Linux: vim isn't moe

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > YOSPOS > Linux: vim isn't moe

«‹›11 »

Nomnom Cookie: Aug 30, 2009

do broadcasters even have OTA viewers at this point

# ¿ Nov 27, 2018 20:10

Adbot: ADBOT LOVES YOU

# ¿ Apr 25, 2024 21:17

Nomnom Cookie: Aug 30, 2009

Tankakern posted:

but is it really necessary?

shh someone with money might notice that computers are unnecessary and stop paying me

# ¿ Nov 28, 2018 02:30

Nomnom Cookie: Aug 30, 2009

Suspicious Dish posted:

The Tao of HashiCorp is the foundation that guides our vision, roadmap, and product design. As you evaluate using or contributing to HashiCorp's products, it may be valuable to understand the motivations and intentions for our work.

The HashiCorp approach is to focus on the end goal and workflow, rather than the underlying technologies. Software and hardware will evolve and improve, and it is our goal to make adoption of new tooling simple, while still providing the most streamlined user experience possible. Product design starts with an envisioned workflow to achieve a set goal. We then identify existing tools that simplify the workflow. If a sufficient tool does not exist, we step in to build it. This leads to a fundamentally technology-agnostic view � we will use the best technology available to solve the problem. As technologies evolve and better tooling emerges, the ideal workflow is just updated to leverage those technologies. Technologies change, end goals stay the same.

i too love programs written in go with weird config syntax

# ¿ Nov 29, 2018 04:00

Nomnom Cookie: Aug 30, 2009

Broken Machine posted:

i put os x on a system like this the other day; it's a hackintosh but it works wonderfully. it was a considerable process to get it up and running properly but it's great, has space for three nvme drives. i looked at all the macs available, and I couldn't find anything that matched what i wanted as far as hardware; the new mac mini was fairly close in ways but the integrated graphics and absurd markup for the ssd made it unsuitable. apple should make a new normal desktop; maybe they'll refresh the trashcan next

this is the furry porn of computer touching

# ¿ Dec 5, 2018 04:20

Nomnom Cookie: Aug 30, 2009

Captain Foo posted:

I don't understand your comparison

a shameful thing that if done at all should be done in secret, or as a joke

# ¿ Dec 5, 2018 04:26

Nomnom Cookie: Aug 30, 2009

Broken Machine posted:

i agree, it's incredibly shameful. i'm so embarassed i now have a portable mac i can also do vr and 4k gaming with as well. terrible. i should have just gotten an imac with equivalent specs so i can lug a huge loving 23" display with me instead

despite what certain parts of the internet believe, a discrete gpu is not necessary for survival. neither is osx. if youre on a trip or something where you need a portable for an extended period, and the only laptop you have is a 4 year old acer with a broken shift key, and this is significantly affecting your quality of life, some reflection is in order

# ¿ Dec 5, 2018 06:58

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

your operating system, running in ring 0 on your personal computer, connected to your home network, is running random software you downloaded from a pirate on an internet forum

because you didn't want to pay for a computer

are you broke, braindead, or both?

pirating OS's turned into a bad idea around the time MS blacklisted the FCKGW key

# ¿ Dec 5, 2018 06:59

Nomnom Cookie: Aug 30, 2009

the only good thing about Darwin is I lol every time I remember they built out a microkernel os and then had to shove bsd in it to get something usable

eat it tanenbaum

# ¿ Dec 5, 2018 08:01

Nomnom Cookie: Aug 30, 2009

r u ready to WALK posted:

It's not pirated though? You install it from an unmodified Apple image
Just make sure your computer has an apple sticker on it somewhere so you're not in breach of the license agreement

I'm not a huge enough nerd to audit the source code of Clover and FakeSMC but I highly doubt there's anything actively malicious in there.
I'm sure you open up some new security holes for local code but if anything you are probably safer running on an obscure combination of hardware and random emulated EFI drivers with a tiny install base than you would be on highly standardized computers where the attacker can develop their exploit on a 100% identical machine.

I guess none of it matters after apple inevitably switches to ARM laptops in 2020 and entirely refuses to run unsigned code

wow

um

wow

# ¿ Dec 5, 2018 08:05

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

that's not really a fair characterisation

the "designers" of what became darwin gave zero fucks about microkernels or macrokernels or os research or anything else. they just wanted to deliver an unencumbered unix as fast as possible

bsd, at the time, was mostly complete in userspace, but had some horrifying ties to at&t licensed code in the kernel. researchers had been hosting a faux-kernel based on bsd code using mach, in order to run test harnesses and poo poo on their research os

the NeXT people looked at that frankenkernel and said "gently caress it, ship it"

they didn't adopt a microkernel out of optimism, nor was it done out of stupidity. it was just greed and haste, like every other corporate project

�they shipped a research kernel� is even better lols

# ¿ Dec 5, 2018 08:10

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

my interpretation was the charitable one -- greed

yours is the uncharitable one -- stupidity.

why on earth would you write drivers, frameworks, and daemons that 100% depend on a single unix server if what you were really interested was mach-native applications?

that's ridiculous

idk i mean if you've already decided to use bsd-in-mach because that's fastest to ship, and you need some message-passing ipc, seems to me like it makes sense to use the mative mach facility for the one thing it does better than bsd

# ¿ Dec 6, 2018 03:35

Nomnom Cookie: Aug 30, 2009

eight megs and constantly swapping

# ¿ Dec 10, 2018 07:19

Nomnom Cookie: Aug 30, 2009

vim HEAD vs. emacs 18 would be a fun comparison imo

# ¿ Dec 11, 2018 05:16

Nomnom Cookie: Aug 30, 2009

Tankakern posted:

btrfs gets proper swapfile support in 4.21! finally

Btrfs Restoring Support For Swap Files With Linux 4.21

tl;dr remember nocow (chattr +C)

lol you use btrfs

# ¿ Dec 12, 2018 16:13

Nomnom Cookie: Aug 30, 2009

el dorito posted:

I legit laughed out loud when I saw this

yup me too

# ¿ Dec 13, 2018 02:07

Nomnom Cookie: Aug 30, 2009

ratbert90 posted:

Always in the mood for a good (and 100% justified) Linus rant.

Some idiot decided that breaking userspace for containers was a good thing, which broke several other things in userspace.

For those of you who don't know, the one unbreakable rule for the Linux Kernel development process is "ya don't break userspace."

Eric W. Biederman was notified of his pull request breaking poo poo back in JULY and he responded with:

Linus was notified yesterday about it and he was very VERY angry.

here is the commit message:

https://github.com/torvalds/linux/commit/94f82008ce30e2624537d240d64ce718255e0b80

no wonder he used to swear so much. the net effect of this change was to replace a failure that makes sense with one that doesn't

ewe2 posted:

Disappointed Linus is much more devastating than Ranty Linus.

if i was this biederman guy i would get a new job and maybe change my name

# ¿ Dec 24, 2018 03:26

Nomnom Cookie: Aug 30, 2009

Suspicious Dish posted:

> It has never been the case that mknod on a device node will guarantee that you even can open the device node.

Is this even true? lol.

it's fishmech true

# ¿ Dec 24, 2018 04:14

Nomnom Cookie: Aug 30, 2009

eschaton posted:

the kernel presents an API and ABI too, both to drivers and to userspace

�the kernel can never break userspace� is a stupid rule; there are all sorts of subtle behaviors a userspace program could wind up relying upon that aren�t guaranteed by spec, and instead of trying to define this problem out of existence a more rational approach is to figure out a way to balance keeping existing software running against not limiting the future

yeah ok there's that

on the other hand, linus says for what 27 years now NEVER EVER BREAK USERSPACE and still devs are constantly sending patches to him like "hmm what if we break userspace, that would be good actually". i can't or at least don't want to imagine what would happen if that stance weakened even a tiny bit

i like the api version idea. or steal from wsl and support windows syscalls lol. although i guess that wouldn't work in practice cause the nt equivalent is partly in userspace

# ¿ Dec 25, 2018 02:55

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

i have a debian desktop for historical reasons (aka laziness)

i would not consider debian to be "state of the art" under any circumstances, but it still shouldn't be broken as gently caress ootb

(also, fedora 30, which actually is meant to be state of the art, hosed me over with nonsense twice)

so your systemd is not under any circumstances ever reaping zombies?

# ¿ Jul 15, 2019 06:23

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

virtualization incurs virtually no i/o penalty on a linux host so you can be quite certain the problem is macos related

unless it does have a penalty ofc, which is the common case for consumers on consumer hardware

# ¿ Jul 16, 2019 04:33

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

pci-e passthrough is literally 0 overhead, but even regular ole' passthrough of scsi luns got you over 95% of bare metal performance, and that was ten years ago

(it is perhaps notable that the last time the question of i/o performance came up for me at work was ten years ago. now it's just a given.)

you do not normally emulate an hba in the year 2019, you use a driver designed for your virt environment that knows it's talking to the host

(not that emulating an hba was ever particularly expensive -- scsi is not rocket science)

this is a known problem on windows which is why a vm on windows is dogshit slow

linux doesn't work that way. open backing store w/ O_DIRECT, and bob's your uncle. no extra steps required.

perhaps there have been developments in the hardware area in the last 10 years that have made the cost of an extra context switch to issue IO more noticeable

# ¿ Jul 16, 2019 22:40

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

i very much doubt it

i can't disprove a negative for you

idk it seems pretty reasonable to suppose that a million hypercalls a second would be quite heavy

# ¿ Jul 17, 2019 04:37

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

the purpose of a hypercall is to be simpler and faster than a full virtualization

if a million hypercalls a second is worse than the prior implementation using a pure virtualized approach, somebody hosed up real bad

of course it's better, but it's also terrible on current hardware. some devices can do literally a million iops. at that point the kernel is a bottleneck just from being a kernel. involving the vmm as well is completely infeasible, even if all it's doing is passing IO straight from the guest to device queues

# ¿ Jul 17, 2019 18:40

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

the whole point of paravirtualization is to not invoke the vmm and trap instructions

at this point you are just loving with me. if you are doing a million iops, virtualization is not going to be what breaks your balls -- you are at the limits of what can be done with general purpose hardware. and this has nothing to do with the original case anyway. you're just moving goalposts to sound less like an idiot

to go back to the original point: doing linux on linux virtualization presents no i/o overhead. it is not a barrier to anything you would do on bare metal.

it may waste other resources (why run two kernels if you don't need to?) but it does not make your i/o mysteriously slow

yes if you are an expert and using fairly slow hardware then you can get the IO overhead down to pretty trivial levels. like everything else on linux, it works ok with significant caveats when set up properly. i did say that before, or close to it. do you really want to claim that this is what happens when a desktop user installs virtualbox and runs a vm

Sapozhnik posted:

Presumably the guest kernel has a page cache too so you're not doing one hyper call per guest write syscall

fsync. but that's workload-dependent ofc

# ¿ Jul 17, 2019 19:11

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

not going to make any claims about virtualbox. it literally taints the kernel with the 'C' flag when loaded. (C, for "crap")

but yes if you take a very ordinary linux desktop and install a VM using the native hypervisor, you should get very, very good i/o performance out of the box without any tuning or fiddling around

smart to not make claims about virtualbox, because virtualbox, the linux virtualization one can use without being an expert, is slow as molasses

and i appreciate that you were able to discuss KVM's out-of-the-box experience in a positive way. that takes some creativity

# ¿ Jul 17, 2019 19:24

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

i don't know why virtualbox even has a linux port

who is it for

does someone actually pay oracle money for this

installing the guest additions on a commercial machine obligates you to pay. ive always assumed its DB customers getting caught in a license audit

edit: not guest additions, the extension pack or whatever BS

# ¿ Jul 17, 2019 19:27

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

why would anyone ever bother to use that poo poo though

it is very bad and there are better alternatives that don't involve giving money to oracle

because you can use it without being an expert

# ¿ Jul 17, 2019 19:31

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

apparently you live in some bizarro world where installing a third party software package that requires its own kernel drivers to be compiled on the spot is "easier" than running a gui tool to manage the native hypervisor

the world of linux on the desktop

# ¿ Jul 17, 2019 19:37

Nomnom Cookie: Aug 30, 2009

thankfully, i lost the need to run a VM several years ago and have forgotten why kvm was dogshit. something involving virsh

virtualbox is dogshit too, but crunchy instead of sticky

# ¿ Jul 17, 2019 19:56

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

the dogshit expert has logged on

instead of doing anything fun or interesting in my youth, i used linux on the desktop. now i understand autotools and hate computers

# ¿ Jul 17, 2019 20:01

Nomnom Cookie: Aug 30, 2009

eschaton posted:

that doesn�t sound right

nothing about x11 is, or ever has been, right

# ¿ Jul 18, 2019 05:40

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

tl;dr: no one in this industry ever learns anything

this is not the first time some brain wizards had a x11 replacement. or the second time. or the fifth time.

this is not the first time vendors thought they could bully users into adopting something new but feature-incomplete

this is not the first time that the X project has found itself unmaintained. nor is it the second time.

X11 has survived people hating it for thirty years

there was never a time when X11 was the best-of-breed anything. it is just what actually had users. that status quo has weathered many challengers

i have no reason to believe, at this juncture, that wayland will be the one to finish this job. it is not impossible, but it doesn't look particularly probable at this moment

reading ur wayland posts but replacing x11 with blackberry and network transparency with hardware keyboard and it�s like I took a time machine to 2009. ofc in this scenario there�s no iPhone just low end androids so it really is hard to say what the outcome will be. my prediction is that wayland will become ubiquitous and so will x emulation. hard to go wrong expecting that an effort to improve Linux will end up making it cruftier and shittier

# ¿ Jul 19, 2019 05:16

Nomnom Cookie: Aug 30, 2009

Notorious b.s.d. posted:

an instructive tale:

NeWS was a pretty cool technology but it had an X11 implementation bolted on for legacy apps

result: no one ever ported anything to NeWS

NeWS is now dead and gone and X11 is still with us

moral: Unix was poo poo, is poo poo, will always be poo poo. otoh being able to shovel that poo poo without gagging is worth a lot of money

# ¿ Jul 20, 2019 17:21

Nomnom Cookie: Aug 30, 2009

Athas posted:

Arch is 2018. Run NixOS.

nixos has some neat ideas but it also breaks really basic stuff like mtime to achieve its ends. last I tried nixos, virtualenv didn�t work because of the mtime on system installed being set to epoch and to the nixos devs this was just impossible to solve, somehow. also the docs are poo poo and it�s so different from every other distro that you can�t really use docs produced for anything but nixos

and if the way they do things seems cool, just use chef or puppet. it�s the same poo poo but actually fits into existing systems rather than trying to reinvent the world

# ¿ Jul 22, 2019 06:16

Nomnom Cookie: Aug 30, 2009

the only stuff that runs better on Linux is horrible nerd poo poo. if horrible nerd poo poo games are a thing, I can believe someone saying they run better on Linux

# ¿ Jul 22, 2019 06:44

Nomnom Cookie: Aug 30, 2009

pseudorandom name posted:

that UI is generated at runtime from some legacy xkb bullshit

this explanation I can believe. overengineered with lovely UX is what you get when a dev tries really hard to do good work

# ¿ Jul 23, 2019 05:30

Nomnom Cookie: Aug 30, 2009

Progressive JPEG posted:

mac keyboards are decorative props used to fill the space between the touchpad and the display

true but they didn�t used to be

# ¿ Jul 23, 2019 05:31

Nomnom Cookie: Aug 30, 2009

what is even the point of desktop Linux if you�re not gonna use sweaty neckbeard Linux. none. there�s no point

# ¿ Sep 25, 2019 03:31

Nomnom Cookie: Aug 30, 2009

Tankakern posted:

you shut your mouth, gentoo with systemd is the best linux experience there is

this guy gets it. staring at console scroll and wrestling with breakage is what desktop Linux is all about. trying to avoid or deny that fact will just lead to even more pain in the end

# ¿ Sep 25, 2019 15:24

Adbot: ADBOT LOVES YOU

# ¿ Apr 25, 2024 21:17

Nomnom Cookie: Aug 30, 2009

alpine is for docker images and your containers are running in k8s so there�s only one ip in resolv.conf. no problem

# ¿ Sep 28, 2019 04:19

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > YOSPOS > Linux: vim isn't moe

«‹›11 »