<infinisil>
Not really documented, but you can customize the arity of the tree. E.g. with 2 it could store `[ 2 7 4 3 0 ]` as `[ [ [ 2 7 ] [ 4 3 ] ] [ [ 0 null ] ]`
<infinisil>
Insertion and access is done with an index
<infinisil>
(above would be a binary tree, but I found that with an arity of 4 it's a bit faster for AoC day 15)
<gchristensen>
the samuela and samueld(r) is tripping me up
<__monty__>
rycee: Can you ban that spammer from #home-manager?
* samueldr
is confused
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-chat
__monty__ has quit [Quit: leaving]
<siraben>
samueldr: there's some spammer on the HM channel
<samueldr>
I didn't think they were related though
<zgrep>
infinisil: Darn. I hit a stack overflow.
<zgrep>
I assume there's no tail call optimization of any kind here? :P
<infinisil>
zgrep: Hint: use builtins.seq to evaluate attributes, so you don't accumulate thunks pointing to thunks pointing to thunks..
<infinisil>
E.g. let foo = { bar = 10; baz = <some nested computation>; } in builtins.seq foo.baz foo
<zgrep>
Mm. Hm.
* zgrep
is not used to lazy languages.
<zgrep>
I... is there a way for me to see what nix has thunked? Or some way for me to figure out what would be causing the stack overflow? (Other than knowing how nix works/evaluates, because I don't. :P)
rajivr has joined #nixos-chat
<infinisil>
zgrep: Hm you could try a bunch of -vv's
<infinisil>
Or --trace-function-calls
<zgrep>
Ah, I'm probably using the wrong command here.
<infinisil>
Though actually probably neither of those will help..
* zgrep
has been doing `nix eval -f 15.nix`.
<zgrep>
At the very least it runs for part 1 up to 2020 without problems! :D
<infinisil>
Nice!
<zgrep>
It's just that 30000000 is apparenly a large number.
<zgrep>
apparently*
<infinisil>
It do be pretty big
<infinisil>
Also recommended: using builtins.trace to debug/print progress
<zgrep>
Mm. I shall try that.
<zgrep>
I can get up to 11330 before it stack overflows.
<zgrep>
Heh. It doesn't print any of the traces when there's a stack overflow.
<zgrep>
Is there a non-stack-overflow-y way to just call a function on its own output a bunch of times?
<zgrep>
Hmm. I guess there might be...
talyz has quit [*.net *.split]
lukegb has quit [*.net *.split]
Raito_Bezarius has quit [*.net *.split]
betawaffle has quit [*.net *.split]
glowpelt has quit [*.net *.split]
{^_^} has quit [*.net *.split]
AMG has quit [*.net *.split]
lucus16 has quit [*.net *.split]
jackdk has quit [*.net *.split]
infinisil has quit [*.net *.split]
gchristensen has quit [*.net *.split]
joepie91 has quit [*.net *.split]
wirew0rm has quit [*.net *.split]
{^_^} has joined #nixos-chat
lukegb has joined #nixos-chat
talyz has joined #nixos-chat
jackdk has joined #nixos-chat
gchristensen has joined #nixos-chat
lucus16 has joined #nixos-chat
glowpelt has joined #nixos-chat
infinisil has joined #nixos-chat
betawaffle has joined #nixos-chat
wirew0rm has joined #nixos-chat
AMG has joined #nixos-chat
joepie91 has joined #nixos-chat
Raito_Bezarius has joined #nixos-chat
<zgrep>
Hmm. Now I get memory exhausted.
<zgrep>
When I give it 10000. But not if I try to give it 30000000. ._.
betawaffle has quit [Ping timeout: 249 seconds]
betawaffle has joined #nixos-chat
<zgrep>
Ooh, okay, I have something that does something.
<zgrep>
But is slow. I want to see how long I'd have to theoretically wait.
<zgrep>
This isn't scaling linearly. :(
<infinisil>
Hehe, yeah it's pretty tricky
<zgrep>
It does 100000 in ~3.6 seconds. It does 200000 in ~13.3 seconds. ._.
guest24024024024 has joined #nixos-chat
guest24024024024 has quit [Remote host closed the connection]
<zgrep>
I don't see why it's not scaling linearly, though. :(
<zgrep>
Maybe I'm making assumptions about data structure speeds.
<infinisil>
zgrep: using // creates an entirely new attribute set, copying both sides into it
<zgrep>
Mmh.
<zgrep>
You are very good at guessing what I'm not saying.
<zgrep>
I assume there aren't any data structures that can be updated in place?
<zgrep>
Or, well, hm.
cosimone has quit [Quit: cosimone]
* zgrep
can't think of a way that isn't O(n), because "updating" any data structure probably amounts to copying it.
<zgrep>
Err, O(n²) total.
<zgrep>
O(n) per iteration.
<zgrep>
I assume I'm just not well versed enough in functional data structures.
<pie_>
are there tools for invoking windows stuff from nix? like idk spinning up a windows vm, running an executable, returning its results to the store
LnL has quit [Ping timeout: 246 seconds]
LnL has joined #nixos-chat
LnL has joined #nixos-chat
LnL has quit [Changing host]
hl has quit [Ping timeout: 256 seconds]
aleph- has quit [Ping timeout: 260 seconds]
aleph- has joined #nixos-chat
hl has joined #nixos-chat
hl has quit [Changing host]
hl has joined #nixos-chat
lunc has quit [Ping timeout: 264 seconds]
<elvishjerricco>
pie_: I wonder if wine would be the easiest way to do it.
<pie_>
elvishjerricco: probably. assuming the thing works in wine
endformationage has quit [Quit: WeeChat 2.9]
<lovesegfault>
Debugging an issue that only happens after 8h of runtime
<lovesegfault>
I hate it
<energizer>
can you change the system clock to make it happen sooner?
<lovesegfault>
Nope, not that kind of bug
<lovesegfault>
I'm interfacing with a device, and after around 8h of usage it triggers he issue (on the device)
<lovesegfault>
unclear why it takes that long
maljub01 has quit [Quit: maljub01]
maljub01 has joined #nixos-chat
<pie_>
lovesegfault: have to wait for the bug timer to overflow to 0 :P
<lovesegfault>
:D
Dotz0cat_ has joined #nixos-chat
Dotz0cat has quit [Ping timeout: 264 seconds]
<pie_>
elvishjerricco: yeah the thing i want doesnt work in wine :P
<andi->
pie_: there are runInWindowVM functions in nixpkgs
<pie_>
every time i try to free up more disk space or ram it fills up faster than I can do anything xD
<pie_>
need mo money for mo hardware >:V
<aszlig>
samueldr: what's the reason it sucks up so much disk space?
<samueldr>
windows?
<samueldr>
heck if I knew
<aszlig>
yeah
<cole-h>
samueldr: I'd think that's the answer
<samueldr>
I just know that by passing
<cole-h>
"Windows"
<samueldr>
I think they now have a simili-A/B scheme that allows them to undo big upgrades
<samueldr>
requiring them to have a copy of multiple files in the system files
<samueldr>
but that's the limits of my knowledge
<aszlig>
samueldr: you mean winsxs?
<samueldr>
not sure if that's it
<aszlig>
or wait, wasn't that scrapped eventually?
<samueldr>
since winsxs was for multiple libraries side-by-side (sxs) no?
<samueldr>
I don't know what I'm thinking of is called, I don't use windows :)
<samueldr>
I only barely remember some factoids like that thing for allowing downgrades when things do go wrong
<aszlig>
samueldr: me neither :-)
<aszlig>
i only need to need to deal with its quirks because *other* people have issues with it
<samueldr>
I *did* install windows earlier this year on a tablet/laptop device of mine to confirm that something worked as badly as on Linux, then promptly dd'd back my system to NixOS
<aszlig>
but IIRC they still have some rollback-mechanism to some extent
<samueldr>
>> Microsoft announced that it would begin using ~7GB of user hard drive space for the application of future updates.
<samueldr>
while I don't really keep up-to-date, sometimes I end-up reading bits of infos and it sticks
<aszlig>
well, you need to go with the times: people have more bandwidth -> bloat your websites, people have more disk space -> dump more crap onto it... =)
lovesegfault[m] has quit [Quit: Idle for 30+ days]
thibm has joined #nixos-chat
cole-h has quit [Ping timeout: 256 seconds]
FRidh has quit [Ping timeout: 268 seconds]
<steveeJ>
infinisil: thanks, I'll take a look at poetry2nix
waleee-cl has joined #nixos-chat
hello has joined #nixos-chat
hello has quit [Remote host closed the connection]
<eyJhb>
The no-exec has not been a problem until now, that I am doing C++ stuff...
<LinuxHackerman>
`mount -t tmpfs tmpfs $HOME/bin` and always build to ~/bin ?
<LinuxHackerman>
Or build everything with nix
<eyJhb>
LinuxHackerman: I have my /nix-builds, that I use for building NixOS stuff so I am just placing it there. However I could actually just build this using NixOS if I wanted to :D
<eyJhb>
What is the correct way to not get Nix to do make install at the end?
<eyJhb>
Whops
<__monty__>
Override the installPhase? Or remove it from phases?
<srk>
isn't that the default for installPhase?
<srk>
"In fact, for Unix packages that use the standard <literal>./configure; make; make install</literal> build interface, you don’t need to write a build script at all; the standard environment does everything automatically. If <literal>stdenv</literal> doesn’t do what you need automatically, you can easily customise or override the various build phases."
<srk>
</literal>
<eyJhb>
Yeah, I of course needed to override the installPhase...
srk has quit [Remote host closed the connection]
srk has joined #nixos-chat
<srk>
maybe pre or postInstall?
<eyJhb>
Just doing the installPhase worked fine, but I might as well have added it to the Makefile
<srk>
maybe pre or postInstall?
<eyJhb>
Now I just get a `Floating point exception (core dumped)`, yay..
<LinuxHackerman>
eyJhb: the "right" approach is setting `dontInstall = true;`
FRidh has joined #nixos-chat
aleph- has quit [Ping timeout: 260 seconds]
neeasade has joined #nixos-chat
Blackraider has joined #nixos-chat
Blackraider has quit [Remote host closed the connection]
cosimone has joined #nixos-chat
abathur has joined #nixos-chat
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos-chat
cosimone has quit [Quit: cosimone]
cosimone has joined #nixos-chat
FRidh has quit [Ping timeout: 256 seconds]
FRidh has joined #nixos-chat
Jackneill has quit [Ping timeout: 240 seconds]
cosimone_ has joined #nixos-chat
cosimone has quit [Read error: Connection reset by peer]
cosimone_ is now known as cosimone
Jackneill has joined #nixos-chat
aleph- has joined #nixos-chat
cosimone has quit [Remote host closed the connection]
bqv has quit [Ping timeout: 260 seconds]
ashkitten has quit [Ping timeout: 260 seconds]
<bbigras>
Maybe I would have more success getting into cloudflare betas if I didn't input "for fun" as my usecase.
<andi->
why would you want to use a single point of failure?
ashkitten has joined #nixos-chat
<bbigras>
I don't see it as a worst thing than gcp or aws.
<andi->
my point exactly ;)
<andi->
(tbh do whatever is fun)
<bbigras>
I would think a cloudflare pages with or without workers could be pretty reliable as long as the whole cloudflare is not down. People seem to overreact all the time, like when google was down the other day. It's still pretty reliable to me. and more reliable than my home internet connection and my electricity. and I don't flip the table every time it goes out.
<andi->
My point is more that I do not like the trend of there being 1-3 "big players" hosting 99% of the services.
<bbigras>
yeah I wouldn't mind if there were more players.
<andi->
bbigras: did you by chance upload your nix files to radicle the other day? I was browsing nix related repos the other day and your name sounds familiar.
<bbigras>
haha yeah
<andi->
have you used it for anything else?
<bbigras>
no. I just saw that we had a PR and wanted to try it.
<bbigras>
you saw my files on the discovery-thingy website?
<adisbladis>
Do you know how I can log in to an existing account?
<adisbladis>
Or whatever you should call it
<adisbladis>
It seems the UI only allows me to create a new identity at startup
<bbigras>
I didn't know repos were public without sharing the hash or whatever. but in my case it's fine since it's also on github
<andi->
bbigras: always assume it is public if you put data in a 3rdparty :)
<andi->
adisbladis: no idea
<andi->
I think that might only work ifyou have the local data backed up
<bbigras>
andi-: oh yeah. I assumed it could be accessed but not published on a public page.
<adisbladis>
That definitely wasn't obvious...
<bbigras>
with some p2p app, your identity is like a key pair. I wonder if it's the case with radicle.
<andi->
adisbladis: I think it is just a private key on your disk (and more?) that is encrypted at rest.
<adisbladis>
Presumably yes
<andi->
yeah, it is just a key in ~/.config/radicle/keys
<andi->
but as usernames do not collide there is probably little loss if you do not have your data anymore at this stage.
<adisbladis>
Oh god... I'm going to have to learn Rust because of this..
<andi->
very good :)
<adisbladis>
I refuse to use anything that uses software based keys
<bbigras>
yes.. come to the dark side.
<adisbladis>
Need to add hw token support
<andi->
adisbladis: just wrap it in luks with a yubikey?
* andi-
hides
<adisbladis>
andi-: Key material should never reside on disk or in unencrypte memory
<adisbladis>
Oh well, I guess it's a good place to start contributing =)
<bbigras>
can we really trust those hardware keys? what if they were backdoors! /me put tinfoils hat
<andi->
best to not require cryptography.
<adisbladis>
:D
<adisbladis>
Hmmmmm
<bbigras>
but seriously the other day I saw that we had a tpm2 module. I think we can store keys in it and use it as a number generator. but people don't seem to recommend/trust it.
<andi->
you generally only add random generators never switch to one
* adisbladis
now has 2 key related features in mind for radicle:
<adisbladis>
- hardware tokens
<adisbladis>
- mnemonics for software keys
<adisbladis>
I'll probably just end up implementing bip39
<andi->
bbigras: I would love to use mine more for all kinds of stuff just have to do it.. ETOOMANYPROJECTS
<bbigras>
I saw some people using tpm for full disk encryption without a passphrase. I guess crazy people with hardware could extract the key, but it might be good enough for if we lose our laptop.
<andi->
how is that good enough if it decrypts itself?
<bbigras>
I think it's assumed that our user password (in the os) is good enough. and people can't change the boot (like to enable single user mode) or boot with a usb stick since the key wouldn't be read by tpm then. maybe it involved secureboot too.
<sphalerite>
well the workingness of secure boot on nixos is still… limited
<bbigras>
I'm not 100% sure but I think I heard that windows do that.
<sphalerite>
yes, that does make sense, but we don't have it for nixos yet.
<sphalerite>
Unless you use Graham's unmerged PR.
<bbigras>
yeah. maybe we should bribe him or something.
<sphalerite>
bribing someone to do work is called "paying" them
<sphalerite>
:p
<sphalerite>
I mean tbh all that's missing, AFAIK, is more testing.
<sphalerite>
danielrf[m]: are there any remaining issues with the tests you wrote, IYI?
<bbigras>
sphalerite: yeah I know about this PR. I'm the annoying guy who pinged everyone multiple times.
<sphalerite>
oops :D
<andi->
I just sign my grub and everything up from there is (at least) encrypted (probably not real authenticated crypto but good enough for me)
<bbigras>
haha. each time I was afraid of being rude. I mean, we all have more projects than time and everyone has the right to spend their precious time the way they want.
<sphalerite>
andi-: secure boot signing, or do you have a custom firwmare or something?
<andi->
secure boot signing
<andi->
I keep one special efi entry on each device and do it manually every blue moon
<andi->
haven't bothered really automating that yet
<andi->
mh, I wonder if there is a modal mode for weechat so I can have the same keybindings as in qutebrowser & neomutt..
<bbigras>
the PR radicle was merged 2 days ago. We might get it in unstable for Christmas.
<andi->
Nice. I just spent my morning (and night) creating a source build for it.
<andi->
But that will sadly not make it into nixpkgs as it requires reading the lockfiles during eval...
<andi->
some people really hate adding those to the repo and instead prefer IFD nonsense..
<andi->
s/IFD/FOD/
<bbigras>
oh. is the PR using the source or an like an appimage?
<andi->
appimage
<andi->
if you have radicle: rad:git:hwd1yrergedjioote6dg1d5q4hdmqiaaqgm39gpm1prwgam44tdse6bpiec
<bbigras>
I'll build radicle right now, well the package.
rajivr has quit [Quit: Connection closed for inactivity]
<bbigras>
I saw that it sets some special git remotes. so is it a daemon that we run at all time?
<andi->
well they are one thing right now IIRC
<bbigras>
or we need to keep the app open.
<bbigras>
gotcha. thanks
<andi->
so there is the "proxy" radicle-proxy that appears to do the heavy lifiting
<andi->
and then there is the electron app with two "threads" (security contexts?) where one is the unpriv. web thread that has IPC to the priv thread that has IPC to the proxy IIRC
<andi->
only briefly looked at that today
<adisbladis>
andi-: Sounds like a very reasonable design :)
<andi->
Yeah. I didn't mange to inject any code into the browser so far (but that doesn't mean other can't)
* adisbladis
imagines using radicle + magit-forge and drools
<adisbladis>
andi-: I started auditing their smart contracts the other day
<adisbladis>
Planning to do a bit more of that this weekend
<andi->
I haven't had any motivation to dive into that dark corner of the internet
<adisbladis>
If you audit the rust code I'll audit the solidity ;)
<andi->
ugh, maybe I'll stick to patches via email then.
<adisbladis>
Because?
<andi->
I do not feel competent enough to judge whatever crypto foo they are doing there.
<adisbladis>
Ahh, right
<adisbladis>
I used to do this for a living
<andi->
I mean I've already implemented spec'ed out algorithms and had that beeing reviewed and there were no huge issues so I kind think I am capable of writing *some* cryto code but auditing something that you haven't studied in depth is something different.
<bbigras>
andi-: I just followed your radicle-upstream-nix repo
<andi->
Great just do not dig too deep :P
<andi->
the yarn.lock parser is gross
<bbigras>
oh it's the radicle source pkg. if it was your nix-config I would be forced to go through all of it.
<samueldr>
github search was actually good at finding things google forgot
<adisbladis>
Wait what
<adisbladis>
How could github code search possibly get any _worse_
<samueldr>
read it, it's short enough
<adisbladis>
It's already really, really bad
<adisbladis>
samueldr: I read it
<samueldr>
and it actually was not bad
lunc has quit [Ping timeout: 256 seconds]
<samueldr>
for my uses
<drakonis>
it indexes only a year a now
<samueldr>
in fact it was pretty good for what I did
<adisbladis>
I can kinda see why
<drakonis>
repositories that have been active for at least a year
<samueldr>
take a textual string that looks unique enough from a binary
<adisbladis>
Having elasticsearch indices you don't use is $$$
<samueldr>
search it with quotes on github search
<samueldr>
and many times it surfaced a repo!
<adisbladis>
=)
<samueldr>
like actually trustably many times
<adisbladis>
samueldr: I also use it, but for a search engine it's not very good
<samueldr>
for searching what you don't know of already, yeah
<samueldr>
for searching a specific needle that is pretty unique, it was amazing
<samueldr>
because neither google, nor bing would find anything relevant
<samueldr>
so now we have yet another amnesia level on the web
* samueldr
sighs
<samueldr>
for android-adjacent development it was a treasure trove
<samueldr>
I found multiple unique kernel repositories with unique drivers for specific hardware
<samueldr>
where the OEM wasn't respecting the GPL
<samueldr>
now it's basically impossible
<joepie91>
uggghhhh
<joepie91>
well that fucking sucks
<joepie91>
"This change will enable the most relevant content for developers to surface in the code search index as well as keeping code search queries fast for all customers."
<joepie91>
it reduces your costs, you mean
<samueldr>
"the most relevant content"
<samueldr>
implying that recent == relevant
<samueldr>
which for code is... wrong
<srk>
exactly
<joepie91>
samueldr: the entire thing is a disingenuous argument, "prioritize by recency" would have accomplished the same thing
<samueldr>
it already was
<samueldr>
so it's a useless change
<drakonis>
searching already annoyed me because it would point to an older commit
<andi->
I guess GitHub has to watch it's spendings now as it was acquired? Wasn't there something else they changed recently as well that smelled like it?
<srk>
quite the opposite with actions
<andi->
waay after actions I think
<bbigras>
it might be a good idea to exclude stale repos. When I'm searching for a potential dependency I don't want an abandoned one.
<samueldr>
actions are marketing spending
<samueldr>
since you vendor lock-in users, and bring in users
<samueldr>
search is a liability, especially since it's commonly meme'd that it's bad and terrible and useless
<srk>
bbigras: and how do you define abandoned one?
<samueldr>
so since it's bad and terrible and useless it's not going to be missed
<andi->
srk: stable, working, untouched for years :D
<samueldr>
(except it is not)
<srk>
andi-: :D
<bbigras>
something like no commit for 6 months. 6 months in dog years is a lot
<samueldr>
bbigras: no; most of my useful github searches were for abandoned years-old repos
<andi->
samueldr: have you looked at the code at github.com/actions as some part of the runner are OSS but i suspect the scheduler isn't
<bbigras>
samueldr: wut, you are looking for old repos?
<samueldr>
not exclusively
<samueldr>
but I want to actually search everything
<samueldr>
it's for "code archeology", in a way
<samueldr>
android vendors are terrible
<samueldr>
you can often find drivers shared between unrelated repos
<samueldr>
to do so, a good method is to search for a unique string
<samueldr>
for relatively-unpopular devices it's been extremely useful
<bbigras>
it could be nice to have an option to search old repos but I think I would like a fresh-repos search by default.
<samueldr>
andi-: I know you can get most of the way there, but then if you want to cross-operate you'll have a big wart in the shape of `.github` :/
<samueldr>
andi-: and yeah, you're in a situation where to be as-useful as actions you have to draw the rest of the owl
<samueldr>
bbigras: yes, if it gave a tool to ask for a span of recency it'd be better
<bbigras>
yeah
<samueldr>
>> andi-: I know you can get most of the way there, but then if you want to cross-operate
<samueldr>
oops
<andi->
and get their internal API stuff.. IIRC dependabot is completly decoupled from actions while it is basically marketed in the same breath
<samueldr>
copy-paste fail
<samueldr>
>> Recent activity for a repository means that it has had a commit or has shown up in a search result
<andi->
Lets search all the repos?!?
<samueldr>
riiiiight, stable software will disappear from search
cole-h has joined #nixos-chat
<danielrf[m]>
sphalerite: Responding to your question about the secure boot testing, the only remaining issue (I can recall) was better organization.
<danielrf[m]>
At some point in the last few months I did some more work to rebase those changes, but I never got around to finishing it
<danielrf[m]>
It's still on the TODO, but I don't know when I'd be able to get to it :)
FRidh has quit [Quit: Konversation terminated!]
<bbigras>
Would a github star bribe on one of your choice repo help with that rebase?
<bbigras>
one of your repo*
lunc has joined #nixos-chat
<danielrf[m]>
While temping, I've also been super busy recently and I'm not sure a star would give me extra time
<danielrf[m]>
Maybe during the Christmas break I can take another look
<samueldr>
is a bribe a legal contract that means you need to do the deed?
<samueldr>
could I bribe a politician to do something and sue them in court if they didn't?
<danielrf[m]>
I've also been wanting to finish my systemd boot counting PR, which also like 90% done
<andi->
right, when I woke up, some 12h ago I wanted to do something.. how could I get side tracked like this /o\
<bbigras>
boot counting? is that about the boot an old generation on boot fails?
<danielrf[m]>
I guess i'd have to return the github star if I didn't complete the task :)
<danielrf[m]>
bbigras: yep!
<bbigras>
sweet :) I probably saw you mentioning it the other day. I didn't know it was a thing.
<eyJhb>
samueldr: I would not recommend to try
evanjs has quit [Ping timeout: 264 seconds]
<samueldr>
"google is better at code search than github"
<samueldr>
literally the first search link I followed is a 404
<ashkitten>
after steam relaunches itself for an update, does launching it with the steam command or the launcher icon or whatever cause a new instance to launch for anyone else?
<eyJhb>
Somebody mentioned Facebook and Google sharing data. I just got this on my Youtube start page, after I sent that Gif a couple of hours earlier on Messenger - https://i.imgur.com/eM7E3tn.png
<eyJhb>
Might just be change, but fuck that is creepy. I am not even signed in, it is a fresh FF instance
<bbigras>
I watched that video today. But I didn't share it on facebook.
<bbigras>
I watched it because the gf is watching the serie for the first time and I wanted to know if she saw that scene yet.
<eyJhb>
bbigras: Twinsies?
<eyJhb>
AHh :D
<eyJhb>
I have a person in my group at Uni that is not that great at English, and she was feeling sick, so she didn't want to do group work physically with us, because as she said "I am feeling dangerous" :D
<infinisil>
andi-: Oh nice I was hoping somebody would do that!
<andi->
I'll probably write a very brief blog post with the whole set of graphs
<samueldr>
it's... it's a nice graph
<infinisil>
andi-: What's up with the drop at the end?
<samueldr>
heartening to see people mostly sticking
<samueldr>
I guess partial month of december?
<andi->
infinisil: december isn't over
<infinisil>
Ahh
cosimone has quit [Quit: cosimone]
<andi->
I am not sure what kind of historic markers I can add there.. I have the switch from SVN to Git (apparently Eelco used git before there was the public repo?) and the Nix releases so far.. not sure what would make sense for "orientation". Perhaps the date we added ofborg and rryantm?