<adisbladis>
qyliss: Did it run successfully before?
<gchristensen>
spacex.com/webcast launc a minute ago
<gchristensen>
or now?
<gchristensen>
aborted
<__monty__>
Too bad.
<__monty__>
Weather?
<gchristensen>
seems the engines were unhappy
<__monty__>
Burn them!
<MichaelRaskin>
Engine unhappiness has ways of becoming contagious, though
<gchristensen>
adisbladis: reading reviews and comparisons of those two CPUs show 3900x >>>> 2920 except for things where the memory bandwidth is useful, and I'm still not sure :P
<adisbladis>
Very hard to say
<adisbladis>
gchristensen: Have you checked openbenchmarking?
<MichaelRaskin>
Bavaria: from Mon 16th all schools are closed. Also Bavaria: Sun 15th is the elections, no changes about that. University: All retakes are cancelled. Also University: nothing changes in administrative work, you know, where at-risk age staff takes physical paper documents from a ton of people.
<MichaelRaskin>
It is hard to beat «we will screen all people from EU flights by putting them into a long line in a crammed space for a few hours», though
<gchristensen>
yeah that was really amazing
<MichaelRaskin>
We-ell
<MichaelRaskin>
Not really amazing
<__monty__>
How about the alleged attempt to bribe CureVac to move from DE to the US?
<gchristensen>
I was amazed ...
<MichaelRaskin>
It was a task of the same people who did the entire «we will confiscate random liquids under pretext of the liquids being possible explosives, and drop them in a thin-metal urn in from of a long queue we have created»
rardiol has quit [Quit: No Ping reply in 180 seconds.]
<MichaelRaskin>
Like, _how_ security theater people can produce anything _not_ defeating the stated purpose?
rardiol has joined #nixos-chat
<gchristensen>
:
<gchristensen>
:)
<ar>
>It is hard to beat «we will screen all people from EU flights by putting them into a long line in a crammed space for a few hours», though
<__monty__>
I think originally they were just naive and thought it'd help but then they saw 🤑 and figured "what the hey."
<ar>
MichaelRaskin: that sounds like what happens regularly when landing at SFO
<adisbladis>
US airports are pretty unique... You have to go through immigration even for transits
<MichaelRaskin>
Well, this time the claim to be screening for a highly contagious disease
<adisbladis>
That's not the case in pretty much the rest of the world
<adisbladis>
And even for transit you need a visa
<adisbladis>
Mind boggling...
<MichaelRaskin>
Indeed. In Riga, I had even a transit situation where arrival and departure _were_ separated by Schengen-only zone without having a visa, well, even this wasn't really a problem
<MichaelRaskin>
__monty__: I think you ascribe too much caring to people running the security theater, both before and after
<__monty__>
What, they keep it up just to mess with people, not for the money?
<MichaelRaskin>
They keep it up to inflate warm-body-count under their command, obviously
<__monty__>
I doubt everyone involved realized the financial potential in advance though.
<MichaelRaskin>
Yeah, bureaucratic was clearly more obvious
<MichaelRaskin>
As for _everyone_ invlolved, well… The line officers probably just think that they are really unlikely to be in a situation to shoot an unarmed person, so, better job than police.
abathur has joined #nixos-chat
abathur has quit [Ping timeout: 255 seconds]
<__monty__>
Curious, felt like checking /etc/password on a mac. It has a ton of users defined. Among which things I don't even have installed indirectly afaik, like dovecot and ClamAV. Weirdest thing is my user's nowhere to be found in the file.
<LnL>
osx uses a separate database for users
<LnL>
dscl . -list /Users
<LnL>
it's ldap under the hood I think
<__monty__>
Yeah, figured. And have actually run into that before. Still kinda strange there's users defined for so many processes that I've never had running though.
<__monty__>
I assume it has to do with apple blessing certain apps?
<LnL>
there are a bunch of builtin users, generally prefixed with an underscore, but not sure where those you mentioned would come from
<__monty__>
Also weird how /etc/passwd is kinda kept in sync but not quite.
waleee-cl has joined #nixos-chat
<LnL>
I'm guessing that's probably manual for unixy things that depend on getent, etc.
abathur has joined #nixos-chat
<LnL>
eg. id nixbld1 looks up the user through pam/opendirectory but getent passwd nixbld1 doesn't
<viric>
how can I know the closure of a nix-shell?
cole-h has joined #nixos-chat
<eyJhb>
There is a dvorak layout for programmers, right?
<eyJhb>
What is the difference between that and normal Dvorak? (US)
<gchristensen>
the number row I think
<gchristensen>
24680 then 13579 for some reason
<viric>
hand mandates the last bit
<eyJhb>
So, it is the number row which is switched?
<MichaelRaskin>
Well, 0 and 1 are probably standing out even more in programming than in normal use
drakonis has joined #nixos-chat
<infinisil>
eyJhb: The main thing is that without shift, the number row are symbols
<infinisil>
not the numbers
rardiol has quit [Ping timeout: 258 seconds]
<infinisil>
&[{}(=*)+]!# <- my number row without shift
<infinisil>
%7531902468` <- my number row with shift
<cole-h>
Poor modulo :(
<infinisil>
The number thing is a bit weird, but I'm used to it now
<etu>
eyJhb: You're confusing Dvorak with Simplified Dvorak
<infinisil>
Symbols without shift is awesome though
<etu>
eyJhb: Simplified Dvorak has the number in numeric order while Dvorak doesn't
<etu>
infinisil: yeah, the other one kinda... never took of I think
<infinisil>
I've been wanting to make my own layout, with super key combos and prefixes and stuff
<infinisil>
But that gets really complex and it's hard to know where to start
slack1256 has joined #nixos-chat
<__monty__>
I think the original dvorak number row only makes sense if you type a lot of numbers.
<__monty__>
Which makes it kinda curious that programmer dvorak kinda stuck with the idea.
<MichaelRaskin>
If you use computers to, oh the heresy, compute?
<__monty__>
Not sure, for serious number work I suspect the numpad is the way to go. I was thinking more of historians who have to talk about dates a lot or maybe drafters who have to list a ton of dimensions or product managers who have to handle IDs with mixed numbers and letters.
<MichaelRaskin>
I think numpad is best for one-handed work
<MichaelRaskin>
Another hand either for mouse or for paper documents
<__monty__>
Yes, "documents"...
<__monty__>
#numberphileisarespectablechannel
aleph- has quit [Ping timeout: 258 seconds]
<ashkitten>
programmer dvorak has numbers mixed in with the letters doesn't it?
<ashkitten>
i figured that was unnecessary so i just learned normal dvorak
<ashkitten>
(i gues simplified)
<__monty__>
No, they're on the number row, but even->uneven, rather than 1-9,0 or just mixed according to frequency like original dvorak.
<ashkitten>
oh i see
<ashkitten>
yeah anyway given the amount of times i use the number row for things that are not numbers, i'm glad i chose simplified dvorak
<gchristensen>
I nope'd out when I saw the number sreordered :P
aleph- has joined #nixos-chat
<gchristensen>
anyone familiar with microphones and setup for recording grand pianos? I'm looking to do live music in the age of COVID-19 with a medium-modest budget :)
<MichaelRaskin>
That's how implied opinions get on record
<MichaelRaskin>
samueldr: looking at Korea's scale, I would expect that they just traded off getting things done vs full formalities and started using whatever sequencing machinery they could grab.
<samueldr>
new measure just announced, publicly-funded day care are opened exclusively for workers in the health networks, and first responders, and supporting staff, even for those who were not using those services
<samueldr>
(without additional costs)
<MichaelRaskin>
Sounds like a natural measure to avoid collapse
<samueldr>
yep
<samueldr>
MichaelRaskin: I was thinking more about how I've heard china has been testing, the swab only happens once a quick CT scan of the lungs has been made to check for telltale signs
<samueldr>
they have CT equipment that allows making it in-situ, rather than bringing the patient in a big thing
<samueldr>
those machines allow order of magnitude more patients to be screened for the next step
<samueldr>
here our equipment can't be used that way
<samueldr>
I wonder if korea has the same kind of equipment that helps not using the limited testingcapacity
<MichaelRaskin>
I think for zero-level sorting of the low-suspicion crowd they just used human judgement as the first step.
<MichaelRaskin>
But it looks like they did a ton of real tests, too
<__monty__>
adisbladis: Doesn't cryptpad or etherpad do that?
<danderson>
list of APIs, toolkits, and various on-tablet linux hacks
<adisbladis>
__monty__: As in not on an LCD
__monty_1 has joined #nixos-chat
__monty_2 has joined #nixos-chat
__monty_2 has quit [Client Quit]
__monty_1 has quit [Client Quit]
<danderson>
looks like there's basically no boot lockdown, just a question of finding the magic door to get into the shell :)
<danderson>
and other people have figured it out, it seems
<danderson>
looks like people have figured out ssh access, so from there the world is yours :P
<adisbladis>
Sweet :)
<danderson>
just need a tiny bit of SDK to support the weird eink refresh model
<danderson>
because it's trying to be like paper, you can erase specific pieces of the page
<danderson>
which can cause some weird artefacting for 1-2s while the OS does the full e-ink page refresh to remove things
myskran has quit [Ping timeout: 250 seconds]
<danderson>
it's kinda cool, at first it looks like you've used an eraser on pencil writing. The writing goes away, but there's still a faded outline visible for a bit
<__monty__>
Hmm, weird. I thought that was an inherent attribute of e-ink. That you could address specific regions.
<danderson>
then the tablet does the full eink refresh, and it goes away
<danderson>
it is fully addressable, but to get a "perfect" deletion you have to toggle black-then-white a couple of times
<__monty__>
Ah, complete clearing requires a higher voltage or something?
<adisbladis>
Ohh, looks like you can even use Qt <3
<danderson>
yeah, something like that. I don't know the details
<adisbladis>
I told myself I wouldn't buy more stuff any time soon....
<danderson>
but the tablet tutorial says "hey when you switch pages or erase stuff, your screen might flash black briefly. That's normal, that's how eink works"
<danderson>
adisbladis: I will say, the tablet is *really* expensive for what it it.
<danderson>
so, it's definitely a luxury.
<danderson>
if I had to give it up now, I would probably not miss it in 2 weeks. But now that I have it, I use it every day.
<adisbladis>
That's the other thing I'm afraid of, that it would end up getting unused.
<danderson>
and I know people who use it to read RFCs and scientific papers, and they love it for that. Like a hackable kindle.
<adisbladis>
At 500 USD I really wanna be certain
<danderson>
so far I use it only for notes. Although I have to write a couple of blog posts soon, and I might try drawing diagrams on the tablet as well
<danderson>
yup, it's a difficult purchase. Can't really test it, so you have to kinda jump and hope you like it...
<danderson>
they do have 30 days no-questions-asked returns, so that's one option
<danderson>
but be careful, with the pandemic right now their shipping was very delayed
<danderson>
I ordered it early Feb, it arrived early March
<danderson>
so if you decide to try it, and shipping is slow, contact their support and tell them you want to change the start of the 30 days to when the tablet ships
<danderson>
they default to "when you click the buy button"
<danderson>
in my case, it took ~30 days from that point until they shipped it, so...
<{^_^}>
puppetlabs/vault-plugin-secrets-oauthapp#4 (by grahamc, 10 seconds ago, open): `oauth2: server response missing access_token` for the GitHub provider
<gchristensen>
adisbladis: in case you were looking at that
<gchristensen>
FYI: if a user PM's you about being unbanned, please don't engage. If you have questions, I can answer them.
myskran has joined #nixos-chat
myskran is now known as abathur
__monty__ has quit [Quit: leaving]
<gchristensen>
social distancing question -- is it inappropriate to go visit my brother in his home, 1:1?
<samueldr>
depends of your social proximity before and after I guess
<samueldr>
plural your
<danderson>
"social distancing cohorts" - small groups that distance from other groups but mingle with each other - has a minimal impact on the effectiveness of distancing
<danderson>
so, depends on the total number of people you both interact with
<gchristensen>
good point, I'm not sure how strict he's going to be
<samueldr>
interact_ed_ and _will_ interact
<gchristensen>
right. he's seen a lot of people.
<MichaelRaskin>
For future interactions, you might want to also weight by age
<cole-h>
gchristensen: (moving here cuz mostly not Nix related) How does the recommended `atime=off` setting affect things in general?
<cole-h>
The wiki notes "might confused some software like mailers", but was wondering if you'd run into any issues related to that (or if you use atime=off at all)
<gchristensen>
I use atime=off always
<danderson>
I run all my filesystems with noatime, and I've never had any problems
<danderson>
if you want a halfway measure, use relatime instead
<gchristensen>
*some* software uses atime for logical purposes, but as far as I can tell it is almost exclusively atime
<gchristensen>
*some* software uses atime for logical purposes, but as far as I can tell it is almost exclusively mailers
<danderson>
it's an optimization that updates atime way less frequently
<danderson>
so you still get roughly useful atime info, without nuking the filesystem updating atimes
<MichaelRaskin>
I wonder if relatime is the default by now
<cole-h>
Any way to see if my current fs uses atime=off (ext4)? Been a while since I set it up...
<cole-h>
Oh
* cole-h
just looked in /etc/fstab and saw noatime
<MichaelRaskin>
Heh
<cole-h>
That answers that, I guess. No real reason for me to be worried
<gchristensen>
it turns out atime is a bit of a bad idea
<cole-h>
And apparently you can change compression algo on the fly-ish, so that's another concern dealt with
<MichaelRaskin>
Good idea, but unscalable
<cole-h>
(I want to use zstd once it merges)
<gchristensen>
back in 2h
<gchristensen>
nice to chat about fire suppression, danderson :P
<danderson>
in general, ZFS responds to config changes by applying the new config to newly written data, but not touching the old data
<danderson>
so you need to do something expensive like zfs send | zfs receive to a new dataset if you want to "upgrade" all your data
<danderson>
which is kinda annoying, but understandable. They want to prioritize not destroying the system's I/O bandwidth with maintenance tasks
<cole-h>
Yeah, that's what I meant with the `-ish` -- apparently you can `zfs send` and `zfs receive` to recompress
<cole-h>
According to a stackexchange site
<danderson>
yup. But my take is: probably not worth it for many situations
<danderson>
unless you have highly compressible data and it's suuuper expensive to store (e.g. an all NVMe flash array)
<danderson>
I'd just eat the loss and only optimize the new data
<danderson>
the problem with send|receive, is that you need 2x the storage while you do that, I believe
<cole-h>
Oh, true. Something I didn't think of.
<cole-h>
Magically compress data in-place >:(
<danderson>
you can't make it do the work in place, you have to construct a new dataset on the side, cut over all users, and then delete the old one
<danderson>
Weirdly, there's no reason zfs _couldn't_ do it in-place. It's already a COW system, so it'd just be a question of writing out new blocks (with the new settings) and updating the hash trees
<danderson>
and then garbage-collecting the old data
<danderson>
but ZFS is developed by pretty specialized storage companies, and they don't need that... So it's not in there
<danderson>
from their perspective, it's okay to say "I'll just send|receive everything to the new storage box and decommission the old one"
* colemickens
swears the irc bridge drops messages sometimes, or the scrollback is confusing and I'm missing something
<danderson>
which bit are you missing?
<colemickens>
Trying to figure out what you had shipped?
<colemickens>
Something about a tablet, maybe just an old conversation that I caught the end of, just being nosy.
<danderson>
oh, right, that conversation happened across 2 channels + twitter
<danderson>
I'm shitposting on twitter about setting up nixos on my laptop, and I posted some shots of my setup notes
<danderson>
I'm taking notes on a reMarkable tablet (remarkable.com), an e-ink tablet you can write on
<danderson>
its claim to fame is that its screen texture is very paper-like, so it feels like writing on paper, except it's not paper :)
<danderson>
and being eink, you can actually just scribble as if it were paper. There's no perceptible lag between writing and the screen updating to reflect your write, because the e-ink display can toggle invidual pixels very rapidly
<danderson>
compared to a traditional LCD's refresh rate
<danderson>
(individually addressable pixels, so instead of redrawing the whole display, the tablet just sends a stream of "darken pixel X, Y, Z, A, B, ..." to the display
<danderson>
It's got a fairly dedicated niche following as a device. Some people love it for reading PDFs and stuff, like you'd use a kindle. Except you can scribble notes as you read.
<danderson>
I use it entirely for note-taking right now, and have used it daily since I got it 2wks ago.
<danderson>
it's also quite hackable (it's basically linux with a fancy SDK for doing eink-appropriate graphics)
<danderson>
so people have built various syncers, open source SDKs for making new eink-friendly apps, there's SSH access, etc.
<danderson>
I haven't tried doing any of that yet, but it's a nice touch.
<danderson>
colemickens: hopefully that helps fill out the context :)
<danderson>
I'm very happy with it, but two downsides of note: it's quite expensive for what it is ($500), so unless you're going to use it frequently, I'd say it's not worth it
<danderson>
and the stylus has consumable tips. To get the paper-like writing feel, the display is slightly abrasive, and the pen tips are a soft-ish plastic that ablates away as you write