/tech/ - Technology

>KDE
Pozzed bloatware. Check out Lumina instead.

>>80
Doesn't look like anything LXDE, Blackbox, or MATE isn't, but maybe I'll give it a try later desu

>>81
It's very lightweight and free from dependencing on RedHat cancers such as dbus, polkit or elogind.

>>82
dependencies*

>>82
>>82
>RedHat cancers such as dbus, polkit or elogind.
only thing that makes linux usable

i had to use vi the other day and i wanted to kill myself
i'm an emacsite through and through

neovim is better

>>126
What's the actual difference between vanilla Vim, Neovim, and Vim6? Ive never cared enough to find out on my own.

>>128
Neovim has cleaner code and is extensible in lua in addition to vimscript. It is aimed at improved extensibility.

Vim6 is just a version of vim as far as I know. Vim is on vim 9 these days.

>>124
The true pinnacle is to use evil in emacs. Emacs and vim are not really comparable, I don't know why everyone tries to do so. A virtual lisp platform is not a text editor, it just happens to do text editing in its default mode.

>>129
Although Emacs style editors have always been the de facto base of Lisp platforms, and GNU Emacs certainly shares that heritage, I wouldn't call it one by itself. ELisp may be a full fledged Lisp, but it's left lacking in many ways that keep it down into being a scripting language.
Some of the magic behind a full Lisp system is lost when you keep hitting C brick walls. I have seen however that work has been done to integrate a native ELisp compiler into newer versions, which may one day alleviate this issue.

>>51
>or best of all: don't use HTTP.
BASED

I sneer at those uninitiated to OpenBSD httpd

they both suck, use Caddy.

>>51
If Uriel knew what was truly harmful, he wouldn't have killed himself. Untrustworthy.

>>125
on a surface, unenlightened level, yes.
however uriel surpassed the buddha and so the universe sought to evict him forcefully for his crimes against the dharma. he is free from both samsara and nirvana now.

>>93
Even if WebAssembly is a better step towards solving the problems the JVM tried to, it doesn't take away from the stance that these problems never needed to exist. Native code runtimes will always outperform bytecode, and the prime advantages of a bytecode VM can be adequately addressed by providing expected features like compile and run time type checking, exceptions, conditions, error recovery and restarts. All features that have existed for decades in Lisp but have only begun to be explored by mainstream languages. Infact, John McCarthy's Lisp compiler is oft to be considered the first JIT compiler, compiling expressions as they are read and running from resident memory instead of saved tape. It was also the first system to utilize garbage collection.
Lisp's own homoiconicity leads to extremely novel and profound development flows that don't entirely exist in other languages all in an impressively efficient runtime, and despite how influential it is remains to be a niche in the modern world.

>>96
The advantage of a bytecode VM isn't speed at all but trust. It's just another form of sandboxing that enables new and more immediate forms of untrusted software delivery. Like how webapps "won" because you can install them by loading a webpage (with PWAs you can now literally install them onto your computer as well, interestingly). WASM and eBPF don't replace existing software, but rather create new classes of software that couldn't previously exist…

I have awalys found in my experience that writing domain specific bytecode runtimes into my programs has created very flexible software; think like TrueType bytecode; interestingly the HarfBuzz text shaping engine now embeds a WebAssembly runtime- you can now inject WASM blobs into font data that can then be invoked by the shaping engine in order to execute the internal logic of the final text layout.

The proliferation of eBPF is also interesting… Linux is now deeply extensible through this mechanism, that allows internal kernel code to trap to this JIT compiled bytecode runtime in ring 0- so you can write ring 0 kernel space programs in eBPF- which allows you to run your untrusted code deep inside the kernel- not just in the network stack anymore, but also inside the storage drivers, and more.

>>100 (checked)
I thought stuff like PostScript and Metafont was really cool, portable and specific processing like that is a good application of VMs, and it's funny to think there was a period where an office printer was more powerful than the machines around it.
eBPF… that sounds insane, but kind of neat. It might actually be a step towards breaking away from the limitations that Linux imposed by being destined to be a UNIX clone.
My remarks about Lisp runtimes were more about program stability (C is a plague). Trust is a social problem over a technical one, and we already let all kinds of arbitrary code run on our machines all the time, so why not go the whole nine yards?
It seems more to me that the industry is approaching the same ideas that were tried years ago in projects like Plan 9, except with none of the simplicity or consistency of those models. Apparently they could get away with disk access being as fast as a file server in 1993.

You could also have a flexible language from the beginning. Embedding an interpreter in a static language adds infinite flexibility, but how about new functionality, and even new syntax without having to change the compiler? That's how the original OOP module for Lisp was written, it was ordinary Lisp code that generated new code to expand at compile time. The simplicity and powerfulness of ordinary Lisp macros is amazing when you get down to it, and I'm wondering if any other languages have done a similar thing.
It seems Clojure is getting its traction in the industry. I've never tried it, and it seems to have its flaws and rounded edges, but if you do work in that kind of field it might be worth looking at.

>>104
Fetching a file from the network has been faster than local disk access for many years now. Look up "race network with cache" in web browsers- browsers will re-request network resources because it's faster than local storage sometimes. The cost of downloading it over ethernet is lower than bringing data all the way up the network hirearchy; if a server has a file loaded into RAM already, it can win a race with your disk.

You should look into modern Linux APIs (or rather the lack of APIs) such as io_uring… Linux is applying the programming techniques that graphics programmers pioneered in the 2010s with AZDO and Vulkan- not APIs to talk to the GPU, but rather the lack of an API- using mapped command buffers- to avoid the cost of calling into the GPU driver. Now we have a system call interface that looks exactly like that- we no longer have to make system calls to do the most common operations. Windows 10 has a similar system to this.

Trust is definitely a technical problem. Just look at all the places we fit x509 certs into- which as a technology is a way of creating accountability- (although letsencrypt has turned it into a rubber stamp on the web). There are so many programs that we couldn't write without the strong sandboxing garuntees to allow a user to feel confident in running it, knowing it cannot scrape data, or engage in other malicious behavior.

I need to read more about lisp… in general high level code. Its easy to write interesting C programs just by exploring new technology, which gets exposed to C by default- but as a measure of power, that's really only truly useful when you use it to make runtimes for high level code.

starforce

>>58
I did this in school, had a lot of help, very informative but wouldn't do it again

Everyone wants to design their own language when in fact it'd be better to try and find a good use for some existing smaller compilers

Tor does indeed glow, and the "deep web" is ironically less expansive than the surface net

Tor is funded directly by DARPA and other US federal agencies, so that's all you should need to know.

>>91
I thought this was a chuck e cheese birthday party for a moment, then I enlarged the image

>>91
>hanging out with people irl is cringe you should be isolated and miserable like me
Nah.

>>91
Leave FreeBSD out of this

Sad to see. The installer is actually a convenient script that set up reasonable defaults.