|
and the there's the people doubling down on the boredom by going "well, the committee determined in paragraph 231:19 that..." posix is a pos far beyond even yos
|
# ? Oct 2, 2023 18:40 |
|
|
# ? May 27, 2024 21:56 |
|
unpopular take: "open standards" are an anachronism from the days of closed source vendors when you wanted to believe that you could just say "*harrumph*, well i suppose i will just take my business elsewhere" if ibm gave you an unfavorable renewal quote (and ibm would of course respond with "lol. lmao"). but you could tell your boss that you have the illusion of choice in a competitive marketplace of vendors and that was good enough for him. these days you have open source infra instead so you don't really need a marketplace of so-called "open standards", you need a specific software package that isn't under the control of one single entity who will always be looking for the ideal time to flip the rent extraction switch. notably we don't quite have anything like that for web browsers right now, because they are all centrally-controlled "open source" projects where the vast majority of the manpower on each individual web browser project all reports to some central for-profit body of senior management. as for kernels, well, try doing any sort of development other than ios or html5 on macos without there being any virtual machines involved and let me know how that works out for you. for backend services posix by itself isn't that useful, you generally need an implementation of the linux cgroups api as well, and the sum total of kernels that implement this api consists of, well, the linux kernel.
|
# ? Oct 2, 2023 18:46 |
|
Cybernetic Vermin posted:one of the things stallman got right is that focusing on which kernel things run is a really dumb way to understand an ecosystem. even dumber to focus on who makes the garbage unix utilities or overengineered libc of course, but correct that the kernel matters little. if the choice of kernel matters so little why aren't we all using gnu/hurd
|
# ? Oct 2, 2023 23:41 |
|
turns out stallman was completely wrong and the kernel is the only thing that matters and the linux shell script detritus that makes up "gnu" is totally worthless (hence the name free software)
|
# ? Oct 3, 2023 02:32 |
|
correct.
|
# ? Oct 3, 2023 03:42 |
|
BobHoward posted:if the choice of kernel matters so little why aren't we all using gnu/hurd what would cause the change if it doesn't matter?
|
# ? Oct 3, 2023 09:44 |
|
all the development ive ever done would run in literally any posix environment with probably an hour or twos effort. this should tell you all you need to know about both posix and my development skillset.
|
# ? Oct 3, 2023 13:07 |
|
oop - wait. theres a shitton of rexx in there and i dunno if orexx is fully posix portable. maybe not then…
|
# ? Oct 3, 2023 13:10 |
|
95% of all development i've ever done could run on any modern operating system with roughly 0 hours of effort but, taking the posix aspect aside, i really wonder how many people going "hell yeah linux rocks" would be the least bit inconvenienced if circumstances somehow developed such that freebsd became the obvious choice for e.g. money reasons. which is why i also don't get being real into kernels unless you're really into like device bringup work or some such.
|
# ? Oct 3, 2023 17:03 |
|
Cybernetic Vermin posted:95% of all development i've ever done could run on any modern operating system with roughly 0 hours of effort nice work if you can get it really pays the bills
|
# ? Oct 3, 2023 17:16 |
|
anyways, the fact that python has been ported and tested to all of the modern relevant computer architectures means that I don’t have to do that kernel work (or creating kludges for weird kernels) since someone else has already done it for me
|
# ? Oct 3, 2023 17:18 |
|
too bad python is utter poo poo
|
# ? Oct 3, 2023 17:21 |
|
fresh_cheese posted:too bad python is utter poo poo Why?
|
# ? Oct 3, 2023 17:31 |
|
load bearing whitespace is the stupidest poo poo ever
|
# ? Oct 3, 2023 17:35 |
|
fresh_cheese posted:load bearing whitespace is the stupidest poo poo ever kinda rude to your mom tbh
|
# ? Oct 3, 2023 17:37 |
|
i have spent more time working around whitespace in bash and yaml then i have in python
|
# ? Oct 3, 2023 17:41 |
|
Yeah, white space in Python isn’t a problem. Use PyCharm and stop caring. Also black, mypy, and flake8.
|
# ? Oct 3, 2023 17:46 |
|
fresh_cheese posted:load bearing whitespace is the stupidest poo poo ever i only recently had to use python and its fine lol just use an ide
|
# ? Oct 3, 2023 17:49 |
|
Significant whitespace doesn't really matter unless you don't normally indent your code in other languages in which case that's gross
|
# ? Oct 3, 2023 17:57 |
|
mystes posted:Significant whitespace doesn't really matter unless you don't normally indent your code in other languages in which case that's gross gob’s program doesn’t need indents
|
# ? Oct 3, 2023 17:58 |
|
i've only recently forgiven python for the incredibly incompetent job of the migration from 2 to 3, and with that out of the way i generally like it. also helps that there's some actual effort to not have it be just willfully slow and inefficient, i get that there's value in keeping things simple, but it was just wasteful how they approached things before 3.10. the typing situation even pretty good now, i don't mind it being optional as long as it is good.
|
# ? Oct 3, 2023 18:00 |
|
It's insanely dumb that python decided to do a backward incompatible break just for changing print to a function and fixing strings, which could have easily just been a version flag in a project configuration file if that was something that python had, and not fixing any of the important stuff like the GIL, package management, etc. otoh all the people who refused to migrate for years out of stubbornness were also loving idiots
|
# ? Oct 3, 2023 18:02 |
|
they also redid the entire standard library and make a bunch of breaking changes to the C API iirc
|
# ? Oct 3, 2023 18:06 |
|
mystes posted:It's insanely dumb that python decided to do a backward incompatible break just for changing print to a function and fixing strings, which could have easily just been a version flag in a project configuration file if that was something that python had, and not fixing any of the important stuff like the GIL, package management, etc. that would be one hell of a load-bearing feature-flag can you imagine feature-flagging every single stdlib function that takes a string, in order to make it take a bytes instead? and feature-flagging the parser, etc? also feature-flagging class definition syntax, and god knows what else? And marking flag-compatibility in literally every module in the stdlib? package management continues to be an insane garbage fire though. Poopernickel fucked around with this message at 18:28 on Oct 3, 2023 |
# ? Oct 3, 2023 18:25 |
|
the worst thing about python (and every other language) is colored async functions
|
# ? Oct 3, 2023 18:29 |
|
Poopernickel posted:the worst thing about python (and every other language) is colored async functions just use java, op it also does lambdas better than python, which is admittedly not hard
|
# ? Oct 3, 2023 22:26 |
|
There's no way the transition to handling strings sanely was going to happen without major breaking changes. That change touched basically the entire language, standard library, and ecosystem. It was always going to blow up. I absolutely resent the fifty different python packaging tools.
|
# ? Oct 3, 2023 22:39 |
|
also that there is no nice way to to package your app into a single exe file they could have owned the LOB CRUD space since like 2000
|
# ? Oct 3, 2023 22:48 |
|
https://www.phoronix.com/news/XOrg-Vulnerabilities-Since-1988quote:It was a decade ago that a security researcher commented on X.Org Server security being even "worse than it looks" and that the GLX code for example was "80,000 lines of sheer terror" and hundreds of bugs being uncovered throughout the codebase. In 2023 new X.Org security vulnerabilities continue to be uncovered, two of which were made public today and date back to X11R2 code from the year 1988. welp
|
# ? Oct 3, 2023 22:52 |
|
Poopernickel posted:that would be one hell of a load-bearing feature-flag we already have that anyway, it's called perl
|
# ? Oct 3, 2023 23:18 |
|
if you're running untrusted code on desktop linux and it's not in a vm then you've got bigger problems than xorg's swiss cheese security. maybe flatpak and bwrap will one day provide a meaningful amount of sandboxing but they sure as hell aren't there yet.
|
# ? Oct 3, 2023 23:26 |
|
Cybernetic Vermin posted:what would cause the change if it doesn't matter? gnu hurd actually existed before linux, and early linux was not the polished juggernaut it is today point being, kernels actually do matter and 99% of everyone switched from hurd to linux because it was obvious the upstart project led by a finnish rear end in a top hat was going places while gnu's own kernel project wasn't delivering
|
# ? Oct 4, 2023 00:13 |
|
Sapozhnik posted:if you're running untrusted code on desktop linux and it's not in a vm then you've got bigger problems than xorg's swiss cheese security. The nice thing about X server vulnerabilities is that you don't have to run the untrusted code on your own machine or outside a sandbox in order for it to be dangerous. It's a lot like a buggy browser running whatever JavaScript it sees. You can only shield yourself by running the X server yourself in a sandbox, which is contrary to what was supposed to be the point of its Byzantine architecture.
|
# ? Oct 4, 2023 10:22 |
|
BobHoward posted:forth is kinda gross though I don’t know if you know this but you’re pretty wrong
|
# ? Oct 4, 2023 10:45 |
|
BobHoward posted:gnu hurd actually existed before linux, and early linux was not the polished juggernaut it is today nonsense. 99% of everyone didn’t use gnu hurd to begin with. essentially nobody apart from gnu hurd developers has ever even attempted to boot a gnu hurd kernel. the switching that actually happened irl was from proprietary unix, and that was driven by cost-cutting more than any technical considerations. (cost-cutters didn’t switch to gnu hurd because ok sure yes, the kernel matters in the sense that it needs to actually work and be stable enough for production use and so on)
|
# ? Oct 4, 2023 12:10 |
|
I don't think anyone could actually even run Hurd back then, even if some code has been written. This was before GNU developed its software in the open. Although the licenses are mostly the same, the way free software was developed back then was quite unlike today.
|
# ? Oct 4, 2023 14:16 |
|
pseudorandom name posted:and that’s why we should use BSD atop Mach
|
# ? Oct 5, 2023 20:32 |
|
BlankSystemDaemon posted:Not even a little bit. POSIX defines a hell of a lot about the APIs a compliant system needs to provide (and their behavior) beyond just userspace and “a few libraries” — despite simultaneously being insufficient for real-world use (see for example the large pile of _np flags needed to make posix_spawn useful as a replacement for fork and exec in real code, despite the obvious superiority of not cloning a process just to replace it) beyond POSIX, there’s the whole UNIX certification suite, which is also large, tedious, and annoying in its detail, but at least loving Stallman didn’t come up with the name (one of the only actually-original things he’s done) both are sufficiently divergent from the real world that you’re allowed to distinguish official certified behavior from the usual behavior by means of settings in the environment, at both build and run time the two biggest successes POSIX has had are getting written into procurement in such a way that it’s not actually required to be used and propagating a bare-minimum portable multithreading API across UNIX variants
|
# ? Oct 5, 2023 20:42 |
|
Soricidus posted:just use
|
# ? Oct 5, 2023 20:46 |
|
|
# ? May 27, 2024 21:56 |
|
BobHoward posted:gnu hurd actually existed before linux, and early linux was not the polished juggernaut it is today don’t forget AT&T loving everybody with the stupid Berkeley lawsuit, that’s a huge part of what got Linux its momentum Linux would have been a footnote without it
|
# ? Oct 5, 2023 20:49 |