<bqv>
I think its plenty fine. Any other way requires both parties to have something installed, or some middleman service, but everyone has netcat/python :)
<eyJhb>
Everyone should have Nix installed as well!
<name54>
why not just pass your friend a public ssh key and have them put stuff on your server? does everyone here not run their own server?
<name54>
or vice versa... sorry I can be dyslexic sometimes
<eyJhb>
Too much work
<eyJhb>
Also, I do not trust some of my friends with that kind of access
<eyJhb>
Especially since many of the pwn to own
<name54>
that's crazy... and it's just temporary. you can comment the key out of authorized_keys after the swap
<name54>
well, everyone has their prefs
<eyJhb>
Seems a lot easier to just `nix-shell -p python2 --run 'python -m SimpleHTTPServer'`
<name54>
maybe you have a point though eyJhb, people are paranoid
<eyJhb>
Well, I would say I have a reason to be
<eyJhb>
I would rather trust someone in here, than those guys. But still, not a very good way to share files. I have done it by spinning up a VPS before
<eyJhb>
Then I can just take it down
<name54>
well, that's basically what i'm suggesting but not taking the VPS part down, simiply commenting the ssh key
<eyJhb>
But else, most are in the security branch of work, and many of them X members of what I guess translates to "Denmarks National Cyber Security team"
<eyJhb>
So a SSH keys gives access to all kind of fun you can fuck with
<name54>
sure, but you're going to put it into root's authorized_keys file O:3
<name54>
instead of spinning up a VPS you make a restricted user and put it in their authorized_keys
<name54>
or actually, you know... make that user in your VPS... or your home computer/laptop for that matter
<name54>
anyway... what'd i miss from my nap. anything good?
<name54>
i noticed the scientific code debate raged on a little
<eyJhb>
user/root, does not matter. User access for sharing a file is generally not great practice.
<name54>
well, I trust you eyJhb. if you need to send me a file I'l happily put your public ssh key into one of my machines for a few minutes <3
<name54>
see, i have a theory about all this. and it includes the scientific code dilemna too
<name54>
and I guess it boils down to basic diplomacy... it's always better to give credit than to place blame
<name54>
people make mistakes, scientists, doctors, software engineers... everyone
<name54>
nobody is perfect
<name54>
so, it's important to focus on the good things people do and be able to kindly forgive the bad. that's not to say we should just ignore misdeeds
<name54>
but we should be careful not to cause further harm to the individual who suffers from a failing, after all... they're already messed up a little
<name54>
or we could just crucify everyone... your call. lol
<eyJhb>
Crucify.
<eyJhb>
Problem is, there is no one to tell them to stop or make propper gode
<eyJhb>
E.g. my brother is currently taking a PhD, where they are helping people in a wheelchair that have spinal cord injuries. So basically can only move their head. So they are making a "robot arm", which can do various things
<eyJhb>
Many of the people collaborating on this, are writing their code in Matlab. Which is insane, as it needs to be mounted on that wheelchair, etc.
<eyJhb>
Matlab is fine for yourself, etc. but not to fucking distribute
<name54>
lol, I'm sure the people at Matlab will grant a license for a modest fee
<eyJhb>
"Lets say, 5.000 USD a week! You want that arm to work, right?"
<eyJhb>
Generally the Matlab fetish... *sigh*
<name54>
well... I mean the people who made Matlab need a place to live and some food too... don't they?
<name54>
Matlab is a pretty sophisticated tool at this point... it's not "hello world"
<name54>
expecting them to sacrifice everything they have on the hope that someone can make something useful isn't fair imho
<name54>
if someone uses their stuff to make something cool the Matlab people should be rewarded
* ashkitten
screams
<name54>
not to mention, the matlab people can't just fall asleep at the wheel... or Mathematica will eat their lunch
<ashkitten>
developers building release artifacts with a dirty git repo makes me so viscerally upset
<eyJhb>
You are missing something name54
<eyJhb>
Matlab is forced down or throat at Uni, instead of using free stuff such as Python which can do the same, and much more
<name54>
guess it wouldn't be the first time... what am i missing
<eyJhb>
Because they were taught Matlab waaay back, and now we are
<ashkitten>
i just found a bug building this project from source that didn't exist in the released binary because they built it with uncommitted changes
<ashkitten>
and now i will scream forever
<eyJhb>
They have even gone so far, as we need to know some of the commands for exams. Fuck. That.
<eyJhb>
ashkitten: That it the best kind of builds!
<ashkitten>
this is why CI was invented
<ashkitten>
>:(
<name54>
yeah, idk... that sounds like a very subjective point. like i suggested... it could just have as easily been Mathematica that targeted the edu community
<eyJhb>
Still something that I and others would see gone
<eyJhb>
As there is no reason for it
<name54>
or like you said, the profs could just have taught python numpy from the start or something more appropriate but I figure you have to start somewhere
<eyJhb>
They teach it because, they know it, sadly
<name54>
I was going to choose SAS for one of my projects but decided R was better for the same reasons you suggest
<name54>
they teach it because they knowit... and that brings me back to my original point. we're all imperfect; professors too
<name54>
being perfect isn't what's important... striving for perfection is
<Taneb>
My overkill idea to have a wiki set up to keep track of what's happened in long-running media is getting more and more overkill
<Taneb>
I'm rewriting the mediawiki NixOS module
<Taneb>
Or, writing a new module that overlaps in functionality, but allows wiki farms
<JJJollyjim>
eyJhb: fwiw i kinda enjoyed matlab at uni
<JJJollyjim>
it's really nice to just have something put together somewhat cohesively for you
<JJJollyjim>
instead of like "i want to read an image, now do i use imageio or scikit or PIL or pillow or opencv, and some of those have incompatible ideas of what an image is, and python2/python3/pypy and python dependency managment and etc etc"
<eyJhb>
name54: and most will not do that. They will strive for it to be easy ;)
<JJJollyjim>
(talking from a "teaching scientists/non-software engineers to quickly get stuff done perspective", not a "good software engineering" perspective)
<eyJhb>
"getting stuff done quickly"
<JJJollyjim>
i found it pretty refreshing coming from a programming background
<name54>
JJJollyjim: guess that's one reason why Matlab is good for scientists and students!
<eyJhb>
Most of the time, we have waited for Matlab to download
<eyJhb>
Or not work because license shit, etc.
<JJJollyjim>
ah yep it was preinstalled and configured on lab computers every time i've used it, that probably helps :P
<name54>
Yeah, there are louse teachers out there... maybe their lessons are even more important, lol
<eyJhb>
But this is now somewhat better, because you have online Matlab...
<name54>
*lousey*
<JJJollyjim>
i had online matlab at uni, it was called X forwarding :D
<eyJhb>
But that would then be the same with Python, preinstall it, and have a simple "lab" doc with such a basic thing if needed :p
<eyJhb>
name54: there are many of them
<name54>
docker! lol
<eyJhb>
JJJollyjim: you have RCE on these
<eyJhb>
name54: They currently run it in Docker
<eyJhb>
And there is Python installed on them!
<JJJollyjim>
i have rce on lab computers :P
<name54>
eyJhb: oh, that make sense
<JJJollyjim>
true, though having an ecosystem where you can google "how do i do _ in matlab" is nice
<eyJhb>
Not saying free ressources from Matlab, but I am saying free ressources
<JJJollyjim>
compared to "how do i do _ in my uni's subset of python"
<eyJhb>
I would say, you can do the same for Python. Also, I just suggested having a basic set of packages :p
<name54>
Oh, the world definitely arcs toward FLOSS... i fully agree
<eyJhb>
Not having our own `from aau import our_own_image_lib` :P
<JJJollyjim>
idk
<JJJollyjim>
maybe
<name54>
but early adopters and pioneers get compensated for the risks they take if they're smart
<JJJollyjim>
it's fun how many of the science-adjacent python packages are exact replicas of the matlab APIs
<JJJollyjim>
or started out that way
<eyJhb>
On another topic. I am not sure if I should allow for implicit declaration of structs in my Go code, and then have a function which puts in the default values..
<eyJhb>
name54: well, only me working on the code atm. But just trying to make the configs files, as small as possible and easy to create without too much boilerplate stuff
<name54>
eyJhb: oh, is there a common library folks use w/ go for cfg file managment... I'd usually rather not recreate the wheel
<eyJhb>
name54: Cyber Security platform for training, so it needs a lot of files for each challenge
<srk>
same here. I'm almost done with my NUR repo and flakes are the new thing .. :D
<eyJhb>
Are you keeping your new background JJJollyjim ?
<JJJollyjim>
i assume itll go away when i restart x
<JJJollyjim>
but otherwise i guess haha
<eyJhb>
*hopes it stays*
<eyJhb>
*and that JJJollyjim will make a lot of lasagna now*
cole-h has quit [Quit: Goodbye]
<name54>
:] Guess you could always load feh again with a different image and press the wrong key again if you don't wan to bother figuring out your window manager or desktop env.
<name54>
That kind of thing drives me absolutely crazy!
<name54>
Thats actually why I keep religious backups of my home... I'd end up doing a diff on the whole thing against my last backup if I had to X)
<eyJhb>
Backup every 5 minutes <3
<bqv>
5 mins?! Thats a long gap. Try 5 seconds
<name54>
lol, just about... actually in all seriousness... I snapshot weekly and backup every time I make a "big" change I like, '(version-control t) in my .emacs of course
<name54>
important projects in git, naturallly... and all that dupulicated off-prem lol
<bqv>
adisbladis: ...woah
<name54>
I never heard of nilfs ... sounds like a good option actually. I'm just using zfs and I have to wonder how it compares wrt to speed, and in general. I have to guess zfs is the better option; nothing against nilfs.
<sphalerite>
name54: why only weekly snapshots?
<eyJhb>
name54: I think many of us with ZFS do every minute or so, and keep it for 15 min, etc.
<name54>
hmmm... why weekly? idk... my life doesn't move that fast I guess, lol. plus i'm not sure all the minute by minute changes are worth keeping. How badly can I mess things up (or lose stuff) in 1 week?
<sphalerite>
one stray rm…
<sphalerite>
I take snapshots every 15min, it seems like a good compromise between recency and avoiding unnecessary IO
<name54>
yeah, but that's what the off-prem is for i guess. I can just boot off a cloned image and pull stuff back in an hour or so
<sphalerite>
right, but in the worst case for you that means losing a week of work?
<name54>
all this is, of course, is why I'm lurking here in the nixos-chat irc for so long... by over arching concern at this point is the "initial" boot image... I want to make sure that's golden so maybe looking for an alternative to makeing clonezilla images of my boot/root
<name54>
sphalerite: rome wasn't built in a week, lol
<name54>
i can do it over in that kind of catastrophe
<name54>
i won't fault another more frequent process though... it's just my own comfort zone
<name54>
can someone still sell me on nixos after I just revealed all that?
<immae>
sphalerite: your sentence is how a friend of mine convince people to do more frequent backup ("Oh, you’re doing backup every month? then you’re ok to lose one month worth of work, good...")
<eyJhb>
name54, nope banished from NixOS.
<eyJhb>
I just have two SSDs in my laptop, where I have snapshots on the primary and on the secoundary as well
<eyJhb>
And it doesn't fill up much, when you use "tmpfs" in my case
<eyJhb>
Are we calling that setup stateless or statefull?
<eyJhb>
Or state-in-between
<ashkitten>
finally rounded up all the changes in my nixos configs and committed and pushed them
<ashkitten>
feels good
<bqv>
> pkgs.ghc.version
<{^_^}>
"8.8.4"
name54 has quit [Ping timeout: 245 seconds]
name65 has joined #nixos-chat
<name65>
lol
<name65>
what that it? just q.e.d. I guess... huh? maybe so
name65 has quit [Remote host closed the connection]
<sphalerite>
immae: :D
<sphalerite>
of course it's a question of risk management
<immae>
yes, but usually people don’t realize how much risk they take
<sphalerite>
I consider myself accidentally removing a file a lot more likely than my laptop's mass storage failing, so I have frequent snapshots locally so I lose a maximum of 15min of work in that likely scenario
<immae>
right. That’s (removing a file accidentally) something I couldn’t afford until yesterday night :p
<sphalerite>
whereas I'd be satisfied with daily backups to my backup server that guard against my laptop's storage failing, because while I lose more work that way I consider it a lot less likely
<{^_^}>
t184256/nix-on-droid#34 (by ShamrockLee, 33 weeks ago, open): Tightvnc vncserver: couldn't find "Xvnc" on your PATH
leah2 has joined #nixos-chat
<pie_>
and i cant strace it because no ptrace :I
<pie_>
apparently strace is availible in termux
<pie_>
why not nix-on-droid
<pie_>
hmm, well what it says is proot warning: ptrace request "PTRACE_???" not supported yet
FRidh has joined #nixos-chat
<infinisil>
ar: Hehe
<pie_>
strace in termux works so theres probably some kind of proot issue
<pie_>
and xvfb also works in termux
<pie_>
so now the question is how can i connect x11 apps from nix-on-droid to the x server running in termux
neeasade has joined #nixos-chat
Jackneilll has quit [Ping timeout: 258 seconds]
Jackneilll has joined #nixos-chat
FRidh has quit [Quit: Konversation terminated!]
<pie_>
i ran x11 using termux and i was able to tell x11 apps from nix-on-droud to use it
<pie_>
now i just need to figure out why things are always hanging
<pie_>
maybe android is pausing processes or something even though i told it to acquire wake lock
abathur has quit [Ping timeout: 240 seconds]
waleee-cl has quit [Quit: Connection closed for inactivity]
cole-h has joined #nixos-chat
abathur has joined #nixos-chat
abathur has quit [Ping timeout: 246 seconds]
<sphalerite>
why do I still feel like GNOME 3 is a new thing?
abathur has joined #nixos-chat
endformationage has joined #nixos-chat
abathur has quit [Quit: abathur]
davidtwco has quit [Ping timeout: 240 seconds]
davidtwco has joined #nixos-chat
raboof has quit [Ping timeout: 240 seconds]
steveeJ has quit [Ping timeout: 240 seconds]
raboof has joined #nixos-chat
steveeJ has joined #nixos-chat
waleee-cl has joined #nixos-chat
<pie_>
dunno if yall seen (if someone would be so kind as to link it) this paper from 3 weeks ago abut (?) nix: Self-Scaling Clusters and Reproducible Containers to Enable Scientific Computing
<pie_>
looks like some HPC stuff
<colemickens>
Does it happen that RAM will be flakey around a mem test? Like, possibly report bad, then later report good, then report bad again? Or does this indicate mobo failure too possibly?
<Church->
pie_: Ooooh
<pie_>
* its on arxiv
<eyJhb>
colemickens: usually indicates a bad ram stick or slot
<eyJhb>
Test them one by one and do multiple passes, and switch slots
<eyJhb>
(if possible)
<colemickens>
yeah, I've been doing the best I can to filter it out but the test results basically have to be flakey. Or there's another bug in the machine today, but zsh and rustc have both been segfaulting.
<pie_>
Church-: paste link :pp
<Church->
pie_: Haven't looked it up yet
<eyJhb>
memtest , and how many passes ? And how many sticks/GB colemickens ?
<colemickens>
it's hard, things "seem" fine with certain configs too, but I have an absurd amount of RAM and rarely dip into all of it so...
rajivr has quit [Quit: Connection closed for inactivity]
<colemickens>
eyJhb: I just use Windows' mem test, hoping that they all work the same.
<eyJhb>
I haven't used that, ever, I would trust memtest more and make a bootable USB. I have used that since always, and trust that
<eyJhb>
How much mem do you have?
<eyJhb>
Sounds like TB
<colemickens>
64GB
* pie_
waits for firefox to start
<eyJhb>
4*16?
<colemickens>
but it's only a 4 core machine that I play games on and sometimes run nixos on and work from
<colemickens>
yeah
<eyJhb>
Make a bootable USB and use Memtest, or enable memtest in NixOS and use that
<eyJhb>
DriveDroid if you phone is rooted. Then see if that is flaky as well
<colemickens>
yeah, at the very least it will avoid the issue that windows's memtest doesn't play nice with bitlocker
<eyJhb>
Never trust windows! :(
<eyJhb>
\s but I would go in that direction
<eyJhb>
Faulty memory sucks. But usually it will fault the entire computer and just crash, not just segfault. At least in my experience
<colemickens>
hmmm
<drakonis>
windows tolerates faulty hardware until it starts failing
<bqv>
linux tolerates it even after :p
<drakonis>
rather
<drakonis>
windows tolerates until its dead :V
<sphalerite>
colemickens: yes, that does happen — that's why it's recommended to let memory-testing software do multiple runs of various patterns across the whole RAM