<samueldr>
I hate vague issues about a vague problem that gets hijacked months after months with, to me, probably or obviously unrelated more or less vague issue about the same thing
slack1256 has quit [Ping timeout: 246 seconds]
<samueldr>
it ends up being totally unactionable, and a complete mess of clues for people with one of the issue that gets dumped into the same truckload :(
<samueldr>
if only github allowed to split a ticket into a new one, like discourse does for threads, it would be more manageable
<gchristensen>
anyone on theaccess list of #nixos-mods can +o channels with that set
<gchristensen>
so now srhb, niksnut, and I
<ashkitten>
is it reasonable to pick out a specific subset of paths to mirror from cache.nixos.org if i were to set up a lan cache?
<ashkitten>
i know it's more complicated because it's not actually a package repo
<gchristensen>
you might just want a squid reverse proxy cache with a disk budget
<ashkitten>
hm
<ashkitten>
we are going to set up squid, i'm just not sure how useful it will end up being
<ashkitten>
i guess it'll probably be fairly useful since my system is set to auto gc
<ashkitten>
so for paths i frequently redownload
<ashkitten>
(like the closure for discord, since i refuse to actually put the client in my nixos config)
<gchristensen>
I .. don't feel good about running that program
<ashkitten>
yeah, but i have to because video isn't supported in firefox
<ashkitten>
and jitsi meet has problems actually maintaining stable video
<gchristensen>
maybe unshare your ~...
<ashkitten>
are you suggesting it's snooping in my files?
<gchristensen>
all I know is its CEO is shady as heck and that it takes a lot to be subject to a class action suit for fraud and invasion of privacy
<ashkitten>
ah
<ashkitten>
anyway gchristensen did i mention i have 10Gbit fiber in my apartment now?
<gchristensen>
no.
<gchristensen>
take it back
<ashkitten>
but it's so overkill!
<ashkitten>
it was actually... fairly cheap and easy?
<ashkitten>
we got a lot of 6 fiber ethernet cards on ebay
<ashkitten>
and some 10Gbit sfp+ modules
<samueldr>
and here I'm proud of having recently done gigabit all the way around
<gchristensen>
I'm not even sure I have consistent gigabit :x
<abathur>
here I am hot-gluing microSD cards to pigeons to beat a gigabit
<colemickens>
gchristensen: whoa I must've missed some Discord news!
<gchristensen>
colemickens: discord's ceo's prior company
<colemickens>
interesting. I'm thinking about your "it takes a lot", but reading this synopsis, I wouldn't be surprised to find lots of companies are doing the same thing. I wonder what got them caught in particular
<samueldr>
I never said I had consistent gigabit! I think I do though
<gchristensen>
colemickens: that is why I assume it takes a lot
<pie_>
MichaelRaskin: didnt you say datamatrix can do arbitrary sizes?
<pie_>
Im trying to use dmtxwrite and its being annoying
waleee-cl has quit [Quit: Connection closed for inactivity]
<bqv>
Hmm
<bqv>
Xmonad is a haskell app
<bqv>
Taffybar is a haskell app
<bqv>
Whats stopping me from just adding a taffybar thread to my xmonad config
<bqv>
I don't want a separate bar app, but I still want a bar
<pie_>
MichaelRaskin: huh. apparently datamatrix, or at least this reader, handles a squished datamatrix just fine
<MichaelRaskin>
pie_: -s should allow you to set a size out of a reasonably diverse list of options
arahael1 is now known as Arahael
<MichaelRaskin>
Only specific sizes are feasible, and the data should actually fit in the available bits, of course
<MichaelRaskin>
DataMatrix has some nice properties w.r.t. squishing, yeah
<MichaelRaskin>
I do have a feeling that libdmtx is plain better at recovery than generic open-source QR-code libraries. But of course that time when I had to write custom code to recover damaged QR-codes, the counterparty was laughably bad at scanning, too, so maybe I do not have a fair comparison.
<MichaelRaskin>
(The one time I needed to mass-recover DataMatrix codes was with them printed with top row cut off…)
<Arahael>
Wouldn't the simplest way to recover those be to have someone re-draw the calibration dots?
<Arahael>
Those orientation dots, whatever they're called.
drakonis has quit [Ping timeout: 246 seconds]
<MichaelRaskin>
If we were talking about «someone», we could just have people type the numbers printed nearby…
<MichaelRaskin>
(which we did for the few percents of most annoyingly damaged codes)
<MichaelRaskin>
But we kind of wanted to avoid large scale manual entry where we could…
<MichaelRaskin>
BTW, repainting the dots is what my code did first (well, after general image-registration stuff to find the code), it… was not enough
<Arahael>
Sad.
<bqv>
I guess i'll just try it
<MichaelRaskin>
Arahael: the scanning was done _hilariously_ wrong
<MichaelRaskin>
Half of the scanners had very local automatic black/white threshold detection, i.e. most of the area of the anchor would get recognised as uniformly coloured and therefore white
<MichaelRaskin>
I mean, sure, I repainted the anchors, but they were not the only damage.
<infinisil>
c74d: Yeah, those aren't very limiting imo :P
FRidh has joined #nixos-chat
<elvishjerricco>
With macos working on the same chips as iOS devices, I wonder if jailbreakers will figures out how to get macos running on iPhones and iPads. That'd be pretty cool
<MichaelRaskin>
Or Apple makes them the same system in a couple of versions…
<elvishjerricco>
They're on record saying they don't want to merge the platforms. Though that may have just been a statement for the time, since they would have been years away from accomplishing such a merge at the time.
<srk>
sphalerite: yup, iirc we've had to enable hidepid for lxc host in shared env
<energizer>
i want something like Dropbox for my personal stuff but i have enough data that i'm in business pricing realm which is too expensive. is there an alternative?
<energizer>
youd think everyone would be offering something like this, but google still doesnt
<Arahael>
energizer: You could either sync to s3 directly, or are you willing to host the data yourself?
<energizer>
i've been self-hosting since dropbox temporarily stopped supporting whatever filesystem i was on a few years ago
<energizer>
but ive had enough of that
<Arahael>
energizer: What is the problem you have with selfhosting?
<Arahael>
energizer: How much data are we talking about?
<energizer>
Arahael: large time commitment, poor uptime, poor bandwidth, moderate risk of permanent data loss, poor cross-device syncing behavior
<energizer>
Arahael: around 3TB
<Arahael>
Weirdly, I don't feel comfortale recommending "s3 sync" for that size.
<energizer>
too big or too small?
<Arahael>
s3 sync is curiously naive.
<Arahael>
On the other hand, if you could mount an s3 volume into an ec2 instance, you could run syncthing on that, perhaps.
<Arahael>
There's also backblaze, which I've never tried.
<Arahael>
Oh, wow - Owncloud has s3 backends. That's another option.
<energizer>
i dont trust syncthing
<energizer>
in the reliability/quality-of-service sense, not the security/privacy sense
<Arahael>
Yeah, I personally trust syncthing more than dropbox, but I'm not going to argue with youa bout that, this is going to be awfully subjective, and 3TB is a lot of data and I don't know what this data is.
<Arahael>
That said, I'd probably trust dropbox more than owncloud. :(
<energizer>
it may be subjective but dropbox has spent thousands of engineer hours working on edge cases; syncthing hasnt and it shows
<Arahael>
Good thing about s3, though: You can let Amazon manage those backups nicely. S3 itself isn't going to be the problem - getting the data there and in particular sync'ing it, will be the issue.
<energizer>
also syncthing requires me to have all my data on my laptop but it wont fit there
<Arahael>
Ah, right, and that's the other thing there. Syncthing is clearly biased towards a typical linux system.
<Arahael>
I use syncthing for my personal stuff, but it's *not* my backup system. I do my backups with rsync.
<energizer>
right ive been using syncthing and rclone on my systems for a long time
<energizer>
im just over it
<Arahael>
Owncloud, and their clones, are the only thing I've found that compete with dropbox. :(
<Arahael>
In that specific niche.
<Arahael>
I mean, there's google drive, and iCloud, but...
<energizer>
the only "but" is no linux client
<energizer>
otherwise google drive would be fine
<c74d>
there's nixos.drive for Google Drive
<energizer>
same problem as Box
<c74d>
I use it
<energizer>
does it have a service module c74d?
<Arahael>
It's a real pity that 's3 sync' is so naive. :(
<Arahael>
The big problem with s3 sync, is that it has to iterate and touch every single file in order to sync it.
<c74d>
I don't think so, but you could write the systemd equivalent of a cronjob to have it "push" a directory up on a schedule if that's what you want.
<Arahael>
And I think it's one-way, only.
<energizer>
tbh that's like saying its a pity that scp is naive
<energizer>
it's just not even in the same league
<Arahael>
energize
<energizer>
arahae
<Arahael>
energizer: Right, I mean, it's difficult to even find an alternative sync'er.
<Arahael>
energizer: s3 has the advantage that aws themselves will ensure that the data doesn't get lost.
<Arahael>
Plus, you can do naive per-file versioning, too. Almost dropbox style. (Originally, I think dropbox was backed by s3, but they use their own solution now, I think?)
<Arahael>
I can see there's s3fs-fuse as well, but I don't think I could seriously recommend that.
<Arahael>
It's a hardware appliance, from the looks of it, though
<srk>
git-annex might be an option
<energizer>
that's one i havent spent much time with
<Arahael>
Oh, you can run the storagegateway as an VM appliance (eg, using Linux KVM), as it turns out.
<Arahael>
Complex.
<Arahael>
srk: What's the UI for git-annex like? I'd not normally suggest git for non-devs - not saying that energizer is not a dev, but I assumed the discussion was for a non-dev audience?
<energizer>
i'm a dev
<energizer>
git-annex-assistant has a web ui
<Arahael>
Yeah, but I assumed that the discussion here was for a non-dev user case, ie, that this wasn't simply your personal stuff, but perhaps your organisation or family.
<Arahael>
ldlework: ipfs, isn't everything there public, though?
<joepie91>
ldlework: IPFS is basically more granular bittorrent
FRidh has quit [Ping timeout: 265 seconds]
FRidh has joined #nixos-chat
sdier[m] has left #nixos-chat ["User left"]
<ldlework>
I'm not confused about the core concepts or even the fundamentals of how multihashes and so on work
<ldlework>
It's the ecosystem that is baffling
<joepie91>
ah right
<joepie91>
that I can agree with :)
waleee-cl has joined #nixos-chat
FRidh has quit [Quit: Konversation terminated!]
cjpbirkbeck has quit [Quit: cjpbirkbeck]
slack1256 has joined #nixos-chat
slack1256 has quit [Remote host closed the connection]
arahael1 has joined #nixos-chat
Arahael has quit [Ping timeout: 240 seconds]
arahael2 has joined #nixos-chat
arahael1 has quit [Ping timeout: 260 seconds]
<abathur>
oh, hmm, I missed a git-annex discussion, however brief; that doesn't happen often
neeasade has quit [Remote host closed the connection]
<evanjs>
Ugh I need to start my own hydra or _something_ already
cole-h has joined #nixos-chat
<evanjs>
If only so I can get (nixpkgs) master builds for stuff lazily, if not set up a channel that gives me stuff from unstable and that falls back to stable automatically for packages as needed
<evanjs>
e.g. in the case of https://github.com/NixOS/nixpkgs/issues/89481, my machines would end up pulling down stable.YouCompleteMe, but the channel would be abstracted from such
<evanjs>
like it would just pull _whatever_ the server produced
<evanjs>
Tired of switching to stable kernel when VBox breaks, etc
<evanjs>
Though, that case in particular is interesting to consider
<evanjs>
When NVIDIA or VBox don't build, what do I do? I have my machines set to the latest kernel. Do I update that to say "latest kernel considering NVIDIA, VBox, and any other applied kernel modules"?
<evanjs>
That seems like it might make the most sense. Hrm. </rubber duck> lol
rajivr has quit [Quit: Connection closed for inactivity]
<abathur>
evanjs I've wondered (different solution, same problem) if there could be a metadata system that enables expressing something like "I'd like to update my channel, but only if the packages I depend on aren't broken" (or perhaps: but only as far as I can update without any of the packages I depend on being broken?)
<evanjs>
exactly that. It's mainly VBox and NVIDIA (kernel modules), but often times other packages will be hit as well. rustraced and Discord come to mind as of recent breakages
<MichaelRaskin>
Of couse, of those only VBox and rustraced have Hydra data
endformationage has joined #nixos-chat
<energizer>
i want a network fileshare for my personal data. do i want cifs, smb, or something else?
<KarlJoad`>
energizer: I would start by asking the question of what hosts you need to serve.
<energizer>
my PCs
<cole-h>
Host OS, they probably meant
<cole-h>
Windows? Mac? Linux? BSD?
<KarlJoad`>
I meant the OS you are serving to.
<energizer>
linux, android
<KarlJoad`>
As far as I know, if you aren't going to be using Windows clients, then say away from CIFS/SMB.
<energizer>
why is that?
<KarlJoad`>
From my VERY limited knowledge, other network file sharing systems are better.
<energizer>
maybe i want nextcloud but i'm concerned that it has a php backend and *video chat*
<KarlJoad`>
I can understand the video chat, but why's the PHP backend so bad?
<energizer>
it's not that php necessarily causes it to be bad; but it's a signal
<KarlJoad`>
I still don't understand why it's a signal, but OK.
cole-h has quit [Quit: Goodbye]
maxdevjs has joined #nixos-chat
hoverbear has joined #nixos-chat
maxdevjs has quit [Client Quit]
maxdevjs has joined #nixos-chat
maxdevjs has quit [Remote host closed the connection]
<aleph->
Okay, teleport service is almost setup. Nice
<samueldr>
there are multiple accounts syncing already
<samueldr>
I don't want to "get started", and DEFINITELY don't want a reminder
<hoverbear>
But you might want a new one!
<samueldr>
and it's not like getting started has an easy way out
<samueldr>
it's either scan a QR code or send an SMS
<samueldr>
good, now what if I wanted to sync between laptops?
<samueldr>
aaaaand, they must have never tested this, or broke it
<hoverbear>
Admittedly I fix that by openinga new tab and just keeping going
<samueldr>
(oh, it unbroke)
<hoverbear>
Can I make SPectacle upload to an s3 bucket?....
<samueldr>
sure, but since it wants to remind me, maybe it will if I ignore it
<samueldr>
these are dark patterns
<samueldr>
and again mozilla has proven they are not above using them :(
<hoverbear>
Based on my interactions with dozens of Mozilla employees over the past 5 years I would say everyone agrees that the leadership/marketing at Mozilla Corp are very, very out of touch with the rest of the foundation.
<samueldr>
that's a sad thing to hear :(
<hoverbear>
But then again the executive is too busy rolling in cash to notice...
<samueldr>
I mean, good to see people *are* caring, but sad to see the corp doing this
<hoverbear>
samueldr: Mozcorp CEO compensation was 2.5mil in 2019... It was under 400k in 2008. Their marketshare dropped 20% during this time.
<hoverbear>
(I can only talk about the CEO since other numbers aren't published)
<hoverbear>
:shrug: I have mixed feelings about Mozcorp. I love the other teams at Mozilla dearly though <3
neeasade has joined #nixos-chat
<aleph->
:/
<hoverbear>
They're kind of in a bad financial situation. In 2008 most of their money came from papa Google
<hoverbear>
Now it's more diverse, which is good :)
KarlJoad` has left #nixos-chat ["sLEEP"]
parsley936 has quit [Remote host closed the connection]
<gchristensen>
hoverbear: I'd like to write an async program which schedules things to run a certain time period in the future, and where these timers could be canceled
<gchristensen>
any fav. libraries?
<hoverbear>
How long in the future? ns? us? ms? years?
<gchristensen>
minutes
<energizer>
what language ?
<colemickens>
I would give so much money to make kryptco actually OSS
<aleph->
Heh saw that q in #python
<aleph->
Thinking, think there's a timer/cron lib you can use
<arahael2>
gchristensen: minutes is far enough in the future to be a bit tricky, but not far enough that you need o worry about weirdities involving timezones, calendars, leap seconds, etc.
<aleph->
gchristensen: N minutes from proc start time?
<aleph->
Or from a predetermined timedate?
<aleph->
s/timedate/datetime/g
<gchristensen>
N minutes from when an event enters stdin
<arahael2>
ah, yes - the key question: using a datetime offset... or a timespan from "now".
<aleph->
gchristensen: Could maybe use threading.timer
<aleph->
Just wondering how to get that to work event-based...
<aleph->
Maybe as a callback func~
<arahael2>
how many of these events do you have? if only a few... spawn a new thread and just sleep.
<aleph->
I'm out of touch with event-loop stuff now. So not really sure there.
<aleph->
Yeah what arahael2 said might work
<gchristensen>
arahael2: oh that is a cute idea!
<aleph->
Oh now this reminds me of some multi-threading stuff in python at work I saw
* aleph-
has PTSD flashbacks of that code
<aleph->
Probably the worst bit of code I've ever seen
<arahael2>
gchristensen: keep in mind that unless the language has ultra efficient threads, such as haskell, that will be an expensive solution, however if there are only a few not so bad.
<aleph->
^^
<arahael2>
aleph-: could always be worse.
<hoverbear>
gchristensen: I agree with arahael2
<aleph->
arahael2: When I say multithreaded/proc. Here's a loop, let's subprocess.run() some other python files in the dir
<hoverbear>
So you'd have one timer and poll over it checking to reduce load.
<gchristensen>
oooh
lejonet has quit [Ping timeout: 256 seconds]
<elvishjerricco>
TIL about `boot.binfmt.emulatedSystems`. Now I'm doing the pointless exercise of building aarch64-linux NixOS with `--no-substitute` just to see how well it works and how long it takes :P