unless your hobby is compiling chromium, firefox, ….
<abathur>
<3 andi- good reference points. I have 16GB on the laptop I'm curious about. It looks like I'll be able to create a tmpfs using up to 8GB, but I can't test it out on real builds yet
<{^_^}>
andi-'s karma got increased to 37
<hexa->
gchristensen: systemd-oomd: @keszybz keszybz merged commit 69c0807 into systemd:master 11 hours ago