You shouldn't remove bugs because that would break some retard's program that was built on that bug

lkml.org/lkml/2018/8/3/621
>you shouldn't remove bugs because that would break some retard's program that was built on that bug.
Why does Linus encourage bad coding?

Attached: workflow.png (278x386, 34K)

Other urls found in this thread:

retroarchleaks.wordpress.com/
yarchive.net/comp/linux/gcc_vs_kernel_stability.html
twitter.com/NSFWRedditImage

never really read his mailing list but he sounds like a typical Jow Forums sperg. wouldn't be surprised if he was shitposting here.

Hes basically saying the kernel needs to be able to upgrade independent of any other piece of software. Not that hard to understand.

Nice strawman.

I can only imagine how does maintaining of such project affects your life (including politeness)

>he sounds like a typical Jow Forums sperg
You should get out more. Generally people get angry and go into rants when someone fucks up their job.
Garlic boys on the other hand stay passive, polite no matter what... And tyen go home and cut their dick to release the tension.

>politeness
Sugar coating shit so as to have someone eating it without gagging or throwing up doesn't change the fact that he is eating shit.
People need to realize that the meaning is the same whether you used polite or offending language.
I despise those who say "you don't know me, you should respect me"... All those fucktards think that they are a special sbowflake.
Linus does this for 2 reasons, first off his words are getting relayed left and right due to his somewhat cringy responses,
Secondly everyone who's involved in this would read it twice and memorize it so he doesn't fuck up and be in the same fate.
Your ancestors were fighting lions, fucking hookers with chlamydia and beating their slaves on their free time... What's wrong with you people? You can't even handle a simple cursing on a mailing list?

This.
Fixing bugs is not an excuse for breaking applications. You should fix the bug so that the applications also don't break.

>never really read his mailing list but he sounds like a typical Jow Forums sperg
Reminder that Linus Torvalds has gone on record multiple times saying that he found Ubuntu to be too difficult to install.

It's good seeing Linus taking backwards compatibility seriously.
If only other developers were as well.

>only thing that matters IS THE USER.
I wish Linus was running something like GNOME. It might have not been such a steaming pile of shit that it is now.

I think the user should upgrade lvm2. Why bother fixing bugs at all? You never know what piece of software used the bug.

>But the "no regressions" thing is _so_ important to me, and _so_frustrating when I get the feeling that people ignore it, that I just go ballistic.
>I'm sorry for going ballistic, but there it is. It keeps happening every once in a while. I thought we'd gotten mostly over it.
Top kek

>I think the user should upgrade lvm2. Why bother fixing bugs at all? You never know what piece of software used the bug.
there are major and minor versions for a reason.
You can't just change the major version of a program just because you found a bug that e.g. breaks userspace.
You have to circumvent that bug and fix it in the next major release.
Most likely this bug will be properly fixed in the 5.x version of the kernel.
You'd think that after 25 years of showing people what is important and what's not in linux development, people would get it right...

Everything here made sense. What exactly do you disagree with? Hasn't Linux been known and alienating enough by instability for the longest time? I'm sure you've heard of "it's not a bug, it's a feature". Why introduce even more options for failure? I don't understand how any of what he said promotes "bad coding".

>newfag doesn't know the linux version numbers are meaningless

If only python had a lead like this, it would be usable...

If only python had a lead like this it would still have a lead.

"Never break the userland", he takes it seriously.

This, don't be a brainlet OP

later in thread
>On a funny note - security exploits had also many users - so why fixing them....

kek, why does Linus allow removing these features

Fuck off.

All bugs become features eventually.

DON'T BREAK THE USERLAND! REEEEEEEEEEE

It's called realism vs idealism. If you don't plan around the way things actually are instead of the imaginary perfect scenario that only exists in your head, you're going to have a lot of problems both in programming and in other aspect of life.

Attached: 1530825539897.jpg (1440x1080, 126K)

Security exploits get fixes years after they are written so it checks out

The head of the RetroArch project is also kind of an ass about breaking stuff in order to do "things right" too, but he'll gladly accept broken code (it's free!) and also introduce suicide-inducing changes that drives all but the most autistic/passionate developers away.

The main code is written in a way that predates C89 because msvc/gcc 2.96 didn't support it properly until few years ago and it sucks because things like having thirty variables as the first lines of the function is annoying as fuck and it is not uncommon for bugs to appear as a result of uninitialized memory access because somebody overlooked something.

One time the project ran into many of those bugs in a single month so he got mad and started adding {0} everywhere and the framerate halved in old computers because some of those initializations happened in loops. {0} turns into memset or rtosb when you compile with gcc, imagine zero-filling multiple PATH_MAX-long variables a thousands of times every frame in a system with slow memory.

Another time there was a bug in the Android builds that resulted in NULL pointer dereference but nobody knew where what triggered it so he put null checks EVERYWHERE even in places where a NULL wasn't possible (e.g. pointer to stack variable). Whenever a crash happened it was a massive work to find out where it happened because every subsystem simply let NULLs pass unnoticed. The code is still littered with these checks.

Yeah I heard, that guy is a fucking autist. There is even a website dedicated to archiving his autism: retroarchleaks.wordpress.com/

>NOT AN ARGUMENT

I've read that before

Attached: 1395862898547.jpg (200x300, 11K)

I just went quickly through a few posts on the site and JESUS FUCKING CHRIST! All the memories and feelings of hate for how ungrateful he is just towards people who spent YEARS helping him with his autistic endeavor came back.

I'm glad I ghosted the project. I wish I'd done it sooner.

lol xdcd
go back

>bug fixes should not break the userspace
It's like all the pansies working on the kernel can't adhere to this simple philosophy.

That's not bad actually. It's ridiculously good. If he exposes an interface to user space, that shit is frozen forever and will remain that way forever no matter how shitty, misguided or buggy it is. He does this because he wants people to be able to upgrade kernels without shit breaking all over the place. Other projects like glibc don't subscribe to this strict ABI support philosophy and as a result they break your programs when their special snowflake devs decide to refactor some shit.

THIS ONLY APPLIES TO KERNEL TO USER SPACE ABI. Shit like device driver developers better track kernel updates or they WILL get left behind. And when they bitch about it they get roasted on mailing lists publicly for all to see.

yarchive.net/comp/linux/gcc_vs_kernel_stability.html

>Cry me a river, Jeff.

Basically, if you're an user space program maintainer and the kernel breaks your previously working program, Linus will revert whatever it is that broke your software and yell at whoever's responsible until they fix it. But if you're a linux kernel contributor (drivers, modules, subsystems, etc.) its in your best interests to upstream your shit so that the whole kernel hacker community can help maintain your shit for you. If you don't, absolutely not a second of their time will be spent fixing YOUR outdated code. They receive like 10 pull requests/hour. It's pretty much upstream or die. This is quite unlike how Microsoft operates, IIRC they'd even support these faggot's mistakes if there was money on the line.

Onions onions

How can one man be so based? He really doesn't seem to give a shit about anything other than having a stable user space ABI and a reliable release schedule for his kernel.

If someone is causing issues with that he'll just tell them to fuck off in no unclear words. Really the hero we need.

Attached: the-only-masterrace.png (680x451, 253K)

You get it, my nigga. Linus is so fucking based I seriously worry about what will happen to Linux when he retires/dies. I believe his 'lieutenants' can push the project forward, as long as they too train a new batch of lieutenants in the Linux ways.

Linus is what makes it possible to scrap the entire Linux user space and put your own shit on top of it. It's the central hub where all the software converges, and if your language can make a bunch of simple machine language calls then you can access the full power of Linux, not even libc is required. Everything in Linux user space is merely using system calls in some meaningful manner. I/O? Shitloads of useful system calls. Video? Mostly ioctls. It's all there, and everything is stable and supported because of Linus.