HomeSearchPrevious dayNext day

00:52:49lunik11 quits [Quit: :x]
00:53:27lunik11 joins
01:05:07Exorcism quits [Quit: Ping timeout (120 seconds)]
01:05:07DigitalDragons quits [Quit: Ping timeout (120 seconds)]
01:05:36lukash98 quits [Quit: The Lounge - https://thelounge.chat]
01:05:40Exorcism (exorcism) joins
01:05:40DigitalDragons (DigitalDragons) joins
01:38:15s-crypt is now known as help
01:38:22help is now known as s-crypt
01:40:07<steering>Today I learned: two bits is 25 cents https://en.wikipedia.org/wiki/Shave_and_a_Haircut
01:41:22<steering>I would've guessed two nickels, or two dimes, not two 12.5 cents
01:56:01hexa- quits [Quit: WeeChat 4.4.3]
01:57:39hexa- (hexa-) joins
02:22:31datechnoman (datechnoman) joins
02:43:43lukash98 joins
02:46:34lukash98 quits [Client Quit]
02:51:42<immibis>nukke: Hetzner doesn't run deals, period. They'd say their prices are low every day, and there's also the server auction.
02:53:28<hexa->they drop setup fees now and then
02:54:14lukash98 joins
03:01:30linuxgemini (linuxgemini) joins
03:31:43BlueMaxima quits [Read error: Connection reset by peer]
03:32:23Exorcism quits [Client Quit]
03:32:23DigitalDragons quits [Client Quit]
03:33:10Exorcism (exorcism) joins
03:33:19DigitalDragons (DigitalDragons) joins
03:50:35AK quits [Ping timeout: 260 seconds]
03:51:06etnguyen03 quits [Remote host closed the connection]
03:53:08<nicolas17>MEGA launched a password manager, lol
04:01:07<nukke>immibis: yeah that's pretty much what I'm reading.
04:01:27th3z0l4_ joins
04:01:36<nukke>I'm still trying to decide between hetzner and ovh. I forgot linode was acquired by akamai so I was confused why akamai kept getting mentioned in posts
04:02:08<nukke>ovh is cheaper due to black friday deals, but it's hard to find specifics about the hardware they offer
04:02:33<nukke>I was also one of the unlucky people that got affected by the france datacenter fires :')
04:03:25th3z0l4 quits [Ping timeout: 260 seconds]
04:11:19StarletCharlotte joins
04:12:11<immibis>IMO hetzner is considered equally reliable and professional as the rest of hetzner, while OVH's cheapest servers are standard bottom rung VPSes where who knows what you get
04:12:34<immibis>black Friday deals are also cheaper, so decide if you want cheap or good
04:13:37<immibis>I got two more ovh black Friday vpses to add on to my dn42 network
04:17:31<StarletCharlotte>Heya
04:25:00riteo quits [Ping timeout: 260 seconds]
04:36:30<nukke>o/
04:40:40<nukke>immibis: am I misreading, or did you mean to name something else besides "hetzner"?
04:40:47<nukke>akamai maybe?
04:41:05<immibis>hetzner is low cost but also stable, ovh is not
04:41:11<nukke>oh gotcha
04:41:37<immibis>at least in my perception
04:41:42<nukke>yeah I'm going with hetzner. the one annoying thing is that they ask for a lot of personal information and payment information _before_ even buying anything
05:08:11Exorcism quits [Client Quit]
05:08:11DigitalDragons quits [Client Quit]
05:09:22<immibis>I meant Hetzner{'s cheapest options} are no worse than the rest of Hetzner's options, apart from being smaller, obviously
05:09:31DigitalDragons (DigitalDragons) joins
05:09:36<immibis>that's how that was meant to be parsed
05:09:41Exorcism (exorcism) joins
05:13:20Fusl (Fusl) joins
05:13:20@ChanServ sets mode: +o Fusl
06:52:53riteo (riteo) joins
07:13:06Exorcism quits [Client Quit]
07:13:06DigitalDragons quits [Client Quit]
07:14:33DigitalDragons (DigitalDragons) joins
07:14:33Exorcism (exorcism) joins
08:04:54pixel (pixel) joins
08:10:10qwertyasdfuiopghjkl2 quits [Ping timeout: 260 seconds]
08:42:24qwertyasdfuiopghjkl2 (qwertyasdfuiopghjkl2) joins
09:30:35<@JAA>This looks neat: command-line JSON viewer with regex search, object/array collapsing, etc. https://jless.io/
09:37:49AK (AK) joins
10:30:18le0n quits [Quit: see you later, alligator]
10:35:04<steering>it's quite sad that ENUM (phone-number-dns) is basically forgotten
10:40:43le0n (le0n) joins
10:40:43yasomi quits [Quit: ZNC 1.9.1 - https://znc.in]
10:41:08yasomi (yasomi) joins
10:52:00le0n_ (le0n) joins
10:53:30le0n quits [Ping timeout: 260 seconds]
11:03:59BornOn420_ (BornOn420) joins
11:06:14sralracer (sralracer) joins
11:07:44BornOn420 quits [Ping timeout: 276 seconds]
11:15:45mls quits [Quit: leaving]
11:38:39<immibis>silly idea: a flame graph of youtube titles. what are the most common prefixes?
12:00:01Bleo182600722719623 quits [Quit: The Lounge - https://thelounge.chat]
12:02:47Bleo182600722719623 joins
12:33:49BennyOtt quits [Quit: ZNC 1.9.1 - https://znc.in]
12:38:08BennyOtt (BennyOtt) joins
12:39:54BornOn420_ quits [Remote host closed the connection]
12:40:29BornOn420 (BornOn420) joins
12:48:05<joepie91><immibis> yeah you described what i said
12:48:45<joepie91>the point I was trying to make is that what constitutes 'a room' is a large collection of protocol design decisions, types, API endpoints, and so on - and the stateres, the part governed by the room version, is only one tiny part of that
12:49:34etnguyen03 (etnguyen03) joins
12:49:59<joepie91>specifically, the part that you cannot change without breaking compatibility, at a very fundamental level, and that is why it has its own versioning scheme
12:50:37<joepie91>which is why it doesn't make any sense to complain about room versions; the alternatives are either a) you literally cannot change the consensus mechanism, and now you cannot fix any security concerns that come up later, or b) the entire Matrix spec needs to be incompatibly versioned which is worse
12:51:48<joepie91>in other words, complaining about room versions means having unrealistic expectations about what is actually possible in a protocol, because one way or another, you will need a way to make changes in the consensus mechanism, and those changes fundamentally carry a high risk of breakage, there is no design choices you could make in a protocol that would make that not be true
12:51:56<joepie91>choice*
13:30:00<immibis>the alternative is that some server is authoritative for each room
13:30:16<immibis>the NAME of the SERVER is part of the room name, so why isn't that server authoritative?
13:30:51<immibis>the other alternative is that all servers are allowed to have different locally-defined moderation rules on each room, like in usenet
13:31:27<masterx244|m>and then you get desync/different data depending on what server gives the data over federation
13:37:56<immibis>yes, that's how usenet works. if I block some spam, I don't give other servers that spam either.
13:38:06<immibis>why would i give the spam to other servers if I know it's spam
13:38:28<immibis>and if my server is marking non-spam as spam, i should use a different server
13:58:45<joepie91>immibis: the server isn't authoritative because that means that it is a single point of failure which makes the protocol functionally useless for general public use; people do not have the time or energy to do a deep dive on the history of a random homeserver in a federation and its likelihood of remaining available before creating a community (which is what a room is), and so the end result is that people either drop out because it's too
13:58:45<joepie91>much uncertainty, or they create their community on a random homeserver which goes down 3 months later, murders their community, and *then* they leave and never come back
13:58:59<joepie91>there is a reason approximately nobody seriously uses XMPP MUCs
14:00:15<joepie91>so no, this is not an alternative; it is not an alternative within the technical constraints because it requires giving up a core design property of the protocol, and it's not an alternative in a practical context either because the implications for availability and accessibility fucking suck and nobody wants to deal with that
14:03:21<joepie91>it's easy to design a protocol if you can just disregard things like availability, accessibility, and usability; but in real-world protocols those things do actually need to be designed for and that is why shit like room versions exists
14:09:00nulldata (nulldata) joins
14:20:15sludge quits [Remote host closed the connection]
14:20:27sludge joins
14:21:00Froxcey_ joins
14:21:00Froxcey quits [Read error: Connection reset by peer]
14:22:15Froxcey (Froxcey) joins
14:23:00Froxcey_ quits [Read error: Connection reset by peer]
14:23:10<immibis>you're using irc right now
14:23:12Froxcey_ joins
14:24:16Froxcey quits [Remote host closed the connection]
14:25:23<joepie91>and?
14:40:18<immibis>it doesn't have these features
14:41:42<joepie91>... and?
14:55:28<immibis>i guess IRC is not a usable chat platform
14:57:44<joepie91>to the general public, no, it is not, which is why it is dying
14:59:29<masterx244|m>joepie91: true on that since even a bad timed short outage in the middle of a conversation eats context due to the nonexistant message history
14:59:40<masterx244|m>or the classic netsplit case
15:04:49<immibis>they aren't using matrix, either
15:05:05<immibis>all this homeserver stuff. and what if you create your account on a random homeserver which goes down 3 months later, murdering you?
15:05:09<immibis>then you leave and never come back
15:05:27<f_>joepie91: TBH this has advantages and disadvantages
15:05:27<immibis>the ideal protocol is one where an oligarch spoon feeds the user exactly what makes the user feel good
15:05:43<immibis>that's why they're called a user
15:06:41<joepie91>okay, I think it is clear by this point that you're just going to keep throwing things at the wall to see what sticks because you hate Matrix, and no reasonable discussion is possible here
15:13:08<f_>joepie91: I think advantages of this is that there's not a single point of failure
15:13:48<f_>THe disadvantage is .. painful, takes ages (this is why joining takes ages), takes lots of space on the matrix homeserver.
15:13:56<f_>*disadvantages are
15:19:10<immibis>no, just according to your arguments, there is no reason considering any protocol which isn't owned by one person who guarantees that all content will stay up forever
15:19:58<joepie91>f_: there are disadvantages like the space use, but eg. the slow joins seem to be more an implementation issue rather than a fundamental one
15:20:54<f_>joepie91: Implementation yes, I did hear of sliding sync though
15:22:08<joepie91>the whole sliding sync etc. feels to me like shitty workarounds for a bad implementation, to be honest. I am not convinced they are addressing the right prolems
15:22:12<joepie91>problems*
15:22:18<f_>hah
15:22:39<joepie91>like
15:22:58<joepie91>for example, the lazy-loading of room members. okay, but why is that needed? why is loading room members so dog slow to begin with? it's a goddamned list of users
15:23:16<f_>Takes a second on IRC™
15:23:21<immibis>wouldn't you have to replay the state since room creation to know it's the correct set of members?
15:23:22<joepie91>and in the past I've found some hilarious performance issues in Synapse too
15:23:36<f_>joepie91: The E in Matrix means efficient.
15:23:38<joepie91>no, there are state checkpoints
15:23:42<immibis>the server could cache it but do you trust the server?
15:24:10<joepie91>I am not just talking about joins here
15:24:15<joepie91>I am talking about all client to server communication
15:24:23<joepie91>in the vast majority of cases, the server already knows who are in the room
15:24:30<joepie91>yet member lists are still slow
15:24:45<f_>matrix--
15:24:46<eggdrop>[karma] 'matrix' now has -39 karma!
15:24:54<f_>-39 karma o.O
15:24:58<joepie91>yes, efficient join is a tricky problem, but member lists, c'mon
15:25:46<Fijxu|m>IRC doesn't has chat history, most people don't use it because of that
15:26:01<Fijxu|m>I use a bouncer so I don't care
15:26:03<f_>Fijxu|m: That's partly incorrect these days
15:26:14<f_>A few IRC networks *do* have chat history now
15:26:26<immibis>sending the last few hours of history isn't the same as having unlimited history
15:26:41<f_>immibis: I'm talking ircv3 chathistory
15:27:17<yasomi>immibis: usually IRC networks don't have chat history so that opers don't have to deal with legal issues because of it
15:27:53<immibis>yasomi: yes, that's what i always thought
15:28:17<yasomi>speaking as a former IRC operator, that's the argument I made (yet we ended up relenting with a hosted znc)
15:28:48<immibis>with hosted znc, the possibly bad messages are stored in each user's private storage space where the oper should not touch them
15:29:32<immibis>there's a spectrum of how long data is stored; IRC does store messages, but only for milliseconds. Slack stores it for a year if you don't pay, I think. Discord and Matrix store it forever.
15:30:21<immibis>a user can force a server to store a message for up to several minutes by not reading from their socket, but there's no legal responsibility to delete things from socket buffers; the stuff is already in the process of transmission.
15:30:42<immibis>a different protocol might be able to extend this legal window up to hours at most
15:33:30<f_>immibis: Ergo does somewhat "unlimited" history, InspIRCd has a channel mode +H that tells how much history will get recorded
15:33:43<f_>Fijxu|m: which bouncer do you use?
15:33:54<f_>I use soju
15:34:54<Fijxu|m>znc
15:35:23<immibis>f_: and when someone posts a link to a drug market and it is still there 3 weeks later and the police ask the operator to remove it, how do they do that?
15:35:47<immibis>or instructions to make a bomb or something
15:35:59<immibis>(btw: you can do it by -------$%^&$I%^#$V%^##$SCDSNO CARRIER
15:36:06<f_>immibis: IRCv3 redaction .. I'm seriously suspecting that you didn't research what ircv3 is
15:36:21<f_>Fijxu|m: I find ZNC slightly janky to me
15:36:36<yasomi>I know ergo has redact commands
15:36:37<immibis>ircv3 was the json rewrite of irc
15:36:46<f_>immibis: there's no json even
15:37:03<f_>you should research
15:37:23<immibis>there used to be
15:37:30<f_>There isn't.
15:37:52<immibis>everyone got upset because they tried to replace the core protocol framing with json. https://github.com/ircv3/ircv3-specifications/issues/55
15:37:53MrMcNuggets (MrMcNuggets) joins
15:38:18<immibis>especially that last sentence that basically says "if you don't like this, fuck off"
15:41:11<yasomi>honestly the JSON based protocol would have made parsing SO MUCH EASIER
15:41:18<yasomi>like you have no idea
15:41:49<immibis>parsing json is not simpler than split(" ")
15:42:03<immibis>it's probably good for tags though
15:42:08<yasomi>different kinds of simplicity
15:42:25<f_>fireonlive++
15:42:25<eggdrop>[karma] 'fireonlive' now has 762 karma!
15:42:49<immibis>JSON is convenient because you can just pull in a library to parse almost any structure, but that same thing is a problem if you want to parse it efficiently
15:43:13<yasomi>JSON parsers go at gigabits per second though
15:43:13<immibis>like, you can be sent {"junk":[[[[[[[[[[[...]]]]]]]]]], "actual message"} and you just have to handle skipping over that
15:44:16<immibis>well i can send you a gigabit and you'll take a second to process it. the 512-byte limit is slightly limiting sometimes (especially in non-latin alphabets) but having a limit is sensible
15:45:50<immibis>IRC could do with command tags as in IMAP
16:03:41<f_>yasomi: May I ask, what IRC network were you operating?
16:04:05<yasomi>f_: i'd rather not say
16:04:15<f_>Fair, your choice.
16:18:09Exorcism quits [Quit: Ping timeout (120 seconds)]
16:18:15DigitalDragons quits [Quit: Ping timeout (120 seconds)]
16:20:00Exorcism (exorcism) joins
16:20:00DigitalDragons (DigitalDragons) joins
16:21:22DigitalDragons quits [Excess Flood]
16:24:21Froxcey_ quits [Remote host closed the connection]
16:25:55Froxcey (Froxcey) joins
16:30:40Froxcey quits [Ping timeout: 260 seconds]
16:30:40katocala quits [Ping timeout: 260 seconds]
16:30:47katocala joins
16:35:35Froxcey (Froxcey) joins
16:42:20katocala quits [Ping timeout: 260 seconds]
16:42:32katocala joins
16:42:55Froxcey quits [Ping timeout: 260 seconds]
16:43:52Froxcey (Froxcey) joins
16:43:58ducky quits [Ping timeout: 260 seconds]
16:47:44ducky (ducky) joins
17:16:20DigitalDragons (DigitalDragons) joins
17:16:56SootBector quits [Ping timeout: 276 seconds]
17:18:36SootBector (SootBector) joins
17:22:45Froxcey quits [Remote host closed the connection]
17:22:51Froxcey (Froxcey) joins
17:24:57DigitalDragons quits [Client Quit]
17:24:57Exorcism quits [Client Quit]
17:25:18DigitalDragons (DigitalDragons) joins
17:25:21Exorcism (exorcism) joins
17:30:15Froxcey quits [Remote host closed the connection]
17:30:42Froxcey (Froxcey) joins
17:30:45<immibis>was it pissnet
17:31:05Froxcey quits [Remote host closed the connection]
17:31:16Froxcey (Froxcey) joins
17:37:10etnguyen03 quits [Quit: Konversation terminated!]
17:46:37<f_>immibis: They said they'd rather not say.
17:53:22<f_>I suggest you don't insist
17:53:25etnguyen03 (etnguyen03) joins
17:55:41systwi_ quits [Quit: systwi_]
17:58:10nukke quits [Ping timeout: 260 seconds]
18:04:26imer quits [Quit: Oh no]
18:05:07imer (imer) joins
18:12:02nukke (nukke) joins
18:14:23ducky quits [Ping timeout: 260 seconds]
18:14:51ducky (ducky) joins
19:35:55MrMcNuggets quits [Quit: WeeChat 4.3.2]
20:15:19<@OrIdow6>JAA: Regex aside I do sometimes just use the Firefox json viewer for stuff like that
20:15:35<@OrIdow6>But ofc a commandline one has the advantage you don't need to interrupt your workflow getting it loaded
20:38:42mls (mls) joins
20:44:39<@JAA>joepie91: I'm not really familiar with the details, so: what happens if your Matrix homeserver suddenly dies? Both in the case when it's your own server that's irrecoverably damaged and when it's someone else's who suddenly vanishes. Is there any way to recover your identity in such a situation?
20:47:20matoro quits [Ping timeout: 260 seconds]
20:49:55<@JAA>OrIdow6: Yeah, for smaller online things, I often do that as well. Anything local or big, I've been piping through jq to less, which is less than ideal.
20:50:11<@OrIdow6>Yep, same
20:50:34matoro joins
20:51:28<@OrIdow6>Well, except what I mean is sometimes I'll save local stuff as .json adn open that in FF
20:53:00etnguyen03 quits [Client Quit]
20:53:37<@JAA>Hmm yeah, I guess that can be done. If I have it local though, I'm probably already in a terminal anyway.
20:59:47<joepie91>JAA: if you have your signing keys backed up, you can recover from a server failure, even room history will typically sync back from other servers just fine. if someone else's server goes down, you'll lose your account on that server, and any permissions associated with it, but rooms will not be affected (and you technically can even be invited back into DMs on another account, because those are also rooms)
21:00:23<joepie91>after recovery from a backup, your server will hurt for a while, but it will eventually catch up and recover
21:10:04jasons (jasons) joins
21:20:03<@OrIdow6>joepie91: How does that work on an end user level? If my homeserver goes down do I open an account on a new homeserver and give it the same keys?
21:20:37<masterx244|m>OrIdow6: the signing key thing is for those that selfhost their own homeserver
21:21:00<@OrIdow6>masterx244|m: Oh
21:21:12<masterx244|m>if you are on someone else's homeserver you need a reinvite into the rooms
21:21:17sralracer quits [Quit: Ooops, wrong browser tab.]
21:21:29<masterx244|m>(for me its luckily the first case, i selfhost my entrypoint)
21:22:26<@JAA>Hmm, I see, that sucks for most people then.
21:22:41<@JAA>Is there any kind of motion towards federated identities?
21:23:12<@JAA>I know it's been a topic in the Mastodon circles for years, though I'm not actively following it, so no idea what happened to it there.
21:24:05<joepie91>I am not aware of federated identities being a working thing yet, and I am not convinced that it is possible without undesirable tradeoffs
21:24:18<joepie91>zooko's triangle and all that
21:24:40<joepie91>like, the best I could see is a webfinger-esque approach like fedi does and even there that barely works really
21:24:58<joepie91>and it's ultimately not really relevant if you don't have your own domain, which the vast majority of people will not
21:25:56<@JAA>Yeah, it'd certainly need to be done very carefully.
21:28:16<joepie91>more generally, for anything that touches the P2P Matrix project, I'd be very skeptical. they probably have genuine intentions but I do not believe that they have done proper feasibility analysis and answers to obvious questions on feasibility never materialized
21:29:08<joepie91>so it's a useful project to be working on, it would be great if it works, but do not count on it (or its components) ever shipping, basically, certainly do not hold your breath for it
21:33:43<@JAA>I stopped holding my breath soon after it first emerged because I kept hearing how awful the only fully-featured homeserver implementation (Synapse) was yet nothing seemed to improve there. And I'm still hearing these complaints now that the protocol is a decade old. (No first-hand experience, just people cursing at it in various places.)
21:35:40<joepie91>oh yeah, synapse is busted
21:40:47<@OrIdow6>I have a Matrix but no one's on there so I don't use it (setting up the STWP bridge was I think the most activity I've ever had on there)
21:41:05<@OrIdow6>And am on somebody else's homeserver so waiting until I get screwed
22:05:35jacksonchen666 (jacksonchen666) joins
22:09:19etnguyen03 (etnguyen03) joins
22:28:15StarletCharlotte quits [Ping timeout: 260 seconds]
22:38:08<@JAA>The netcup RS deals will happen in a bit over 8 hours (07:00Z).
22:40:34<nicolas17>the what
22:41:40<imer>huh, is that an actual 2.5g flat? how's IA peering? :D
22:44:22<@JAA>nicolas17: netcup's Black Week deals for their RS line.
22:46:04<@JAA>imer: Not flatrate, there's throttling, but it's pretty fair considering the price. Can't say anything about IA peering as I haven't been uploading much directly.
22:46:25<imer>that would've been too good I suppose :)
22:46:50<@JAA>Yeah
22:47:43jacksonchen666 is now known as jacksonchen666_
22:47:51jacksonchen666_ is now known as jc666
22:50:35jacksonchen666 (jacksonchen666) joins
22:56:43jacksonchen666 quits [Remote host closed the connection]
22:56:46jacksonchen666 (jacksonchen666) joins
23:11:42jc666 quits [Client Quit]
23:18:40Froxcey_ joins
23:18:40Froxcey quits [Read error: Connection reset by peer]
23:48:18etnguyen03 quits [Client Quit]

HomeSearchPrevious dayNext day