[01:13:46] Miku.. [03:03:07] @blankeclair i looked up how enby terms work in Portuguese as a gendered language [03:03:15] https://nonbinary.wiki/wiki/Gender_neutral_language_in_Portuguese this article was a fucking experience [03:32:33] languages are haaaaard [03:35:01] you should have seen my reaction to some of these things [03:35:16] oh gawd [03:35:19] That mostly aligns with how people try to get it to work with Spanish works too [03:35:25] with the -e suffix [03:35:46] the spanish article is longer [03:35:58] and @ [03:36:11] how do you pronounce an @ [03:36:28] also, latinx feels like such a hack to me [03:37:03] you think thats a hack? [03:37:04] https://cdn.discordapp.com/attachments/615786602454581249/1403220443279921283/image.png?ex=6896c260&is=689570e0&hm=b5af09a518d1bb3a42819151d2984a4a9a4a2df340a6450d5dd866a42e8af863& [03:37:21] i was so confused at that table that i just glossed over it [03:37:39] the more i focus on it, the more confused i get [03:37:52] https://cdn.discordapp.com/attachments/615786602454581249/1403220645612884108/image.png?ex=6896c290&is=68957110&hm=c811aca8fcb175221583056631f82ca5e5fd8a0c0f8a4e167c6cecbba8caa478& [03:37:59] that's why most people just prefer to use the standard that already exists in the language [03:38:06] amiguy :3 [03:40:16] all of this feels like a lingustical hack patch worthy of the mediawiki ecosystem [03:40:23] lmao tyes [03:42:25] shout out to vcaa for having Gender X btw [03:42:33] enby vs. the cooler enby [04:00:23] checking out the new arbcom cases [04:00:44] > Wikipedia is a place where a group of adult humans believes they need a central arbitration committee to find a peaceful answer to whether the title of some articles should be capitalized or not. [05:09:12] Hi [05:09:28] Especially Japan [08:49:49] I always use a/o [08:49:59] Either es or pt [08:59:03] [1/2] Those of you, Interested in GPT.... [08:59:03] [2/2] https://youtu.be/hmtuvNfytjM [08:59:35] [1/2] Quite long, but very interesting. [08:59:36] [2/2] Huge! If true* [10:02:42] https://losermakes.straw.page/ [10:09:06] [1/2] Someone told me to block UAs with the rule `(http.user_agent contains "Windows" and any(http.request.headers["sec-ch-ua-platform"][*] contains "Linux"))` and it turned out to be highly effective: in less than 12 hours more than 160k requests were blocked. I suspect that these are AI crawlers from alibaba looking for qwen's training data. [10:09:06] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1403319099668303872/image.png?ex=68971e41&is=6895ccc1&hm=1a6dae1ee7b9cd28c748e5646c68135120125ce73fdf85dbaa4cf7b5fa0e9f75& [10:09:49] I would assume more than 50% of traffic to my wiki consist of bots, so this probably blocked off a fraction of them. [10:11:36] [1/2] Ah yeah 1.3 million requests from Hong Kong alone. Definitely not suspicious at all. [10:11:36] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1403319728738533458/image.png?ex=68971ed7&is=6895cd57&hm=36e824a25139d2430384c1442001a327c9eb3cb65500ed69c7ad2876411edbf2& [10:12:08] I've also noticed huge bot traffic from HK [10:12:13] what are they cooking over there [10:15:09] [1/2] I heard bad things about alibaba's bots scraping training data for LLMs (at least the IP address comes from Alibaba, though it could be some other company abusing Alibaba's cloud servers). These bots typically use hundreds of IP addresses, which make them hard to block effectively. [10:15:10] [2/2] https://arstechnica.com/ai/2025/03/devs-say-ai-crawlers-dominate-traffic-forcing-blocks-on-entire-countries/ [10:16:04] That was a while ago. These crawlers definitely seem less sophisticated because no human would set Windows in their UA but have Linux as their `sec-ch-ua-platform`. [11:44:23] i wonder if it could be an actual browser, but with its UA spoofed [11:44:29] would be interesting to see if it loads js [13:12:05] @cosmicalpha I'm gonna leave that draft PortableInfobox PR open, but I've found a way to fix this for real this time and it should be 100% more stable and doesn't rely on the old parser whatsoever, or hacking around parsoid. [13:12:58] (this took me several hours of stepping through every single function call from the legacy parser in PortableInfobox and seeing what data is exepcted, and seeing if we can get the same data from Parsoid, and it turns out we can) [13:19:08] so portableinfobox is saved? 👀 [13:25:40] I think so [13:26:18] Idk if they have updated parsoid since this was first looked at and now its possible, or if it was always possible and nobody thought to try what I’m trying [13:26:44] (i bet the former, as im sure this has been looked at by s few people who are much more experienced with Parsoid) [13:30:40] O h ? [13:32:59] extremely painful because a) the vscode interface for debugging is fucking horrific [13:33:15] b) the request kept timing out so it would close the debug session and I'd have to start all over again from the beginning cba [13:33:19] but alas [13:33:39] i debugged with var_dump() [13:33:53] What was the solution [13:36:21] [1/2] I don't have the full solution yet, but what I do have is that I'm at a point where I know the data is there, and i know which functions PortableInfobox calls and which data is passes, and I've been able to match them up 1-1 so it's just a case of me doing that. [13:36:22] [2/2] That being said, I'll probably have to rewrite a a large portion of some stuff to match what Parsoid expects etc. which is not a problem I'll just copy it and shove it into a separearte [13:36:54] (which isn't really an issue although the code will be duplicated to some degree, eventually the legacy parser stuff will go away( [13:38:01] [1/2] > i debugged with var_dump() [13:38:02] [2/2] I hate that [13:38:14] its probably not a lot worse than the vscode interface though [13:54:59] I also think it would be nice if there was a way to make PortableInfobox return a table based infobox, something super super simple and basic, when the raw HTML of a page is requested [13:55:23] because currently it does what it used to do when you used parsoid when you request the raw HTML [13:56:01] idk what you mean? Why would it return a html table? If its going through the legacy parser it should return the html of the infobox? [13:56:08] Do you have an example? [13:56:33] yeah hang on a sec [13:57:41] oh god I’ve forgotten the parameter now [13:58:02] ?action=raw maybe? [13:58:18] I think that’s it lemme test [13:58:33] I always wondered why they didnt just rename that view=raw but [13:58:58] no that returns wikitext [13:59:14] this returns like parsed HTML [13:59:17] oh [14:00:03] is it an API parameter? [14:00:09] possibly [14:00:44] [1/2] I don't think this is what you mean but is it: [14:00:44] [2/2] `api.php?action=visualeditor&format=json&paction=parse&page=$TITLE&uselang=en-gb&editintro=&preload=&preloadparams=&formatversion=2` [14:04:16] no [14:04:25] hang on lemme check what it was, I have it somewhere [14:09:48] ok got it [14:09:57] so if you call https://chinafake.wiki/wiki/Funny_Crawl_Spider?action=render for instance [14:10:08] that causes issues with PI [14:10:21] and [14:10:28] a lot of custom parser stuff actually [15:52:02] i think the issue here might be that action=render seems to doesn't load any css [17:01:06] chat should I buy an RPi5 with the NVMe hat and use it unironically as a daily driver? I like my computer but it is kind of a powerhungry beast and I don't use it's full capabilities most of the time, not even close, and I'm thinking most of the time I could get by on a Pi and save energy? [17:01:42] I doubt I'll do it until upstream Linux support improves so I can use it with a clean Debian instead of RPi OS but I've been seriously thinking about it [17:03:05] thought you said a MP5 then [17:05:53] or use Gentoo so I keep the RPi-specific crap under control https://wiki.gentoo.org/wiki/How_to_install_Gentoo_on_Raspberry_Pi_5 /s but also kinda not [18:11:57] I remember there was a patch that would have made it possible I dont know if was finally merged but seems possible then yeah. [18:22:06] None passed the Cloudflare challenge, so I’m assuming they’re scrapers. [19:00:32] if you think it will work, hey, why not [19:00:58] but you seem to have assessed clear mindedly and I personally don't think going over to arm for desktop computing is a good bet at this time [20:18:08] rpi with a hdd is slowwwww [20:18:15] even slower than flash storage lmfao [20:18:25] hopefully the nvme hat improves thingz [20:18:47] could be basic scrapers [21:55:12] it's supposed to, the benchmarks I've seen sure look good at least [21:57:03] fortunately I'm in the extremely niche group of users that only uses open source user-space software so I don't lose software access since most Linux distros cross-compile everything, but yeah, the moment you need to rely on anything for which you only have an x86 binary you'll curse the day you bought an arm device [21:57:36] for example windows has x86 emulators for their arm version but they kind of make you wish you just had an x86 device to begin with [21:57:52] from what I've seen, I haven't tested windows for arm [21:59:14] ooooo, you know you want to run windows on a raspberry pi [21:59:17] and as I've said, even if I have to give up most raw processing power what I'm really after is that sweet energy efficiency [21:59:26] sorry, not into masochism [21:59:31] damn [22:00:16] it would really open a window to new perspectives though... [22:00:28] 🙄 [22:01:29] weed [22:07:47] just found out there are people seriously running win11 on pis https://forums.raspberrypi.com/viewtopic.php?t=385149 [22:07:57] today you learned [22:08:40] we shall replace our servers with pis running hacked windows server [22:08:45] one day [22:12:44] @blankeclair do you know if there was a ticket for that hCaptcha issue with mobile frontend? [22:13:00] didn't know it affected MF tbh [22:13:17] i knew about DT though [22:13:25] right, DT [22:13:57] I didn't know where the issue was exactly, I just said MF without thinking [22:14:16] I just noticed another person got caught up with that on Meta https://meta.miraheze.org/wiki/Community_portal?curid=6947&diff=484617&oldid=484576 [22:14:51] I was wondering if you knew what is preventing this from being fixed [22:15:37] nvm just found it https://phabricator.wikimedia.org/T356599 [22:44:27] I'm typically much too lazy to glue binaries together myself [22:56:06] From your previous comments on Miraheze tech, I would not expect any other stance on FOSS tbh.