Time |
Nick |
Message |
00:29 |
|
Artea joined #luanti |
01:36 |
|
shinbet joined #luanti |
01:49 |
|
wallabra joined #luanti |
02:45 |
|
sparky4 joined #luanti |
03:13 |
|
amfl2 joined #luanti |
04:00 |
|
MTDiscord joined #luanti |
04:16 |
|
ltuyen joined #luanti |
04:19 |
ltuyen |
luanti player, what's your opinion about the discord? |
04:19 |
ltuyen |
i don't use discord and i never had an discord account because it's a cesspool place trying to harm me. |
04:20 |
[MatrxMT] |
<Blockhead256> it unfortunately has the most features, including free uploads, but yeah, ultimately it's a platform that tries to dominate the chat and internet space and take up your time, is nonfree software |
04:21 |
[MatrxMT] |
<Blockhead256> I think it's a particularly bad choice for a free software project like Luanti, because it's not value aligned at all |
04:27 |
|
ltuyen left #luanti |
04:30 |
|
sparky4 joined #luanti |
04:49 |
|
repetitivestrain joined #luanti |
05:23 |
|
fanshen joined #luanti |
05:52 |
|
sinvet joined #luanti |
06:00 |
|
SwissalpS joined #luanti |
07:26 |
|
zleap joined #luanti |
07:26 |
|
zleap joined #luanti |
07:29 |
celeron55 |
the worst thing about discord is that it's an information black hole |
07:30 |
celeron55 |
not even that it's non-free, even though that's bad also |
07:30 |
celeron55 |
matrix really has the same information black hole issue. these things are not good places to publish information |
07:30 |
|
alpiquero joined #luanti |
07:31 |
[MatrxMT] |
<Blockhead256> there is supposed to be a way to view matrix from open.matrix.org, but the Luanti room stopped working |
07:31 |
celeron55 |
irc channels with public logs are better, and a forum is best in terms of not being an information black hole |
07:31 |
[MatrxMT] |
<Blockhead256> discord is closed off by design on the other hand |
07:31 |
[MatrxMT] |
<Blockhead256> the forum unfortunately is basically getting flooded all the time now though |
07:32 |
[MatrxMT] |
<Blockhead256> sorry view.matrix.org |
07:33 |
[MatrxMT] |
<Blockhead256> https://view.matrix.org/room/!minetest-general:tchncs.de/ |
07:33 |
celeron55 |
the main problem of irc is not really even that it's inaccessible to gen Z, it's that it's a completely different world. when you post to discord, you are writing to a US corporation database. when you post on IRC, you are sending a TCP packet. they operate at completely different levels with huge implications on everything |
07:36 |
MTDiscord |
<redundantcc> Makes a forum post: Uses HTTP over TCP Makes an IRC post: Uses Cloud IRC over TCP insert that one meme about it being the same picture here |
07:37 |
MTDiscord |
<redundantcc> Although I would tend to agree that Discord is kind of trash when it comes to searchability, but when you get down to it everything nowadays is just packets. |
07:38 |
[MatrxMT] |
<Blockhead256> the internet is a series of tubes |
07:38 |
MTDiscord |
<redundantcc> What do children use for communication nowadays anyway? |
07:38 |
MTDiscord |
<redundantcc> Back in my day we used to use soup cans and a bag of rope? |
07:41 |
|
bodiccea joined #luanti |
07:42 |
celeron55 |
IRC is always distributed in terms of storage, unlike these other platforms. if you send a PM, only you and the receiver will have a chance to store it and even that is a choice. and if you post here, you can use the [off] tag to skip the public logs. IRC shows you who will see your TCP packet as you send it. after that, it is the responsibility of the receivers to not store it or somehow safekeep it |
07:44 |
celeron55 |
discord and matrix store everything you send in a more or less centralized database. that means now choosing to store your message is, in addition to the receivers, your responsibility, and the database maintainer's responsibility |
07:45 |
|
est31 joined #luanti |
07:45 |
celeron55 |
and they also generally make personal and public logging more difficult, so now it actually isn't the receiver who's job is to safekeep your message. which is very backwards in my opinion |
07:45 |
celeron55 |
s/is to/to/ |
07:46 |
celeron55 |
discord and matrix are in this sense some kind of weird very closed off nanoblogging platforms. your messages are essentially blog posts |
07:46 |
MTDiscord |
<redundantcc> It depends on your perspective I suppose, I prefer to look at it as ephemeral rather than distributed. And that all online recipients received a packet, and if the intended target of the packet does not exist it is simply ignored. But I touch networks all day so I probably have worms in frontal lobe. |
07:46 |
celeron55 |
irc is more closely related to CB radio than these nanoblogs |
07:47 |
MTDiscord |
<redundantcc> I would tend to agree |
07:48 |
MTDiscord |
<redundantcc> But I suppose it's you're looking for in a chat platform, if you want the ephemeral experience of going to a farmers market and chatting with whoever's there then IRC works well. If you want the experience of going online to a standard social media platform, then you need logging and a mailing system and a database to store it all. It's just a difference in what you want out of the experience no? |
07:48 |
celeron55 |
this is very technical, but the technicalities are important, because those completely form the end result |
07:49 |
celeron55 |
yes. udoubtedly people find these database based social media platforms useful |
07:49 |
celeron55 |
undoubtedly* |
07:49 |
[MatrxMT] |
<Blockhead256> for my part, there's a persistent problem in turning ephemeral conversations into useful, search engine findable information so that the conversations don't need to be repeated |
07:50 |
MTDiscord |
<redundantcc> You're not wrong in the sense that CSS defines the layout of a page, but I think it's more important to look at it from a conceptual standpoint. Because it's the ux designer that actually designs the pictures used to generate the css, and that higher order thinking is what ultimately brings out the potential in some of the best software. |
07:50 |
MTDiscord |
<redundantcc> Sorry that might be softwares, I'm unfamiliar with the plural for that word. |
07:50 |
[MatrxMT] |
<Blockhead256> software is an uncountable noun in english |
07:50 |
celeron55 |
i think the main gripe with discord (and matrix) is that technically and privacy wise they are not ephemeral, but user experience is ephemeral. they collect data while not actually giving the proper advantage of that data collection back to users |
07:51 |
celeron55 |
forums collect data, but they give you a giant return on your posts in terms of publicity and archivability |
07:51 |
celeron55 |
forums are in this way very optimal |
07:52 |
celeron55 |
discord and matrix are essentially making a lot of waste in utilizing their collected data and to a privacy conscious person that's very iffy |
07:53 |
MTDiscord |
<redundantcc> May I float the idea of a properly distributed IRC, or rather a layer to the protocol on top of the standard IRC protocol. Similar to the way that torrent clients can store data and use peer-to-peer transactions, something like that but for IRC messages where parts of the logs are logged everywhere and can be requested. Obviously limitations including storage space would limit the time that was logs were stored, but a week of data |
07:53 |
MTDiscord |
wouldn't take much more than a megabyte if it was just text. |
07:53 |
[MatrxMT] |
<Blockhead256> I think it actually speaks, at least in some way, to the forums usefulness as a source that they are getting hammered with traffic. At also just speaks to the scale of data scraping in this decade... |
07:53 |
MTDiscord |
<redundantcc> Why not provide a proper API for scraping it, then people will be disincentivized to use bots and you can rate limit them individually? |
07:53 |
celeron55 |
yeah AIs are obviously crazy for forum data. i just wish they'd compensate me for it |
07:54 |
celeron55 |
forums are a massive benefit to society and anyone hosting a forum should be compensated a lot by countries and companies |
07:54 |
[MatrxMT] |
<Blockhead256> having a proper API means nothing, bots operate as they do not because they are lacking an API but because they are strip mining--trawling the internet |
07:55 |
celeron55 |
they're building the information records of today as much as books are, if not more |
07:55 |
celeron55 |
forums, i mean |
07:55 |
MTDiscord |
<redundantcc> If I can be nosy how much do they cost to run, is it particularly expensive? |
07:55 |
celeron55 |
it doesn't cost much money, that is not the issue |
07:55 |
celeron55 |
the issue is time, which i have almost none of these days. i mean time where my brain has any energy left |
07:56 |
celeron55 |
i can't buy that kind of time |
07:56 |
celeron55 |
or if you have that for sale, please give me an offer |
07:57 |
MTDiscord |
<redundantcc> Not to be rude but you programmed one of the coolest piece of software I've used, what part of the administration process is so Hands-On but you couldn't just use a script even after all these years? |
07:57 |
celeron55 |
use a script to... do what? |
07:57 |
MTDiscord |
<redundantcc> Well I just wonder, because you said it cost a lot of time and effort to make decisions about what to do and not to do on the forms? |
07:57 |
celeron55 |
i wrote minetest when i was a student and didn't actually study almost at all. that was easy |
07:57 |
MTDiscord |
<redundantcc> I was wondering if there was a way to automate some of that? |
07:57 |
[MatrxMT] |
<Blockhead256> I have to ask, does it take care of rebooting itself when it falls over? or is it just the proxy that falls over? |
07:58 |
celeron55 |
the only pressing issue with the forum is that the machine it's running on is equivalent to an old version of raspberry pi in terms of power |
07:59 |
MTDiscord |
<redundantcc> Wait you run the entire thing on a silicon wafer cookie running on a potato battery, no wonder it's kind of rough sometimes. |
07:59 |
celeron55 |
it's a dedicated machine at a french datacenter. really the disk could be cloned and spun up in any other physical machine |
07:59 |
[MatrxMT] |
<Blockhead256> improving the hardware would make it more responsive, inducing more demand for forum traffic and slowing it again lol |
08:00 |
celeron55 |
or, well... i suppose it couldn't because of cpu architecture and all that |
08:00 |
MTDiscord |
<redundantcc> Can't have anything nice on the internet, someone will try to bring it down... if even by accident. |
08:02 |
celeron55 |
i like to think about it positively: it's amazing that it's running at all |
08:03 |
MTDiscord |
<redundantcc> Is there anything you don't like about it architecture wise, like if you had a blank check to work on it for a year... is there anything you would want to redo or change about the way that the forums are architectured? |
08:04 |
celeron55 |
https://gcdnb.pbrd.co/images/AaxUcnFZK3VH.png?o=1 |
08:04 |
celeron55 |
https://gcdnb.pbrd.co/images/eOnq7bXIMuCm.png?o=1 |
08:04 |
MTDiscord |
<redundantcc> Like in this Fantasyland a government really did come up to you and said " here live in this house, eat this food. go make good software", would you really have anything to change? |
08:04 |
celeron55 |
that's what the graphs show for the past weeks |
08:05 |
celeron55 |
essentially, the crawlers are going crazier than ever |
08:05 |
MTDiscord |
<redundantcc> Have you tried fingerprinting and blocking some of them, not all at once just occasionally to reduce traffic over the proxy? |
08:05 |
celeron55 |
also you can see the database can serve about 100 queries per second |
08:06 |
MTDiscord |
<redundantcc> That's honestly not a lot for a major project. |
08:06 |
MTDiscord |
<redundantcc> But off peak it should be more than enough |
08:07 |
celeron55 |
i'm a founder of a startup since 3 years ago. if the government came and said that, i couldn't change anything because i've signed a lot of papers to be stuck in this for years still. it's probably worth it and i can be my own government after this, but that'll take years still |
08:07 |
MTDiscord |
<redundantcc> What doesn't make sense is those spikes in traffic are consistent with the usage hours of a human sleep/work cycle. So if it is bot traffic it would still be fundamentally humans using those Bots create queries? |
08:08 |
[MatrxMT] |
<Blockhead256> If I were trying to run a bot to look human-ish, I would specifically make it follow expected traffic patterns |
08:08 |
MTDiscord |
<redundantcc> And what's that weird zigzag on July 5th? |
08:08 |
celeron55 |
fingerprinting and blocking crawlers is exactly the kind of web admin work i absolutely hate |
08:08 |
celeron55 |
if you want SSH access just tell me and i'll let you do it |
08:09 |
MTDiscord |
<redundantcc> I would but I'm kind of busy preparing for the contest you guys host at the end of the year, but I could definitely ask around and see what would work best in this situation. I'll let you know. |
08:09 |
celeron55 |
also if someone wants to try to clone the machine, just tell me |
08:10 |
celeron55 |
but |
08:10 |
celeron55 |
the thing is, it's just a phpbb forum. it's possible to set it up anywhere by just copying the files and the database |
08:10 |
MTDiscord |
<redundantcc> Also blockhead256, there's no perfect way to emulate human behavior in an absolute fashion. You could Implement all the transforms and bell curves and transitions you want but humans are messy and meaty and they bleed over all the data in unexpected ways and no amount of pseudo random number generator it's going to be able to perfectly encapsulate all of that. |
08:10 |
MTDiscord |
<redundantcc> Fingerprinting is always possible, except when it's Google cuz they use lava lamps and using physics is cheating! |
08:11 |
[MatrxMT] |
<Blockhead256> All I am saying is that I expect bot makers are at least trying to not make their activity completely obvious |
08:12 |
MTDiscord |
<redundantcc> I mean if they're publicly available and I can set up my own instance I could have the boys hit with the standard "we'll see what happens package", at worst they'll use it to take notes which would still probably incentivize some optimizations? |
08:12 |
celeron55 |
>I expect bot makers are at least trying to not make their activity completely obvious |
08:12 |
celeron55 |
??? |
08:12 |
MTDiscord |
<redundantcc> Where can it be downloaded, is the code publicly available? |
08:12 |
celeron55 |
crawlers for AI systems are the stupidest things i have ever seen on the intenret |
08:12 |
celeron55 |
internet |
08:12 |
[MatrxMT] |
<Blockhead256> forget I said anything |
08:12 |
celeron55 |
they essentially launch a DDoS attack with no rate limits |
08:12 |
MTDiscord |
<redundantcc> > Also Celeron, you have to add a space otherwise it looks weird on my end |
08:12 |
celeron55 |
and don't cancel if the website is giving out solely errors |
08:13 |
MTDiscord |
<redundantcc> But that seems like unknown behavior that can be corrected with a proxy, like if they don't rate limit themselves just do it for them? |
08:13 |
celeron55 |
i'm not working for them |
08:14 |
MTDiscord |
<redundantcc> That seems very solvable by tying their IP address directly to their transactions measuring and then just rate limiting |
08:14 |
MTDiscord |
<redundantcc> Here hold on let me look something up |
08:14 |
[MatrxMT] |
<Blockhead256> if they're well resourced, they'll switch IPs anyway |
08:14 |
celeron55 |
i have an ip block based rate limit on my nginx reverse proxy, and that is what is keeping the forum accessible most of the time |
08:15 |
celeron55 |
i mean, i don't rate limit ip addresses, i limit /24 addresses |
08:15 |
celeron55 |
that's what made it work reasonably |
08:15 |
celeron55 |
originally it was ip address based and that didn't help at all with the DDoS crawlers |
08:16 |
celeron55 |
but the DDoS is getting worse every month |
08:16 |
celeron55 |
they have so many resources at hand it's stupid |
08:17 |
[MatrxMT] |
<Blockhead256> spam is coming fairly frequently too, and it's insidious, because they use an LLM to write something that looks innocuous, but then edit it later with spam |
08:17 |
[MatrxMT] |
<Blockhead256> you can always spot a fake from their profile |
08:18 |
[MatrxMT] |
<Blockhead256> there was also a post that looked fairly legit, but the actual post content was an actual cognitohazard of AI fuzziness that hurt to try to parse, so that got banned too |
08:19 |
celeron55 |
i wish we had a better community platform optimized for luanti |
08:19 |
celeron55 |
and optimized to keep the AI crap away |
08:19 |
MTDiscord |
<redundantcc> Just going to toss this here, I have to look into it: https://github.com/owasp-modsecurity/ModSecurity |
08:20 |
celeron55 |
phpbb has been painfully outdated for a decade in terms of being able to exist at all on the internet |
08:21 |
[MatrxMT] |
<Blockhead256> having a phpBB sucks, but needing to find anything else and migrate and try not to lose data also sucks |
08:21 |
MTDiscord |
<redundantcc> What about integrating IRC directly into the engine, it would incentivize a new wave of people communicating through and while it is possible to connect as a bot non-activity it's fairly easy to detect? |
08:21 |
celeron55 |
i read the first chapter on ModSecurity and my head already hurts |
08:22 |
MTDiscord |
<redundantcc> I've heard of it before, nothing bad |
08:22 |
celeron55 |
i should make a forum which literally uses luanti as the client and server |
08:22 |
celeron55 |
or someone |
08:22 |
MTDiscord |
<redundantcc> But as it is a Tuesday, I wouldn't have the time to fully set up a lab for testing. Maybe Saturday I could pop an SSH socket mirror your files and take a look at what I could do? |
08:22 |
celeron55 |
then it could generate static pages to be published to the AI crawlers to the open internet |
08:23 |
celeron55 |
there's no way AI bot and spammers are willing to make a custom client for a custom protocol until like 2030 |
08:23 |
celeron55 |
or 2040 |
08:24 |
MTDiscord |
<redundantcc> Why not store data about public posts as a kind of Json object, and then just load it via JavaScript. I know it's not great in terms of pop in, but it would allow you to basically serve your entire website as a static page from something like GitHub pages... which while owned by Microsoft is essentially "free" |
08:24 |
celeron55 |
i mean, some day AI will ruin everything that's out in the open and you'll need to authenticate using a official digital passport of some kind to prove you're human, but that'll take some time still |
08:24 |
celeron55 |
an* |
08:25 |
celeron55 |
eh |
08:25 |
celeron55 |
well you do that and tell me when i'll point forum.luanti.org to it |
08:25 |
MTDiscord |
<redundantcc> I don't think so, if the internet stops being a Haven for cool things protocols like gopher and encryption methods like tor will pick up the slack. There is no such thing as killing pirates, where you spilled blood others will smell it. |
08:26 |
celeron55 |
ah tor. maybe i should move the forum to there |
08:26 |
MTDiscord |
<redundantcc> Sure yeah let me just code that up real quick, shouldn't take more than a couple minutes |
08:26 |
MTDiscord |
<redundantcc> ;-; |
08:26 |
MTDiscord |
<redundantcc> meanwhile I've been studying how to make a mod for 3 years and failing |
08:27 |
[MatrxMT] |
<Blockhead256> reminder that Luanti is still an unencrypted protocol. I don't think running chats on it is good |
08:28 |
MTDiscord |
<redundantcc> By that standard so is IRC though, and also everything is sent through clear text unless you use a plug-in like pgp? |
08:28 |
celeron55 |
luanti can do https requests |
08:28 |
celeron55 |
but yeah, that's not the luanti protocol |
08:28 |
MTDiscord |
<redundantcc> Doesn't it also need a special permission though? |
08:28 |
[MatrxMT] |
<Blockhead256> the client component remains unencrypted without CSM |
08:29 |
celeron55 |
usually these days people connect to IRC using TLS |
08:29 |
celeron55 |
but yeah, it's not enforced |
08:29 |
MTDiscord |
<redundantcc> Wait so theoretically you could just connect to free node for what have you directly through the client? Why doesn't that exist somewhere, that's wacky and cool! |
08:30 |
celeron55 |
what? |
08:30 |
MTDiscord |
<redundantcc> If you can connect over HTTP TLS, wouldn't that mean it's compatible with just sending and receiving data directly through the client? |
08:31 |
MTDiscord |
<redundantcc> Or do you still need some kind of custom socket to connect to an IRC? |
08:31 |
MTDiscord |
<redundantcc> I think I've fundamentally misunderstood something you said... |
08:32 |
celeron55 |
probably |
08:33 |
MTDiscord |
<redundantcc> Sorry, mi lengua Indiana no es inglés. In fact sometimes I can barely be said to speak it. |
08:34 |
MTDiscord |
<redundantcc> I was saying that if you could create custom sockets you could theoretically connect to an IRC server from within the client, I think that would be cool as just an experience you can give players who might need help when they first open the game? |
08:38 |
MTDiscord |
<redundantcc> Alternatively if the forms were organized to just be json files wrapped in some shiny CSS/ML you could show it directly without much more been an HTTP request to requisite the necessary files directly from the forms page? |
08:38 |
MTDiscord |
<redundantcc> Which could be cool, but probably not as cool as introducing people to IRC... and it wouldn't really solve the conflict of the scrapers. |
09:10 |
|
qqe joined #luanti |
09:17 |
|
mrkubax10 joined #luanti |
09:36 |
|
qqe joined #luanti |
09:42 |
|
qqe joined #luanti |
10:00 |
|
shinbet joined #luanti |
11:08 |
|
fanshen joined #luanti |
11:54 |
bgstack15 |
On the topic of the forum, I have noticed it is really slow. What can I do to help the forum load faster? Do you need some donations to get beefier hardware? |
11:55 |
bgstack15 |
I have a vested interest in making sure the forums continue, and the rest of the Luanti infrastructure. |
11:58 |
gry |
can you open web browser console network tab, load a forum page and take a screenshot of its output |
11:58 |
bgstack15 |
OK |
11:58 |
gry |
and link to that forum page, so i can try that too :) |
11:59 |
bgstack15 |
well, it's all forum pages, generally. Like the hardware is undersized (or under attack) |
12:01 |
sfan5 |
forums have a known load issue |
12:03 |
repetitivestrain |
Just bumped into https://github.com/luanti-org/luanti/issues/15419 |
12:04 |
repetitivestrain |
unfortunately i crashed my debugging session in attempting to extract the ModError within the ~Server destructor |
12:04 |
repetitivestrain |
but i should be able to reproduce the deadlock if anyone is interested |
12:04 |
bgstack15 |
Hm, my initial observation while preparing these screenshots is that my browser didn't seem to cache the repetitive stuff like stylesheets and scripts and images. |
12:05 |
bgstack15 |
https://bgstack15.ddns.net/screenshots/2025-07-08-080005_1920x1062_scrot.png |
12:05 |
bgstack15 |
https://bgstack15.ddns.net/screenshots/2025-07-08-080052_1920x1062_scrot.png |
12:07 |
bgstack15 |
https://forum.luanti.org/viewtopic.php?p=445494#p445494 |
12:07 |
bgstack15 |
after page https://forum.luanti.org/viewforum.php?f=18 |
12:41 |
|
wallabra joined #luanti |
12:44 |
|
yezgromafic joined #luanti |
12:57 |
|
silverwolf73828 joined #luanti |
13:05 |
|
cheek_pain joined #luanti |
14:07 |
|
SFENCE joined #luanti |
14:31 |
|
qqe joined #luanti |
15:03 |
|
mrkubax10 joined #luanti |
15:13 |
|
jstein joined #luanti |
15:18 |
|
jaca122 joined #luanti |
15:25 |
|
SFENCE joined #luanti |
16:17 |
|
sinvet joined #luanti |
16:44 |
|
SFENCE joined #luanti |
17:26 |
|
Talkless joined #luanti |
17:28 |
|
ineva joined #luanti |
17:54 |
|
SFENCE joined #luanti |
18:40 |
|
SFENCE joined #luanti |
18:43 |
|
gregon joined #luanti |
18:43 |
|
gregon left #luanti |
18:48 |
|
SFENCE joined #luanti |
18:56 |
|
SFENCE joined #luanti |
19:04 |
|
SFENCE joined #luanti |
19:22 |
|
SFENCE joined #luanti |
20:25 |
|
wallabra joined #luanti |
20:39 |
|
shinbet joined #luanti |
20:43 |
|
sinvet joined #luanti |
20:43 |
|
sinvet joined #luanti |
20:45 |
|
sinvet joined #luanti |
21:06 |
|
shinbet joined #luanti |
21:06 |
[MatrxMT] |
<Bracket> how do scrapers even know what to scrap? do they just try every possible address or do they follow links? If they follow links could you make the entry page have no links but only text telling you what address? |
21:07 |
|
shinbeth joined #luanti |
21:15 |
|
shinbet joined #luanti |
21:29 |
MTDiscord |
<luatic> scrapers follow links. but what you suggest is a pretty bad idea. it would make browsing very awkward, and some regular expressions that look for "addresses" (things that look like URLs) would still catch the links. it also doesn't help at all against external links, which there will be plenty of. |
22:03 |
|
Kimapr_ joined #luanti |
22:29 |
|
FileX joined #luanti |
22:33 |
|
panwolfram joined #luanti |
23:05 |
|
Eragon joined #luanti |