/tech/ - Tech


Mode: Thread

Max message length: 8192


Max file size: 80.00 MB

Max files: 5


(used to delete files and postings)


Remember to follow the rules

(50.94 KB 1024x1024 pythonlogo.png)
Python general Comrade 03/12/2020 (Thu) 10:41:36 No. 454 [Reply]
Post your python projects, ask help, anything related to python is welcome
13 posts and 2 images omitted.
>>701 this is exactly what i mean by /tech/ and /g/ being filled with meme hobbyists who don't know what the fuck they are talking about. >In python you have to write a bunch of trivial, stupid tests that are rendered pointless by a good type system. do you really think type errors are just abounding in apps? if they are in your app, you're just a bad programmer. I've literally never had a type error be the source of my problems after the first two years of a CS degree. How many actual, experienced industry devs are actually having type errors as their main source of concern? This is a problem for freshmen CS students, not professional devs who have actually had syntax and types drilled into their heads. Not saying it can't be a problem but in my experience 99% of problems come down to logic errors, performance, etc. Not simple syntax or type errors >the only moderately fast dynamic-typed language is common lisp Common lisp is indeed fast, pypy is 4 times faster than regular python, and Clojure is in the same ballpark as vanilla java since it's compiled to bytecode and runs on the JVM. >There are languages that compile extremely quickly. This is not an advantage at all. You have never seen an actual java enterprise codebase with a million+ lines of code. Go recompile that and say it's trivial I dare you. And before you say muh functional programming, the FP languages compilers are even slower (ex: scala) and the only reason you think they aren't is because you are writing toy programs with them and not enterprise level programs where you have dozens of programmers collaborating. >Some compiled languages have this, and it's pointless anyway if you are writing unit tests. ignorant. Units tests have literally fuck all to do with the REPL. go do some google research on this, before opening your mouth >Rails is a framework, which Rust doesn't have yet. Compare how the teams would do with Rocket/Actix vs Django. Or fuck even Servant or Yesod. NO ONE IS SAYING RUST OR HASKELL ARE BAD But if you think something like Rust is even comparable to python you are just high. Rust is on the same level of abstraction/productivity as C++, rust ownership is comparable to C++ smart pointers. So yes, I'd be willing to be the python guys would create an workable app at 2x the speed of the rust guys, regardless of framework >It turns out that good type systems actually make possible advanced macros that reduce workload Lisp has macros and it's a dynamic language.
>>710 >do you really think type errors are just abounding in apps? if they are in your app, you're just a bad programmer. so basically, you have zero assurance that type errors, or worse, code that functions when being sent the incorrect data type, don't exist in your code. lmao >How many actual, experienced industry devs are actually having type errors as their main source of concern? This is a problem for freshmen CS students, not professional devs Plenty of veteran devs complain about it. >logic errors, performance Types are a form of logic, no types means you can't express that logic in a simple way. And we've discussed perf already. >You have never seen an actual java enterprise codebase with a million+ lines of code Go and C compile fast. Jai if it ever becomes relevant. It's a fact that there are languages that compile fast, that many of them don't is just another problem with the industry, same as your awful enterprise Java. >NO ONE IS SAYING RUST OR HASKELL ARE BAD >But if you think something like Rust is even comparable to python you are just high. Have you ever used Rust macros? The Django-esque framworks for Rust and Haskell are absolutely beautiful and cut out a lot of dumb boilerplate using macros. You can define the API using types and automatic JSON serialization/deserialization. Any data sent to the endpoints that doesn't match the type is rejected. You get exactly what you expect. People comparing C++ to Rust are just scared by a bit of syntax. It is nowhere near as much a clusterfuck as C++. In practice Rust code can be just as abstracted and simple as Python, in fact moreso in many cases.

Message too long. Click here to view full text.

>>716 >You can define the API using types and automatic JSON serialization/deserialization. >Any data sent to the endpoints that doesn't match the type is rejected. You get exactly what you expect. you don't need rust for that, things like clojure already have type/shape validation using things like schema/spec in the REST api and automatic json serialization/deserialization to regular data structures using middleware. >Go and C compile fast. Never used go, but I can assure you C doesn't compile fast at scale >Jai if it ever becomes relevant. no idea what that is >Plenty of veteran devs complain about it. They're bad programmers tbh >It is nowhere near as much a clusterfuck as C++. and that's good, I sincerely hope it replaces C/C++ as the low level language of choice. I hope we have Rust game engines and shit in the future >Its macros don't have a type system to work with. S-expression macros are probably the most flexible of any, you can't out-macro a lisp
>>723 >Never used go, but I can assure you C doesn't compile fast at scale I can compile the Linux kernel on my computer in a few minutes. If your C project has worse compile times than the Linux kernel, everyone responsible for it should be shot tbh. >you don't need rust for that, things like clojure already have type/shape validation using things like schema/spec in the REST api and automatic json serialization/deserialization to regular data structures using middleware. the point is Rust has those really nice abstractions, and they are much more idiomatic to the language than adding type specs to random parts of a non-typed language.
>>727 C++ and Rust are pretty slow for compiling code. It can be pretty annoying in C++ but you know what annoys me more? When a random error stops a running python program. I can't stand the white space rules or it's lack of types. The only thing python has is really awesome libraries.

Brain to machine interface Comrade 03/27/2020 (Fri) 19:53:58 No. 649 [Reply]
What do you think about BMIs and relatively recent advancements by companies such as Neuralink? For those of you who don't know what BMIs are, it's a technology that implants electrodes from a chip inside your brain to read neurological activity and possibly send electric signals to your brain in an attempt to interface with computers. Here is Neuralink's white paper in case you are curious https://www.biorxiv.org/content/10.1101/703801v4.full.pdf And a popsci video of it, if that's more your thing https://www.youtube.com/watch?v=0jOjh6lwp9w&t
4 posts and 3 images omitted.
>>692 Google glass, in the form it was sold years ago, just wasn't a useful device. Even if it mildly increased your efficiency the social stigma of the nerdiest, pervert looking glasses wouldn't be worth it. I think we can all agree no one will be drilling holes in their heads. However the brain works on a series of electrical impulses, I don't see how theoretically we can't handle it non invasively. In regards to 'downloading' a map or something, I don't know enough to say whether it would be possible, there may be work arounds to the issues you discussed. Neurologists let me know Indirectly I think implants will become common place, they have managed to get every one to take a tracking device with them everywhere they go, there's no reason why people wouldn't put a little contactless debit card in their wrist.
>>649 >Brain to machine interface We already have that. It's the mouse snd keyboard in front of you.
(32.53 KB 302x325 fmri.jpg)
(10.69 KB 213x320 mody-mod.jpeg)
(9.54 KB 230x265 borg.jpeg)
>>708 I think that brain interfaces for computers will be possible with a solid state miniatured and simplified and directional MRI machine that will look inside your brain and read out the patterns of certain neurons in your prefrontal cortex to make an input proposal that you can then acknowledge with a single finger input. There's been recent advances in the miniaturisation of fmri machines from the size of a small room to the this(pic) creepy head-sized apparatus, if they can have another magnitude of size reduction it could probably fit inside a head-band you could wear as a input-device. I don't think you will get people to accept many implants unless they have medical benefits. There simply isn't enough benefit in implanting a credit card chip into your hand compared to having it as a plastic rectangle. I don't think there will ever be a normalisation of inserting stuff into your body, we have been using needles for medical stuff for centuries and people still try to avoid it as much as possible, I don't think you can get past evolutionary ingrained revulsion of having your skin pierced or having a foreign object stuck in your body. With some exceptions like the body-modification crowd. As far as inserting chips in the brain go, if you take the neural-ink interface and instead of using it for a Bluetooth device you do something simpler like having a scientific calculator chip that is so low in energy consumption that it can run of harvested body-heat, you'll get people that have several orders of magnitude better maths-skils, but more importantly it will make maths effortless and fun, you have no idea how transformative this would be, and unlike most other application it will never get outmoded. It's a much smaller step than turning people into a hive-mind but consider that maths is the closest thing to a universal language we have and by enabling people to express ideas that way you'll get closer to the ideal of unified communication without having any risk of unleashing the borg
>>713 Have you not seen the tech-startup Elon Musk fanboy types? They would get an implant just for the minimalism™. 20 years ago people would have said "I don't think there will be normalisation of constantly being tracked" or "being extremely open with all your personal details online" etc. Tattoos and piercings are less than exceptions, the majority of the population has piercings in a lot of cultures.'uncomfortable' is just a spook. I do however think it will take a long time until we have normalised injecting stuff into our brain. Having our car keys in our palm under the skin wouldn't be too crazy though. I'm not really sure, as a mathematician being able too do calculator-stuff in my brain just wouldn't be that useful. Anyone with any reason to have an implant, would already be too advanced to be using basic calculations in that way. The only place I can see it being useful is with kids, so that they don't get put off math at a young age, but I'm not sure since learning maths is about the 'way' in which you think as opposed to being fast.
>>715 I doubt that even among fanboys, you'll get many people that'll go through with it. Tattoos and piercings are fringe in every modern society. You are framing injecting implants as futuristic and people avoiding this as anachronistic. That is sort of a dishonest way of distorting my argument, I'm not saying that there is a cultural momentum preventing this, I'm saying that people will not adopt this because it hasn't got enough advantages to outweigh the downsides. Opening a car or paying with your hand is just a novelty, it's not substantially different from having that functionality in a separate item. Do you understand it's not a new ability it's just a different package. I clearly stated that the calculator-chip had to have the feature set of a scientific calculator, not just a basic one, you would benefit from this too, I sincerely doubt you can do very advanced maths entirely in your head. The main point is about reducing the mental effort as well as improving accuracy, it's not really about speed. I'm not sure how to convey this: consider that you could in principle look for mathematical patterns in every object you look at, but usually you don't because that would be very exhausting, and that's the bottleneck this would solve. Just consider that people could have accurate intuition for stuff that involves large numbers because they could calculate probabilities in their heads, with such low effort that it would feel like "intuitive knowing". You can't really dismiss ho much benefit this would bring.

(129.26 KB 744x389 serveimage (7).jpeg)
Electrical Engineering / Learning Hardware Comrade 03/03/2020 (Tue) 22:53:51 No. 339 [Reply]
I'm looking for recommendations on resources, learning materials, or general advice when it comes to learning about electrical engineering. Books, videos, MIT courses, you name it. I'm open to starting with the absolute basics from Physics 101, which I admittedly have not brushed up on in a while. In addition, I'm also interested in tinkering with hardware as part of the learning process, so any advice on that front would be greatly appreciated.
I'm looking forward to participate in this thread. I am a comp sci student and wanted to get more deeply into electrical engineering, but my exams are keeping me busy, so I can't start yet. If you speak German I would recommend to you Elektronik-Fibel, which gives a brought oversight over the basics, with easy explanations and a bunch of visualizations. Otherwise, perhaps look into The Art of Electronics by Paul Horowitz. That's a book that was recommended to me, but I haven't bought it yet.
https://www.toppr.com/guides/physics/electricity/ohms-law-and-resistance/ Start here lads, get some batteries, 3 cheap small lightbulbs preferably incandescent because the output from them is proportional to the input and replicate the two circuits here I can get youse up to flip flops gates at least
>>347 Oh yeah you'll want a few copper wires as well
>>348 Feel free to ask questions if you're stuck
Electrical Engineering isn't what most people think of it anymore. Try to learn about digital design and learn a HDL like Verilog or VHDL. Buy a cheap FPGA board and tinker with it by writing designs for it. You can dowload vivado from Xilinx for free if you take an older version.

Comrade 03/26/2020 (Thu) 20:23:25 No. 636 [Reply]
Intel (R) Celeron (R) CPU J3060 @ 1.60GHz 1.60 GHz Those are my PC specs, obviously I will search this by myself but I'm opening this thread for when I come back and see if you guys know if my computer is just useless for any sort of vidya, that's what I heard once, playing Java Minecraft is impossible and smaller games like Darkwood can only run at super low specs, and even then still with lag.
5 posts and 3 images omitted.
(5.89 KB 786x508 Nethack_releasing_a_djinni.png)
A perfect time to get into roguelikes.
>>646 You can use this site to see what games you can run https://www.pcgamebenchmark.com/ https://www.systemrequirementslab.com/cyri Scroll down until you see 'What games will it run'
>>650 Thank you so much, if I knew this site I wouldn't have even made this thread.
>>648 I'm going to shill for Dwarf Fortress because it's fun as fuck.
>>660 Honestly, Dorf requires a decent machine to run. Yes, it's basically ascii shit but it is very resource intensive ascii shit due to all of the incredible number of variables it has to run

(914.25 KB new_socialism.pdf)
Kickstarter workers unionize Comrade 02/19/2020 (Wed) 01:00:29 No. 161 [Reply]
https://news.ycombinator.com/item?id=22356972 https://www.vice.com/en_us/article/3a8pp5/kickstarter-employees-win-historic-union-election >The decision to unionize at Kickstarter follows a series of victories for union campaigns led by blue collar tech workers. Last year, 80 Google contractors in Pittsburgh, 2,300 cafeteria workers at Google in Silicon Valley, and roughly 40 Spin e-scooter workers in San Francisco voted to form the first unions in the tech industry. In early February, 15 employees at the delivery app Instacart in Chicago successfully unionized, following a fierce anti-union campaign run by management.
How big a victory even is this? Isn't Kickstarter basically just a parasitic finance capitalist?

bunker performance issues Comrade 03/12/2020 (Thu) 22:33:35 No. 461 [Reply]
As I'm sure most of you have noticed, bunkerchan has been having a lot of performance issues lately, with Nginx gives 500 internal errors while clicking on random threads or the catalog. When replying you occasionally get a 'connection lost' message. Image links are broken or disappear or don't expand when clicked. The site itself is much more sluggish and slow. The fact is the bunkerchan userbase has grown from 200ish people to about 750+ people and high amounts of engagement and posts per hour. This is good news but also brings the need for a higher level of technicality. What exactly is the problem? some possibilities: - Lynxchan is fundamentally unscalable. as many anons have pointed out this is actually the largest and most active lynxchan install ever created, so its possible that the code itself is shit at scaling having never been tested at that level before. - The server hardware is inadequate. It's possible the app server hardware is not adequate to the problem. The expensive solution, scale vertically by buying bigger, buffer servers. - Configuration. The fact that the errors are 500 seems to indicate not a problem with nginx, but the app on the backend that its reverse proxying (lynxchan). Maybe its possible to increase performance and scalability by using Nginx caching/microcaching of images/pages. This would allow present servers to be able to scale to many hundreds of more users, if done correctly, there are guides to this online but it requires mucking around in the dirty details of nginx config. Cloudflare itself already does alot of caching so the perf improvement from this might not be as great as a non-cloudflare site. any other ideas? (not a member of the technical staff of bunker).
20 posts omitted.
(69.67 KB 866x768 1579818286058.jpg)
>>506 I don't have one
(72.83 KB 650x507 gore.jpg)
>>506 >>519 I mean I voted yes in the poll but lets speak frankly. This was supposed to be a temporary space, then 8ch was obliterated and it seems like the new site owner isn't overly concerned for the users wants. If the only problem that could have ever happened due to that trait was with the rogue mod, we should be assured of that. Space_ himself wasn't very apologetic about it and was last seen saying he'd rather delete the whole site rather than do what the users wanted. So it'd be nice if when he asked for server money said something to the effect of "yeah that was fucked up, i cannot foresee this ever being a problem in the future however you should trust me :-)" the minimum gesture would go a long way. And if space_ doesn't want to do it, then the mod team of leftypol is going to have to once again step up and take the responsibility for what they're asking us to do. When this all possibly goes to hell and the people who donated feel robbed they need someone who will accept the blame for it.
(103.35 KB 562x574 idktho.jpg)
>>520 >>490 So the poll should really be: should we pay for a better server or should we start making plans to move to another site. And my answer in that case would depend on the future, if the tech guy who is supposed to be on break in a few weeks(maybe sooner now with the 'chink flu' as trump calls it) can fix all the little user-side annoyances and inconveniences in lynxchan then we should stay here. If the software is never going to get any better then i think maybe it is worth losing some users who get lost in the move and seeing if we can move to leftypol.org. I'm sure comsopolitonmongrel will give you the domain if you ask, he's said as much in discord when i met him. or another domain if that doesn't work out. Cause we're currently hampered by bad software and an absence of leadership. And we can survive one but not both. just my opinion.
>>621 I get these as well from time to time

Cyrilic Encoding in the 1990s? Comrade 03/24/2020 (Tue) 12:40:38 No. 611 [Reply]
>>>/leftypol/391425 >There are some Russian posts but they have some weird encoding >>376661 >I was reading about usenet a couple of months ago and stumbled upon a website that has an archive of posts during the August coup from the 19th of August to the 22nd. It has some posts on the website but most of them must be downloaded. They are in the .tar format but they can be read with notepad or notepad++, or any other text editor. >collapse.su
Fug second link should be >>>/leftypol/376661
It's either KOI8-R or KOI8-U, there's not much difference. I uploaded the original files along with them converted into UTF-8 here: https://files.catbox.moe/twrgxo.7z I accept payment in the form of meganekko.

(166.40 KB 1164x960 C_kk-kjW0AAJU5m.jpg)
Capitalism Ruins Everything Comrade 03/08/2020 (Sun) 08:59:52 No. 412 [Reply]
Thought this story was worth sharing. A bit of an interesting case example as capitalism ruining everything in the end no matter how benign it starts out and a good cautionary tech security tip/tale. https://krebsonsecurity.com/2020/03/the-case-for-limiting-your-browser-extensions/ The Case for Limiting Your Browser Extensions >Last week, KrebsOnSecurity reported to health insurance provider Blue Shield of California that its Web site was flagged by multiple security products as serving malicious content. Blue Shield quickly removed the unauthorized code. An investigation determined it was injected by a browser extension installed on the computer of a Blue Shield employee who’d edited the Web site in the past month. >The incident is a reminder that browser extensions — however useful or fun they may seem when you install them — typically have a great deal of power and can effectively read and/or write all data in your browsing sessions. And as we’ll see, it’s not uncommon for extension makers to sell or lease their user base to shady advertising firms, or in some cases abandon them to outright cybercriminals. >The health insurance site was compromised after an employee at the company edited content on the site while using a Web browser equipped with a once-benign but now-compromised extension which quietly injected code into the page. >The extension in question was Page Ruler, a Chrome addition with some 400,000 downloads. Page Ruler lets users measure the inch/pixel width of images and other objects on a Web page. But the extension was sold by the original developer a few years back, and for some reason it’s still available from the Google Chrome store despite multiple recent reports from people blaming it for spreading malicious code. You can click through for the rest but long story short the malware adds ads to the page(how late stage capitalism is that? Everything's about the ads.)

Message too long. Click here to view full text.

4 posts omitted.
>>437 If they are proprietary then yes it is, lol
>>437 dumb post. it's always fine until it isn't
(376.07 KB 1514x1308 inbox.png)
>>437 Even if you vet your addons (I always read the source code), there is a real risk from automatic updates and acquisitions. I'm the developer of a moderately popular open-source extension (< 100k users), and I constantly get spammed with this shit.
>>492 holy fuck porky really is hard for your extension
>>492 How do you monetize it if it's open-source?

(19.01 KB 361x327 D56ub8bWwAEirHj.jpg large.jpg)
When you HAVE to use Social Media Comrade 03/16/2020 (Mon) 04:05:25 No. 494 [Reply]
Sometimes you just NEED to use a certain social media site for something considering how most-all activity on the internet now goes through about 12 or so sites. I hate these social media sites, and you probably do to, but their ubiquity among the public means that sometimes you just need to use them for whatever reason. So, how can you use a social media site whilst maintaining some security? Obviously you can make a fake account, but what is the best way to do that? What resources can we use to do this? Thanks in advance.
I'm probably just an extreme introvert, but I never felt the need to use facebook, all I ever used it for was playing some of these browser games back when I was 12. but that's really only talking about facebook, I still use youtube to listen to music and use wattsapp to communicate with some people and there's probably not much of a difference there in terms of security and privacy.
>Sometimes you just NEED 'no'
>>504 I mean in our case if there is a lefty group or whatever that unfortunately still uses FB or whatever.
Many of these have become unbelievably invasive. Just as one example, you basically have to get yourself a burner cell phone just to make a Google account these days, because eventually they lock you out of your account unless you respond to a text message for two-factor authentication (re: identity triangulation). An option I have heard might be a possibility but still haven't tested out is to use JMP/Jabber with XMPP to simulate a phone for text messaging.
(118.58 KB 1280x872 stickerpack-2-iday.jpg)
>>494 Use alternative front-ends which reduce the amount of direct data the adtech companies get. Unfortunately they are read-only. If you need an actual account to interact with others >>513 is on the mark about how ridiculous it gets. Twitter - https://github.com/zedeus/nitter/wiki/Instances Youtube - https://github.com/omarroth/invidious/wiki/Invidious-Instances Instagram - https://github.com/cloudrac3r/bibliogram/wiki/Instances Try to use an instance that's different from the main one. (e.g. don't use nitter.net right now, it's being bombarded by Japanese people, and invidio.us gets its scrapers blocked somewhat frequently by Google).

(20.03 KB 720x673 FB_IMG_1558618234148.jpg)
Comrade 03/09/2020 (Mon) 12:16:13 No. 427 [Reply]
Hey so I installed mint on a laptop I got from Walmart after my old one exploded. (literally, buss error) So when I was installing I decided to wipe everything off the laptop in favor of an encrypted partition, well, after I rebooted I noticed something: I still have a working windows partition located on /Dev/sdb1 && yes, that is the actual file name it's located on. So what gives? Why the fuck do I still have a version of Windows installed and can I get the d of it with out breaking my computer?? Help /tech/
8 posts omitted.
>>450 man dd
>>450 just find and back up your configs if you dont want them lost. Or for some stuff like browser configs i just make a list of extentions, settings, etc. i want to keep
dont install mint lul
>>444 if you have another drive available, rsync -avxHAX /dir1 /dir2
>>443 actually, ignore him and just turn the ssd into a bcache cache drive. that way you won't have to move anything https://wiki.archlinux.org/index.php/Bcache


no cookies?