/tech/ - Technology

Technology & Computing


New Thread[×]
Name
Subject
Message*
Files* Max 5 files32MB total
Tegaki
Password
Captcha*Select the solid/filled icons
[New Thread]


Imagine using PHP lmao


acf7c76504dffb0c45aace417b12a90d093683cf744d8de70b48d8910ca30a58.jpg
[Hide] (115.2KB, 1280x1832)
I have a lot of books and documents I want to read but I would prefer to do so on a device that doesn't emit bright light all the time and is portable as well.

An e-reader seems like the best option, but I know little about them and I have a hard time figuring out what to get. My requirements are:
- FOSS operating system, as I want to possess full control over my devices.
- Good durability (/price). This is my biggest concern. From what I understand the cells can only be toggled a limited amount of times. I don't intend to watch anything animated on it but I can see myself burning through quickly if flipping through a lot of pages.
- Ideally a touchscreen + pen, though I can live without it.

I'm also considering rigging one together myself, but finding a screen of appreciable size is hard and the few I do find cost as much as a whole e-reader.
8 replies and 8 files omitted. View the full thread
facbeb971de736a298d17acb8ae00577205d1d14330f4b16c30f5db545b3bc4e.jpg
[Hide] (5.2MB, 3024x4032)
a7c4d6f00630006d5ca43f3be40efb08b84ac44997500f907cfe0ce6a18d7d3b.jpg
[Hide] (5.4MB, 3024x4032)
6fb0e798f3f3ae93d9b25c40cf219953694fbce1b8bd24847e9534a536dd6c37.jpg
[Hide] (4.4MB, 3024x4032)
4b60cee1322596b854ec67e7b7b10700404928217ac6657a1ab78a38d081635c.jpg
[Hide] (4.2MB, 3024x4032)
4489902942ae376f522bed6cbf9e1037b6bc9efa8b80f7d7ff10d24e63603d29.png
[Hide] (33.6KB, 966x308)
I've hacked something together that's good enough for now. I can go forward with a single press and back with double press (I keep long press reserved for other functionality). Going back is a bit awkward though, since waking from deep sleep is slow and you need to time the presses properly for it to register (I'd say ~1/3s press, ~1/3 release then 1/3 press again).
The documentation is atrocious. They seem to have nuked their previous documentation and replaced it with something that's blatantly incomplete. I wouldn't be surprised if it's partially "AI" generated. Their own README's have a bunch of links that are dead. e.g.:
>We support peripheral mode over UART! The commands are listed here [dead link]
Amazing...

The examples are better, though still nonsensical in a lot of places. e.g. the example that loads from HTTP manually does a lot of manual work, including pointless double buffering on the stack, while making no mention they provide a downloadFile function.
You need to be careful with having both WiFi turned on and drawing to the display. It seems a brownout occurs and the MCU resets because both the display and the WiFi slurp a lot of power. Digging through the provided library I see that WiFi gets taken out of sleep at the start downloadFile, then back to sleep if it was (and not if it wasn't). I just put the WiFi immediately to sleep right after connecting which is stupid but an easy hack.

Loading PNG images is very slow. I first thought something was wrong with their HTTPClient since I could see the request in the log, yet needed to wait multiple seconds for the image to appear. However, it takes multiple seconds to load an image from SD card too. Poorly optimized PNG decoder I guess?
FWIW I found that with 3-bit grayscale PNG compresses significantly better than JPEG. Yet, most images hover around 200KiB, while  1200*825*3/8=371250 bytes if you were to use raw bits to store the image. It probably makes sense to use a custom format with lightweight compression on top.

In any case, here's the code I cobbled together. I plan to make something more advanced, where the Inkplate is basically a dumb terminal for my desktop PC, but I'm not going to use arduino-cli for it since I find Arduino wrappers to be shit. Too bad the Inkplate library is specifically designed for "Arduino IDE". By the way, I did find e-reader software https://github.com/turgu1/EPub-InkPlate but I don't seem able to convert from PDF to EPUB properly, hence why I just convert to images and display that. For now this suffices.

(python3 -m http.server is pretty useful)

Message too long. View the full text
Replies: >>17899 >>17906
6789678998.gif
[Hide] (1MB, 350x324)
>>17897
>Oh, they claim the Inkplate 10 is "not suitable as e-reader". Deceptive marketing?
So what the fuck is it good for?
Replies: >>17904
>>17899
An educational experience, like falling off your bicycle or buying a Mac.
>>17897
From the problems you described, I assumed it must have some kind of dinky 8-bit microcontroller like an AVR or PIC, but no, it's got 8MB RAM and an Xtensa LX7 740MHz 32-bit VLIW, equivalent to dual PII/400.

I guess the provided libraries are just astoundingly unoptimized
Replies: >>17914
>>17906
>Xtensa LX7 740MHz 32-bit VLIW
This one doesn't have that, assuming I have the right datasheet.
BOM here says ESP32-WROVER https://github.com/SolderedElectronics/Soldered-Inkplate-10-hardware-design/blob/main/OUTPUTS/V1.3.1/Soldered%20Inkplate%2010%20BOM.csv
Assuming it's the -E variant: https://documentation.espressif.com/esp32-wrover-e_esp32-wrover-ie_datasheet_en.pdf
>ESP32-D0WD-V3 or ESP32-D0WDR2-V3 embedded, Xtensa dual-core 32-bit LX6 microprocessor, up to 240 MHz
... which should still be pretty fast. The Inkplate library uses pngle https://github.com/SolderedElectronics/Inkplate-Arduino-library/tree/master/src/libs/pngle which is optimized for MCUs with little memory. Probably not the best choice for an ESP32 which does actually have a lot of memory.

huggingface.png
[Hide] (124KB, 1488x1374)
Discussion about "AI"s, deep learning, llms and others.
84 replies and 23 files omitted. View the full thread
>>17840
Well. I'm retarded.  I just had to make a symlink.  That's what I get for looking online for a solution instead of spending a few minutes thinking about it, I guess.

It does take minutes to load a big LLM like Mistral-Small from a spinning platter HDD, instead of seconds, but I can deal.  Never mind.
I noticed that AI search engines, aside from citing the top search results or Reddit, Wikipedia, Tripadvisor, etc tend to favor articles written by other AIs rather than ones written by real humans as sources. Has anyone else noticed this and have any ideas as to why, aside from the obvious (AI content being written with SEO in mind)?
Replies: >>17891
>>17850
>SEO
You have successfully answered your own question

In a related phenomenon, GPT-4 required all the English-language content on the entire Internet, including gorillions of terabytes of digitized books and auto-translated content from other languages.  GPT-5 requires four to five more orders of magnitude of training material than actually exist, so the researchers have hit upon the brilliant strategy of having it write its own training data, in a never-ending digital Ouroboros of bullshit.  Sending the output of the sausage machine back into the grinder will surely create more sausage than existed before, and surely nothing can go wrong, amirite guise?  Amirite?  Because clearly the AI isn't just glorified ELIZA plus glorified autocorrect, clearly the AI is actually making new sausage that did not previously exist.  Amirite?

If you own stock in any company that's putting a lot of resources into this, I think it's just about time to start selling short.
Replies: >>17894
>>17891
Muh AGI is a complete meme at this point, yet it's also the thing all the kikes invested so much money into they have to invest more to maintain the bubble so financial physics don't happen.
The real shekels lie in deepfakes, robotics and real-time conversational speech models, which also just so happen to see steady progress with plenty of room upwards compared to LLMs.

>>17840
>decent alternative
llama.cpp and vLLM exist.
>>17036
If you just want to coom then Rocinante-12B is probably the best option. Should be plenty fast as long as you have a decent GPU.

youtubeatitsfinest.PNG
[Hide] (57.5KB, 582x392)
I'm looking to exit Youtube as they've been blocking VPNs, ad blocking software, creating age verification, pushing globohomo channels more and more. I started this half-heartedly some years ago, so I know two channels I main outside Youtube, but stopped after that. Yes, I know I'm late to the party. Better late than never.
I know this doesn't seem too /tech/ related, but could we suggest good channels outside of Youtube and help others find their fix outside of Youtube? I know the tendency is going to be to shit on others for their tastes, but keep in mind that if you do that the individual involved is likely to stay on Youtube with that sort of attitude. The goal of this thread is more to get people off of Youtube whatever it takes.
I'll start:
Alternative channels I've already found:
Privacy news: https://odysee.com/@AlphaNerd:8
General news: https://rumble.com/c/RonPaulLibertyReport?e9s=src_v1_cmd
Alternatives I'd like to find outside Youtube/mainstream tech:
Hoe Math/that Romanian Vee guy/culture war news
Sabine/general sci news
Vtubers
Bokoen/Paradox map game shitposters

Thanks for any help.
20 replies omitted. View the full thread
>>16760 (OP) 
If you want to be in control of your content, peertube (or any other self-hosted federated alternative) is the only choice.
>>17843
stop using mpv+yt-dlp. use yt-dlp to download vid then watch it with mpv. if it's the space/cache that bothers you, download to /tmp or tmpfs, effect with be identical to mpv+yt-dlp
Replies: >>17880 >>17885
>>17879
While I do exactly that, mpv used to load video practically instantly and with settings you wanted automagically.
You have to wait for full download of audio and video which is a wait no matter what, I assume you can make a script to download audio first and then start downloading video and pipe both files to mpv for a faster way of opening videos depending on your connection, but I have no idea how to do that.
>>17843
cookies expire as soon as you close tab or browser in my experience.
>>16760 (OP) 
Freetube.
yt-dlp
>>17879
The problem might literally just be MPV, as instead of chaining yt-dlp more directly, it uses either a hook script or a virtual protocol. Other players chain in a more straightforward fashion, for instance, SMPlayer, and might have less/different issues.

ClipboardImage.png
[Hide] (401.3KB, 1007x573)
https://www.youtube.com/watch?v=7fGB-hjc2Gc
This nigger made a two hour long video explaining why C++ sucks, only to shill Rust while admitting that Rust has awful metaprogramming, no support for circular references, takes even longer to compile (right after bashing the compilation time of C++), only having a single compiler backend and sucks for prototyping and iterative development.
He should have shilled 'D' instead.

Okay there are some other gross errors in the video like telling people to use WebGPU over Vulkan because of claimed support by Apple when WebGPU is just a safety wrapper and even slower than MoltenVK. All his problems with MS Visual Studio seem quite irrelevant too.
At one point he talks about the problems of Qt but never mentions the most glaring one:
The Qt MOC compiler you have to use. If you're writing C++ with Qt, you're actually writing Qt++.
11 replies and 5 files omitted. View the full thread
Replies: >>17756 + 1 earlier
46608b1c52b1b1363fe2e9c5dd6d0a41a44f7150f784d30f9d29824783e5d775.png
[Hide] (618.9KB, 998x1080)
>>17662
>some out of the box system
Is this the latest jeetslang or what?
Rapid prototyping doesn't exist, nobody does that. What "rapid prototyping" means is that the developer half-assed some spaghetti and began randomly changing it without any thought and praying it works. The final version is the first version where the output looks right. 

In C++ you go through the entire compilation and you're able to produce a working binary that you can ship with such a development process. It may spit out 6 warnings per second during compilation, but it "works". In Rust you get a compilation error about trait bounds or other fancy terminology that is utterly meaningless to someone who doesn't actually know the language and development stops right there for a jeet who watched a 20 minute video and started copypasting stackoverflow and AI prompts, hence "why Rust sucks for game development".
Replies: >>17723
>>17709
Nobody managed to make a game in Rust yet.
Replies: >>17724
>>17723
Veloren.
>>17572 (OP)  
C[++] is infinitely worse than rust, let alone any language, and your some tunnel visioned neckbeard who can barely walk for not understanding something so obvious that first 4 month programmers understand

4dceffef47893be982e74c7fe78ce0cd.jpg
[Hide] (2.8MB, 2460x3540)
I ignored it at first because I already bought a new computer 5 years ago and probably won't get a new one for 5 more years, but I keep seeing people randomly talk about this across the Internet, so it might actually be a funny happening making consoomers cry.
But I wanna hear a quick rundown about this before I decide to waste any time taking a deeper look.
11 replies and 5 files omitted. View the full thread
historical_dram_spot_pricing.jpg
[Hide] (70.3KB, 1206x674)
It's just the RAM cartel up to their same tricks they've been doing periodically clear past the '90s, but camouflaged behind paper thin SJW AI derangement moral panic. It'll pass just like every other time once the lawsuits start hitting the dockets.

>>17661 
FAKE NEWS
Reposted from QTDDTOT
Unlike, for instance, the previous PoW ponzicoin fad that specifically gobbled up retail consoomer gayman HW, this current RAM price spike looks like just the same plain old pricefixing the cartel's been prosecuted for in 2002, 2006, 2010, and 2017:
https://www.xda-developers.com/dram-prices-spiking-dont-trust-industry-reasons/
Intentionally cutting production for years, hoarding inventory, shifting remaining production away from less profitable to the most lucrative chips, consolidation, all against a backdrop of constantly increasing demand and revenues.

This definitely would've happened even without some trendy datacenter bubble exacerbating it
Replies: >>17677
Oh, also
>>17665 
>globohomo special ram you can't even put in your pc
There is no such thing

Much cheaper "new" memory is populated by desoldering used DRAM chips from server DIMMs and even other devices like phones to reprogram and recycle onto standard consumer DIMMs:
https://www.tomshardware.com/news/used-ddr4-ram-chips-are-being-reused-in-new-pcs-dram-sticks
Replies: >>17677 >>17678
>>17675
>>17676
Thank you for actual good post.
>>17676
What about HBM and DDR5?
Replies: >>17679
>>17678
That depends on the course mobo standards take. Will we see more desktops with CAMM sockets instead of DIMM slots to increase pincount? So far, even servers don't do that yet for main memory.

But for devices that do use weird RAM, chips can be swapped or upgraded even on GPUs:
https://www.tomshardware.com/pc-components/gpus/modders-have-figured-out-how-to-slap-32gb-of-vram-on-the-rtx-5080-as-memory-shortage-intensifies
The market for which may open up significantly due to nVidia loosening the terms of their agreements with AIBs:
https://www.tweaktown.com/news/109145/nvidia-stops-bundling-vram-chips-with-gpu-dies-tells-aibs-to-source-their-own-gddr-chips/index.html
And even on hyperfascist MCM crap like Apple's:
https://www.idownloadblog.com/2021/04/06/apple-silicon-macs-ram-ssd-unofficial-upgrades/

The only way
Message too long. View the full text

ClipboardImage.png
[Hide] (22.5KB, 576x328)
Qt Company wants to bring the UI part (Qt Quick) of the framework to 'every' language out there.
>initially Rust, Python, .NET, Swift, Kotlin and Java. This is not a “yet another binding”

'Will it do LISP? Will GTK survive? Is the Electron apocalypse prevented?'
Come in this thread and talk about it.

https://www.qt.io/development/qt-bridges
13 replies and 9 files omitted. View the full thread
>>17637
Because it only works in an emulator nigger. Qt runs on real hardware.
Replies: >>17642 >>17649
>>17638
You're making shit up. I booted the CD on a crappy amd64 laptop a long time ago. I doubt it changed much since then.
Replies: >>17644
>>17642
On some 2008 laptop maybe but it doesn't work with modern hardware at all, no bios, no usb stack, no ahci, relies on bios interrupts that are no longer present at all, requires bios vga, no support for vesa...
Replies: >>17649
>>17638
>>17644
And Qt doesn't require an entire OS to work? How is this relevant? Your shill thread sucks. Get better material.
Replies: >>17651
>>17649
Terry used Ubuntu as his main OS.

8bb49823e6037db863f717df6d58d94e22b71e3db8d759ea15a134759765826f.jpg
[Hide] (19.1KB, 581x392)
Can we discuss main stream social  media and Google using drivers license to verify accounts? Is this being the norm now? You cannot get a Gmail using a burner phone anymore.
23 replies and 8 files omitted. View the full thread
>>16044 (OP) 
I have a single Google account to receive job postings and I never bother to use it for anything other than that. If YouTube flagged something as nsfw, I just don't bother to watch it.
>>16065
I would say it's less to do with sites and retarded parents and government overreach. Especially after propoganda slop like adolescence, they're just being safe than sorry.
>>16221
where exactly are you from?
Also, what alrernatives are you thinking of?
>>16347
what is this PRISM you are speaking of?
>>16044 (OP) 
>he wants a google account
Can we just admit that people wanting a botnet account on a botted platform with endless and constant censorshit is the problem? Because it's actually the problem. This idea that the entire internet is just google youtube facebook is such a new normalnigger golem concept and its incredible we went from typing urls to whatever the fuck dystopian monster this is.

Its very concerning how uncaring people are spreading their info around to "verify" whatever bullshit from these nose men technocrats. Ive seen people that really know better just upload their id to watch age restricted slop while coping about it. Letting them data mine you is participating in the humiliation ritual. It's also dangerous. Not just to the golem that obeys but to others that don't.

d90241b934a3105d264192295d667e9f.jpg
[Hide] (109.4KB, 703x1000)
It was well known that C was obsolete in the 90's. It could have been replaced with region inference a la ML Kit, Cyclone, etc. By 2000 computers were fast enough such that 90% of applications could just be written in a memory-safe garbage collected language[1]. Now what about that other 10%? Glad you asked.

What we need now (2010) is a return garbage collected languages with hardware support. Manual memory management is a complete waste of time and is a large source of program defects. Analog computing can be used to implement O(1), deterministic, garbage collection, suitable for even hard real time applications. Imagine a pipe from your water heater to a hot water tap on your kitchen sink. Here's a quick experiment: Have someone go to the water heater in your basement and listen while you turn on the hot water tap. Did you notice it *instantly* turns on?

Now here's one way O(1), deterministic GC can be implemented in hardware. There are millions of tiny water pipes. Each time you have an allocation, a water pipe goes from it to the code using it. If a structure has the pointer to the allocation, a water pipe goes from it to the allocation. When someone (code or a struct) is done using an allocation, it closes its end of the pipe.

All water in the system is in constant flow. When all ends of an allocation's pipe are closed, water overflows from that pipe, and it knows it's time to free it.

Q.E.D.

The CPU will have to be cubic shape to fit this system. Water do
Message too long. View the full text
12 replies and 2 files omitted. View the full thread
>>17569
Both are high level languages and there are no C cpu instructions.
Replies: >>17580
>>17569
Because many LISP features, for instance GC, are inherently inefficient. Think of how Emscripten transpiling C/C++ to JS results in code that runs way faster than idiomatic JS, specifically because it avoids using features such as GC as much as possible.
Replies: >>17579
>>17578
Most LISP languages are scripting languages with dynamic typing. I think that has a larger impact than GC.
Replies: >>17582
>>17570
Implement Lisp in CPU microcode?
>>17579
True, even more so for some languages such as Python.

meme-pfp-13_11zon.jpg
[Hide] (9.7KB, 512x512)
go check this out guys: 

https://github.com/LeviSantegoets/MOW

may it serve you well
Replies: >>17510
>>17506 (OP) 
>wangblows
Why would anyone here need it?

e956b6d9e44b7e663b3c81f254a96d33e5ab49057d58ea1eda0925e5df2e8560.jpg
[Hide] (20KB, 150x150)
i want to go back to when the internet wasnt commodified as well as our data and existance. Should i scrap everything and just go on anon imageboards? do i only go on .tor websites? or is there an internet 2 i dont know about?
4 replies and 1 file omitted. View the full thread
>>17064 (OP) 
You can always check out gopher. If you have no idea about anything then just install bombadillo, it works well enough,  and the default opening page is full of links and info. And if you are on gopher, you might as well visit bitreich.org for some extra autism.
Replies: >>17081
doors3.ans.png
[Hide] (5.5KB, 640x384)
>>17078
I don't know anything or care much about mastodon or social media sites in general. I pretty much just ignore them all.
As far as SDF.org, their shell servers have been online since the 90's, and I don't know of anything similar. There used to be a lot of shell servers (in the 90's...) but they're mostly extinct now.
Of course anyone can setup a shell server, but then you'll just have an empty box with no users (very much like if someone sets up a new imageboard). One of the interesting things about the 90's Internet is the interaction on your ISP's local shell servers. You could see who else was logged in and their running process ("w" or "who" command), and like if someone was telnet'ing to a MUD you could join up with them. You could even "write" and "talk" to them from the shell prompt in realtime. And when you played some /usr/games, there would actually be other people in the high scores and some kind of incentives besides just playing by yourself.
Dialup BBS's had a similar feeling of local user interaction, because all the local message bases were just those users, and you'd play door games with or against them. But most amateur BBS's only had one dial line or two, and being logged in to a Unix server with dozens of other users at the same time was something else altogether. Yeah, I don't know of any other site than SDF.org where you could experience that today.
>>17079
Completely forgot about gopher. Talking about usenet and IRC, how could I forget what should have beaten http? Also that site is awesomesaucebrah, bookmarked.
images_-_2025-09-01T232027.962.jpg
[Hide] (31.6KB, 609x452)
>>17064 (OP) 
I'd say smaller imageboards like this, and the few active forums for specific interests still online, are the closest there is to it.
>>17064 (OP) 
I wrote a guide on this:
https://websitereview.neocities.org/search

Show Post Actions

Actions:

Captcha:

Select the solid/filled icons
- news - rules - faq -
jschan 1.7.3