<abathur>
I've got a TODO somewhere about seeing if I can update my prompt with one or more indicators about an annex if I'm in one
slack1256 has joined #nixos-chat
<infinisil>
I've seen git annex before, but this is the first time I actually took the time to read what it is
<infinisil>
It sure is rather interesting!
<drakonis1>
it truly s
<drakonis1>
is
<drakonis1>
you can use it to sync files in the cloud too
<abathur>
it's nice; it makes me a little anxious but I think that would ebb if I forced myself to use it on the reg; until then I wish there were a more integrated git+annex-aware CLI
waleee-cl has quit [Quit: Connection closed for inactivity]
cjpbirkbeck has quit [Quit: Good bye and take care.]
<colemickens>
pie_: so, now every time rsync comes up I'll think of this and inevitably not be able to find it
<colemickens>
oh hey. it's #2 for "rsync oneliner" on google. Of course that will change as soon as I hit send, but for now at least.
<pie_>
colemickens: so it goes
AluisioASG has quit [Read error: Connection reset by peer]
AluisioASG has joined #nixos-chat
cjpbirkbeck has joined #nixos-chat
<colemickens>
every once in a while I'm surprised at the things the kernel can do that doesn't exist in library form
<colemickens>
also, is 7zip and mounting the only two options for reading from UDF?
<joepie91>
colemickens: I'm guessing there are probably parser implementations, but other than that, I suppose? I'm not sure what other usage you're envisioning
<colemickens>
Just surprised at the lack of diversity of tooling, there are lots of random utils/libs for working with ISO, but not UDF.
<joepie91>
mounting vs. opening as archive seems like a fairly exhaustive set of things to do with a disc image :P
<joepie91>
there are?
<colemickens>
I feel like working with filesystems in general, while trying to avoid root or runInLinuxVM type schenangians is... harder than it ought to be
<colemickens>
a quick google of iso9660 util|rust|<language> etc have non-trivial results, less so for UDF and whatever it's ISO number is.
<colemickens>
isoinfo, etc
<joepie91>
aside: pretty sure that UDF is considered to be "ISO" by enough people that ISO tools will probably work on it
<joepie91>
hm, never seen those tools before
<colemickens>
You're right though, p7zip fundamentally does what I need, but there was some 7zip security kerfuffle lately so I thought i'd poke around and see if there was another such tool
<joepie91>
right, makes sense
<colemickens>
joepie91: I'd seen the same, however the particular UDF ... thing... I have doesn't like anything but 7zip so far.
<joepie91>
odd
<ashkitten>
wasn't p7zip development picked up by some other person/org?
<joepie91>
disc formats are a bit of a mess though
<joepie91>
source: have implemented ISO9660
<colemickens>
I've been bumping up against them for ... a long time now and I've always had a hand-wavey sense of "cross my fingers" :)
<colemickens>
I'm having flashes of mucking with daemon tools at 10 or something, ha
<joepie91>
supposedly UDF has less weird caveats and variants than ISO9660 but whether I actually believe that to be true is a different matter entirely :P
<joepie91>
colemickens: my tool of choice nowadays is fuseiso
<colemickens>
<insert xkcd about another standard>
<joepie91>
in like 10 years I've run into one, singular disk image it couldn't mount
<joepie91>
anything else, does it look vaguely like a disk image? it'll probably mount it
<colemickens>
oh nice, I hadn't heard of or remembered that, that's handy.
<joepie91>
vaguely look like*
<joepie91>
(despite what the name suggests)
<joepie91>
it's completely userspace too
<joepie91>
it's pretty much the command-line, FUSE-y, Linux equivalent of Daemon Tools
<colemickens>
cool, thanks!
* joepie91
shares the Daemon-Tools-as-a-teenager flashbacks :P
cole-h has quit [Quit: Goodbye]
cjpbirkbeck has quit [Quit: WeeChat 2.7.1]
<ashkitten>
woah, gpt-3 is incredible
__monty__ has joined #nixos-chat
<eyJhb>
samueldr: did you make any OEM progress?
<eyJhb>
How I wish I could remember all the timezones people where in
parsley936 has joined #nixos-chat
<__monty__>
I think they said Quebéc yesterday.
<__monty__>
*Québec rather.
<eyJhb>
Uhh, that isn't that much off
parsley936 has quit [Remote host closed the connection]
parsley936 has joined #nixos-chat
<joepie91>
eyJhb: for English-speaking channels, 17 - 23 GMT is a pretty safe availability bet
<talyz>
lovesegfault: yay!! :D
rardiol has quit [Read error: Connection reset by peer]
<adisbladis>
Cross-posting from #nixos-on-your-router: Does anyone have recommendations for a usb wifi module for use with hostapd?
<Valodim>
#nixos-on-your-router oO
drakonis1 has joined #nixos-chat
<philipp[m]>
#nixos-on-everything-with-cpu
<gchristensen>
we already have #nixos-power9, how much more could you want!
<gchristensen>
#nixos-mips
<philipp[m]>
#nixos-c64?
<Arahael>
#nisos-apple-watch? ;)
<manveru>
#nixos-pebble :D
<__monty__>
#nixos-on-apple-silicon?
<Arahael>
I nearly bought a pebble the, the day before they annoynced that pebble was no-more. :(
<__monty__>
Think big, people.
<adisbladis>
My first question with any given system is "How can I get this built with Nix", so it's not that far off ^_^
<manveru>
Arahael: still using my pebble, rebble has done an awesome job of keeping it alive :)
<manveru>
and it helps that it's the pebble time steel, with none of the plastic fatigue the other models have
rardiol has quit [Ping timeout: 265 seconds]
<Arahael>
manveru: Nice. Yeah, I nearly bought that one. I currently have an apple watch, which is nice - I use it for the vibrating alarm, but it isn't pebble and I have to charge it at least every 2 days. :(
<philipp[m]>
So I can buy a steel pebble for probably next to nothing and keep it alive with foss indefinetly?
<manveru>
well, RebbleOS isn't finished yet
<manveru>
so not fully FOSS
<philipp[m]>
Still: Sounds not uninteresting.
<manveru>
yeah... i just wish the pinetime went with e-ink :P
tilpner has quit [Quit: tilpner]
tilpner has joined #nixos-chat
<manveru>
i'd even be fine with monochrome, but IPS and touchscreen are really downgrades i can't stomach :(
<__monty__>
Sounds like unhealthy habits catching up with people because there's no consequences attached.
<gchristensen>
adisbladis: having so many of my colleagues in eu makes it really hard for me to slip in that way
<adisbladis>
Having experienced this before: It's on of the telltale signs of clinical depression
<adisbladis>
I think a lot of people are experiencing it right now for the first time
<gchristensen>
tell me about it
<gchristensen>
yeao
<__monty__>
I'm sure it can be. But isn't that simpler explanation that most people just feel less guilt about snoozing because they don't have to show up at work?
<adisbladis>
I don't think so
<adisbladis>
I think a majority of people want to do as good as they can, but it's hard to stay motivated when the world is on fire and there is no end in sight
<gchristensen>
I doubt it
<gchristensen>
I feel guilty as hell.
<gchristensen>
and that guilt is part of why I'm taking a sick day today
<__monty__>
gchristensen: But you can't afford to sleep-in because then you'd completely miss EU colleagues. If you worked with predominantly US-based colleagues there'd still be sufficient overlap.
<gchristensen>
I don't feel guilty for missing those colleagues' work hours
<gchristensen>
I feel guilty because I'm doing a bad job
<Valodim>
it is Nix's only weakness
<Valodim>
people love it so much they burn themselves up
<adisbladis>
__monty__: I don't think you shuold attribute malice here.. A lot of people are in really bad spots right now and experiencing a sense of hopelessness that's not really been felt by most.
<philipp[m]>
I think both effects are there for most people (in techy now suddenly work from home jobs) and they can amplify each other.
<philipp[m]>
I wouldn't attribute malice though because that's just what happens to be conditioned to do your job because of pressure and not because you like it.
<pie_>
feeling motivated to get up in the morning is totally a thing
<pie_>
i had it when I had a girlfriend lol
<pie_>
or maybe it was one of those ADHD "everything only works once" tricks :P
<pie_>
im xposting the fiber-seeking-backhoe
<joepie91>
<pie_> or maybe it was one of those ADHD "everything only works once" tricks :P
<joepie91>
relatable
<joepie91>
:P
<adisbladis>
pie_: +5 years in to my relationship, still helps.
<joepie91>
once your brain figures out that you're trying to trick it...
<gchristensen>
adisbladis: 11yr here and 100%
<philipp[m]>
Yeah, I really don't know how I would have managed this year without my so.
<joepie91>
it's that I moved to a sensible house just in time, otherwise this would not have gone well for me at all
<joepie91>
not that it's going super great now, what with being alone and general depression and anxiety issues returning and all
<joepie91>
but... it could have been significantly worse, had I still been trapped in my old house
<pie_>
well, good that workd out then :V
<pie_>
rationalization is soo goooood, but sooooo baaaaad xD <joepie91> once your brain figures out that you're trying to trick it...
<joepie91>
yeaaaaaaah
<__monty__>
I'm not attributing malice. It's just that most people have horrendous sleep hygiene. They rely on an alarm clock to force them awake and pretty much the only reason they respect the alarm they set for themselves is having to show up at work.
<pie_>
[my sleep hygiene screaming internally]
<joepie91>
__monty__: from experience: good sleep hygiene ain't gonna save you when depression decides to take over your life
<__monty__>
That's not what I said.
* gchristensen
's gonna depart this conversation now
<__monty__>
I'm just skeptical of so many people all being depressed. I know there's people out there depressed because of the lockdowns. I just expect lacking motivation to be more common than clinical depression. That hardly seems like a controversial perspective.
<__monty__>
I'm also not attaching any value judgements.
<joepie91>
__monty__: so, I'll try and put this politely: I very, very strongly doubt that you have the clinical background necessary to actually make that determination - and "hey let's put people's experiences in doubt", *especially* when that is with undertones of "you're just lacking discipline", is an extremely unwelcome discussion for the people who are suffering from this... so just don't
<adisbladis>
So much this ^
<__monty__>
Those undertones aren't mine.
<joepie91>
yes, they are. and instead of trying to defend your points, consider that you are being told "hey, don't do this, this is making people uncomfortable", and just stop doing it.
<adisbladis>
Being depressed is filled with enough self-doubt as it is.
<joepie91>
self-doubt, and a constant social expectation to justify yourself
<joepie91>
at least for me personally, that is why these sorts of talking points grate on me especially much
<__monty__>
I'll stop. But I don't appreciate being painted as a villain. All I'm doing is having a discussion. If clinical depression is really more common than people lacking motivation because of circumstances, that's a very serious issue and I'd like to know more about it.
<joepie91>
__monty__: stop trying to make this about you.
<joepie91>
__monty__: and if you had any real interest in the problem, you would be talking *with* people suffering from these issues, not *over* them, like you are doing right now.
<bqv>
This clusterfuck of a conversation aside, have any of you ever taught someone to drive?
<pie_>
bqv: is that like, allowed?
<bqv>
In the uk, yes
<pie_>
huh , neat
<bqv>
Pretty sure in america too
<__monty__>
Rules here have been made more severe but here too.
<pie_>
i dont have a drivers license so idk
<adisbladis>
pie_: I thought that was allowed pretty much everywhere ?
<joepie91>
I don't think it's allowed in NL?
<manveru>
bqv: tried once to teach my wife... with my sisters car... and got into a hail storm, fun times
<bqv>
I'm trying to be helpful but its impossible to teach someone whe flips out at the slightest bit of stress, and I don't know how I'm meant to navigate that situation
<bqv>
manveru: how did you manage
<joepie91>
so it seems that in NL, it is allowed to drive on private land without a driver's license, but not outside of that
<__monty__>
To anyone who felt slighted by what I said, I want to apologize.
<joepie91>
the practical exam always needs to be done by a certified instructor I think
<__monty__>
I think on private lands pretty much anything goes anywhere?
<manveru>
bqv: wife still doesn't know how to drive, and luckily we had insurance
<adisbladis>
The rule in Sweden is that the person you drive with needs to have had their license for that class of vehicle for at least 5 years
<joepie91>
__monty__: 👍️
<joepie91>
__monty__: nah, you're still not allowed to do a lot of things on private land :P
<joepie91>
like killing people for example
<philipp[m]>
bqv: Maybe start really slow, take them out to a deserted street somewhere on the countryside and let them train there for a while and see what happens?
<bqv>
Fair enough
<adisbladis>
joepie91: But muh freedoms :/
<__monty__>
Does mean you can teach almost anyone to drive at grocery store careparks until employees/managers tell you to stop : )
<bqv>
philipp[m]: tried that
<bqv>
__monty__: yes
<bqv>
Been using a car park
<joepie91>
__monty__: well I translated it roughly, I believe that "private land" in this case means "not publicly accessible"
<joepie91>
__monty__: private vs. public land in NL is more about accessibility and effective purpose, than it is about ownership and designation
<joepie91>
so a carpark would not be allowed, most likely
<philipp[m]>
bqv: So it didn't go well?
<joepie91>
(which would make sense; if the point of the laws is to prevent danger to third parties, it would make sense to prohibit it on any public-use land, regardless of technical ownership or designation)
<bqv>
Idk, instructors have better experience of dealing with this sort of stuff regularly, but I don't understand how I can even help
<__monty__>
Ah, in BE car parks are useful flat empty areas on sundays : )
<bqv>
philipp[m]: she can't even use the clutch to set off properly, nor steers in a safe manner, but if I correct those or anything similar, lesson over and I get screamed at
<joepie91>
bqv: so this isn't *exactly* the same, but in teaching programming, I often deal with people who are afraid to try things out - I usually address that by generic boosting of self-confidence + very clearly describing what the consequences of the worst case are, so that they have a guarantee that it can't be worse than that no matter what they do
<joepie91>
also explaining that the actual result will be somewhere inbetween best and worst case, so probably fine
<philipp[m]>
Then it sounds like you should tell her that you can't do it any more and she has to find sbd else, maybe a professional.
<bqv>
joepie91: that's essentially what I've been trying, but at a point its genuinely unsafe
<pie_>
bqv: kind of sounds like the person has some issues they need to deal with...
<bqv>
pie_: lets not even go there
<joepie91>
(this does depend on my reputation for being blunt about things and not telling people "it will be fine" unless I actually believe so)
<joepie91>
bqv: then yeah, this is probably the point where a professional should handle it
<joepie91>
ideally one specialized in whatever they are dealing with
<bqv>
Mm
<pie_>
(im just going off the getting screamed at part, i can totally understand being scared of stuff)
<__monty__>
bqv: Maybe talking about it *outside* the scenario is easier? Rather than in the car, at the spot.
waleee-cl has joined #nixos-chat
<pie_>
Im scratching my head at what I'd do in such a situation, so I guess imma observe what other people say here, but like, do they _want_ to learn? Where in the process does something go wrong on their side? Do they just have a thin fuse in general??
parsley936 has quit [Remote host closed the connection]
parsley936 has joined #nixos-chat
<pie_>
did i say something wrong ._.
<__monty__>
bqv: Have you gone over theory yet? Maybe leaving the technical aspects for later works better? It's usually how driving schools teach afaik.
<bqv>
__monty__: yeah, that's pretty solid, it's the practical parts
<bqv>
pie_: no
parsley936 has quit [Remote host closed the connection]
parsley936 has joined #nixos-chat
parsley936 has quit [Remote host closed the connection]
<__monty__>
bqv: Is there maybe an opportunity to include some sort of safe driving theory. Like an educational video of what happens if you pull on the steering wheel too hard or apply gas too suddenly? Maybe even discovery channel type videos of how a car works, including clutch/transmission?
<gchristensen>
sounds like they might benefit from some therapy for stress management
<pie_>
im not sure i understood right, are you correcting them before or during doing stuff?
<pie_>
*after or during
<adisbladis>
Don't do stuff, stop it.
<pie_>
:D adisbladis if only
<bqv>
I mean, I think I'll just leave it to pro instructors
<__monty__>
Unless they end up with an instructor like mine.
<adisbladis>
pie_: Haha :D
<pie_>
i wish i could stop doing stuff and relax, but i have to pay all the scarfolk fines from something :P
<adisbladis>
Also very good
<pie_>
(print more monopoly money)
<__monty__>
Instructors aren't psychologists they may not be equipped to deal with the issues. Ime the difference between professional and amateur instructors is the former tell you what to do from a position of authority and don't expect talkback.
<__monty__>
*backtalk?
<manveru>
sometimes more like fallout
<pie_>
dont expect you to talk back / dont expect backtalk
<pie_>
backtalk does sound kind of weird now that you mention it
<pie_>
i would say english is not german but...apparently actually it is or something?
<pie_>
bqv: i think its also kind of hard to tell whats going on from what youve said so...yeah. Easy to jump to conclusions. At least from my perspective.
<bqv>
I don't want this to be a "let's psychoanalyse this person" but also its slightly related so yeah I was a bit vague
<pie_>
right
<pie_>
but thats like...the obvious path as posed?
<bqv>
I was hoping I was just missing something obvious with teaching driving
<pie_>
mm. yeah, sanity check.
<adisbladis>
pie_: Holy crap, Scarfolk is a scam! You get stuck in an infinite loop of fines :O
aleph- has quit [Ping timeout: 246 seconds]
<pie_>
adisbladis: the real question is who fines the people that do not fine themselves
rardiol has joined #nixos-chat
<adisbladis>
pie_: Well, I guess no one
<adisbladis>
Who would fine you if they are also not allowed to do anything?
<drakonis1>
but the typical container libraries are in Go
<drakonis1>
reduced external dependencies are always great
cjpbirkbeck has joined #nixos-chat
<philipp[m]>
Nice! zfs encryption decreases the write speed for me just from 212 MB/s to 178 MB/s
<philipp[m]>
And with aes256-gcm it's even still 191 MB/s.
<__monty__>
With or without compression?
<philipp[m]>
Without compression.
<samueldr>
eyJhb: none yet
<srk>
ldlework: wasn't able to focus much last few days, some progress with pa + jack is at https://github.com/NixOS/nixpkgs/pull/93431, with that you could just run ardour DAW for mixing and feed what's needed to pulse sink (with alsa_in for microphone in). it's also achievable with just pulse e.g. https://github.com/toadjaune/pulseaudio-config but ardour gives you multitrack recording
<{^_^}>
#93431 (by sorki, 1 day ago, open): nixos/jack,pulseaudio: fix pulse connection to jackd service
<srk>
*pulse source :|
<samueldr>
eyJhb: but given the week-end just ended, it's not been a full business day since I think
<abathur>
at least it sounds like they need physical access to the charger for a moment
<samueldr>
(elsewhere) people have been saying this could end up fun in places with always-accessible chargers
<samueldr>
e.g. airports
<samueldr>
sure, we are unlikely to use them
<samueldr>
but other people will
<abathur>
when I first saw that headline I was assumed it was malware on the device, which could really be a pretty big problem if it was exploitable on many no-longer-updated androids
<abathur>
samueldr: yeah, I had that thought; I guess someone would need to separately test on those sorts of units. Not sure how many are fast chargers.
<samueldr>
or how many are fast chargers *yet*
<samueldr>
and speaking of malware on android
<samueldr>
it could be an entry point into chargers
<samueldr>
so e.g. user X has an old device that doesn't do PD, but that can still connect to the charger, and prime them into being bad chargers
<abathur>
on the up side, at least in airports I'd guess there's some rate limiting from how many people are going to risk "do something strange in a highly-surveilled setting", and how many different USB ports they'd feel like they could reach without attracting attention
<__monty__>
Are you talking about phone malware that exploits a fast charger that gets plugged into the phone to make it blow up the phone?
<__monty__>
o.o
<samueldr>
the last part is highly theoretical
<samueldr>
only confirmed in lab settings: pd chargers tha blow up the phone
<samueldr>
but depending on how it can be installed on the chargers, the theoretical thought of phones worming it along their way is fun
<samueldr>
fun as in FRIGHTENING, UNNERVING, NOPE
drakonis1 has quit [Ping timeout: 260 seconds]
<abathur>
yeah
<abathur>
I've been in nope mode for over a decade just due to concern about attacks against my device in a public USB outlet
ixxie has quit [Ping timeout: 256 seconds]
<abathur>
I suppose I'd admit that my opinion as softened on them a little since devices started defending against this, but not enough to risk it outside of a desperate scenario?
drakonis has joined #nixos-chat
endformationage has joined #nixos-chat
cole-h has joined #nixos-chat
<philipp[m]>
Any crypto people here that can explain to me why LUKS defaults to xts over gcm?
<gchristensen>
probably because it is stuck in time
<__monty__>
I do know my crypto prof was highly suspicious of GCM.
<philipp[m]>
Ah, because the MAC isn't part of the ciphertext and needs to be stored separately.
<philipp[m]>
This is why zfs can easily use gcm and luks has more trouble with it.
<__monty__>
philipp[m]: As opposed to with xts? Can't they simply store it alongside?
<philipp[m]>
They could, but then they'd need to reserve a spot in each block and I don't think they want to do that in the default operating mode, at least.
<drakonis>
i've been waiting on it reaching a somewhat usable state
<drakonis>
i bet it could do nix-y things
<drakonis>
rather, nixos itself
<__monty__>
Oh, I've been thinking about backup rotation. Does anyone use magnitude of diff rather than age? I.e., I've been thinking I'd rather keep a slightly older version if it's a lot more different. Mainly to prevent very regular backups from kicking out valuable backups through sheer quantity but very few differences among one another.
<__monty__>
drakonis: Maybe you could make things ever so slightly more immutable by having mgmt restore things whenever they change.
<drakonis>
hmm, an worthy attempt
<drakonis>
gonna try to achieve that with fedora
<drakonis>
hmm, how much work would it require to reduce the delta between mgmt and nix?
<drakonis>
how much redesigning would it require
<gchristensen>
the innovation in Nix is showing that ideas like chef/puppet/mgmt are all of the same flawed "stock"
<drakonis>
most definitely
<drakonis>
i see it as baby steps though
<gchristensen>
it is a fundamentally broken idea, it can't be fixed
<__monty__>
Hmm, but nix can't manage mutable state. Isn't that where config management might still come in?
<gchristensen>
otherwise we would have found a better idea than the same old, rehashed idea from before even CFEngine in 1993
<samueldr>
isn't that mutable state what's the fundamentally broken idea?
<gchristensen>
yes!
<samueldr>
any kind of layer on top of that fundamentally broken idea is not going to fix it
<gchristensen>
precisely
<__monty__>
But some mutable state is unavoidable?
<gchristensen>
yes
<gchristensen>
push it, maximally, to the edges
<__monty__>
So might not the systems designed to manage mutability help manage that mutability?
<samueldr>
is it state that needs to be _managed_? (and what does management of that state mean here?)
<gchristensen>
no, because they do a miserable job of it
<samueldr>
there are different kind of states
<gchristensen>
they don't know better. they have one hammer and it looks like FHS and perl
<gchristensen>
but instead of perl, yesterday it looked like ruby, and today it looks like ... uh, frankly, perl again
<samueldr>
some of it can be (likely) thrown away, like some of the runtime state
<samueldr>
some of it can (likely) not be thrown away, like the data of your database
<samueldr>
and then there's a whole infinite range in-between
ixxie has joined #nixos-chat
<gchristensen>
puppet's syntax even looks like cfengine's
<samueldr>
before working with nix properly, the only conclusion I could make from all those puppet^W cfengine-likes was that they were, at best, misleading and dangerous
<drakonis>
its a bit of throwing the baby with the mathwater here
<drakonis>
bathwater
<gchristensen>
I know the reference, but I don't know what you mean
<samueldr>
and my solution at the time was to have a bespoke system that completely *trashed* whatever the current state was (at best as it could) and reset it to new values
<samueldr>
miles ahead of trying to resolve the state, but still, archaïc and misleadingly dangerous
rajivr has quit [Quit: Connection closed for inactivity]
<__monty__>
mgmt could do the job of systemd timers but more sophisticated.
<abathur>
gchristensen: see my mention/question earlier re: resholved?
<gchristensen>
hmm no
<drakonis>
i'm saying that there are some things that you cannot solve with just immutability
<drakonis>
the world isnt ready for that yet
<gchristensen>
I think the world is ready for it
<gchristensen>
and is becoming more ready for it
<gchristensen>
it was surely not ready 17 years ago
* srhb
wonders if gchristensen just did 2020 minus 1993 -- except :-)
<samueldr>
I believe something important happened in 2003 srhb
<gchristensen>
(I can count backwards in increments of 7 ....)
<drakonis>
is this something to handle the long standing issue with using existing scripts and watching them fail because you didnt fix the paths?
<drakonis>
if that's correct, then, fantastic!
<drakonis>
holy cannolli
<abathur>
hopefully
<drakonis>
nice
<drakonis>
now that's the stuff
<abathur>
my personal use-case includes being able package interdependent shell projects
<gchristensen>
truly life changing
<drakonis>
it is very life changing
<manveru>
oh, i want that check for all my systemd units :D
<__monty__>
Is it a shim that intercepts shebangs for sh/bash?
<manveru>
so i don't deploy them and find out half the executables are missing again...
<gchristensen>
abathur: you and I been talking about this in here for weeks and they finally pay attention :P
<abathur>
so, like, my darwin-config builds a package for my bashrc, which depends on a package for a shell history thing I'm working on, which depends on like, eh, a shell "orchestrator" library I decided to factor out and release separately, which finally depends on an open source "promise" implementation for bash
<abathur>
gchristensen: heheh :]
<__monty__>
abathur: The first step to recovery is admitting you're a shell addict.
<abathur>
well
<abathur>
in my defense, if it wasn't for really wanting to fix my shell history
<abathur>
because it kept goddamn overwriting itself
<abathur>
I wouldn't be here
<srk>
I'm considering running shell via wrapper that would do the logging to db
<abathur>
gchristensen: anyways; I think I have a few ideas pulled together to consider the strict/debug-mode sort of use case; I'm bearish on whether it'll actually be able to catch much that resholved isn't already catching, but I wrote an early draft last night. It'll help to have some real targets to evaluate to what extent it's productive and/or a dead-end?
<abathur>
srk yeah, my history thing writes to an sqlite3 db
<abathur>
I may at some point re-evaluate dolt, when I first wrote it dolt didn't exist yet and nomz or whatever the precursor was had obviously gone stagnant
<srk>
interesting
<abathur>
it does use PROMPT_COMMAND, though, so presumably whatever thinking you have there applies to it as well
<abathur>
there are a small handful of database-based ones out there, though; presumably at least one uses the debug-trap-based bash-preexec approach
<samueldr>
is the .so loading approach something of interest for the purpose of logging history?
<__monty__>
What's your reasoning for a db for shell history?
<abathur>
__monty__: who?
<__monty__>
abathur: You : )
<__monty__>
Is it reliability or speed or querying?
<abathur>
a few things, but I think querying is probably the best umbrella
<abathur>
I'm not necessarily obsessed with ALL HISTORY EVER
<abathur>
some people who are basically un-cap their history
<abathur>
since mac Terminal can persist the scrollback, I keep a lot of terminal tabs for not-dead projects around; history is fairly fast to load into the browser but as the size grows it definitely does induce some noticeable lag
<abathur>
I'm not getting as much out of the DB yet as I'd like to, and haven't really touched or tried to bolt on any of the interactive history searchers that exist these days
<__monty__>
Ok, cool. So it's kind of about quantity but specifically keeping it useful?
<abathur>
but the main thing I get is that most invocations can go in the db, and I can let bash manage its own short/fast history files without worrying about their durability; if anything happens to one I can re-synthesize a short one (currently 500 entry, but even this is probably also overkill once I have a better search interface) from the db on demand
<__monty__>
Nice.
<abathur>
it's also got a little concept of like a "project", which I call a "purpose", so when I start a new tab/window, if I re-use an old "purpose", it'll synthesize a histfile for it based on the database even though it has no scrollback
<srk>
ash can do that based on directories, also filtering based on exit codes sounds nice
<srk>
I was thinking about tab completion based on history too
<abathur>
so I can in theory decide that I'm mad at some old project's tab taking up a few pixels of screen real-estate, toss the tab away, but still have a simple way to hop back in with at least the command history from whenever I last touched it
<abathur>
yeah
rardiol has quit [Ping timeout: 240 seconds]
<gchristensen>
cpu throttling on battery power, but with cgroup limits on browsers
<abathur>
srk: mine tracks the user, hostname, "purpose", pwd, start time, duration, pipestatus, un-expanded cmd (as entered), and expanded cmd (as presumably interpreted*)
<__monty__>
Licensing question. I'm talking to a company that releases a font which is free to use non-commercially. However, usually you have to register to download this font. They can provide a way to download the fonts without registering which would come with an amendment to the license explicitly allowing this on a per-platform/os/distro basis. I assume they want to provide per-distro urls to keep count
<__monty__>
of installs per distro. Would this be acceptable for a nixpkgs package? Marked unfree, of course. The current package in nixpkgs is technically violating their license afaiui.
<__monty__>
Maybe this is more appropriate for #nixos-dev?
<samueldr>
yes, more appropriate to #nixos-dev :)
<samueldr>
many devs/users of Nixpkgs and NixOS are on #nixos-dev but not in here
<__monty__>
Alright, I'll repeat my question there.
<samueldr>
hmm there may be an issue on the nixos hydra instance too
<__monty__>
That sounds like the discussion in #nixos-dev : )
<samueldr>
oops
<samueldr>
I misclicked
<samueldr>
it was actually a follow-up to a discusison there :)
<__monty__>
Getoutahere with your on-topic stuff!
<__monty__>
(ง'̀-'́)ง
<samueldr>
actually yes, it was a mistake
<__monty__>
Np, just figured you wanted to tell people there.
<abathur>
a couple weeks ago I had an entire conversation in nixops that I *thought* I was having in nixos-chat :[
<cole-h>
heh
<abathur>
didn't even notice until a couple days later when I was trying to find some reference in it and couldn't find it anywhere
<__monty__>
: )
<pie_>
i have a uh python question
<pie_>
if im using multiprocessing, and i try to call a function that changes the state of an object
<pie_>
does it uhh..change it in this process or the other process? because i want it to change it in the other process
<pie_>
ok it hit me
<pie_>
its probably fork and i probably need to start a second thread to do message passing and interpretation
<__monty__>
Yeah, if it's multiple processes it would be multiple interpreters, right? So no way to affect the state of another process except for message passing?
drakonis has quit [Ping timeout: 260 seconds]
<pie_>
well thats what i was scratching me head about
xd1le has quit [Quit: Quit]
drakonis has joined #nixos-chat
<pie_>
__monty__: next question is how do i make sure the right fork is consuming messages
<pie_>
im probably doing this all wrong
<pie_>
i imagine i have to somehow check pid
<__monty__>
Hmm, I don't know much about multiprocessing tbh. But when you call fork it returns twice. In the parent it returns the child's pid (or is it 0?) and in the child the parent's pid.
<__monty__>
No, pid of the child in the parent, and 0 in the child.
<__monty__>
So that way you can tell which of the processes is the child and go on from there.
<pie_>
hm maybe the problem is something else tho, i gotta poke around more
<pie_>
yeah im not calling fork directly but apparently with multiprocessing you can choose between fork and starting a new interpreter n stuff
<pie_>
and also apparently pass a process name
<pie_>
not sure that does what i think it does yet
<colemickens>
Well sure enough it happened. Even though APS readded totp, they didn't publish it, and now I'm out in the wild completely screwed because they thought it was cool to silently remove a critical feature and ship it and then not ship the revert. JFC.
<gchristensen>
aps?
<samueldr>
from the context, the android password store app
<colemickens>
It's all good. It's OSS that I'm very thankful for and it will be fixed in due time.
<samueldr>
one issue though is that even if you built it yourself, you can't replace the install on your phone without losing its data
<samueldr>
so it's kinda annoying in that sense
<samueldr>
(unless you're using something like titanium backup or whatever is used these days to backup app data)
<samueldr>
or uh, I guess it doesn't save those in the app data?
<ashkitten>
it saves its own settings in app data, it optionally saves the actual password store either there or on the user storage
<ashkitten>
fun fact: riot android seems to use some sort of hardware key storage, so you can't backup and restore the app data and expect it to work, it'll crash until you wipe its data
<samueldr>
yeah, those apps that do are annoying :|
<samueldr>
signal is one of them, so changing roms is quite garbage
<ashkitten>
might be a good thing depending on your willingness to trade convenience for security
<samueldr>
yeah
<samueldr>
exactly that
<ashkitten>
with cross-signing i don't have a problem with it, since i don't actually have to tell my friends anymore when i get a new device and go through a whole key signing process yet again with everyone
<ashkitten>
so really it's not that inconvenient to me anymore
<ashkitten>
it used to be that i'd want to try out a new matrix client and then go "ehhhhhh but then i'd have to tell all my friends who care, and refresh the fingerprints file on my website, and...."
<ashkitten>
but now it's really easy, on mobile i can even scan a qr code on another device
__monty__ has quit [Quit: leaving]
<pie_>
the gnu readline api seems kind of crappy?? how the heck do i redraw the buffer
<pie_>
or is the line buffer something the terminal does
parsley936 has quit [Remote host closed the connection]