Linus Torvalds On GCC 4.9: Pure & Utter Crap

[Marc Driftmeyer]

The amount of self-righteous anti-anything-new morons on these forums are amazing.

GCC is and have always been a big pile of dog shit, if you can't admit this, you're either a hypocrite, didn't have glibc blow in your face or just an idiot that's using 4 years old version of it with 100 different backported patches.

Michael posts benchmarks all the time against clang, but guess what, as a compiler, clang is superior to gcc, live with it, also I'll bet money you're anti-systemd too, also guess what? it's better than init, deal with it and pull your head out of your ass.

Very bluntly and astutely stated.

[jimbohale]

A sane person:
There is a serious regression in GCC 4.9.

Linus Torvalds:
GCC 4.9 is utter crap.

No no no no you don't understand. A regression is something getting worse. This is literal breakage. Like breakage. A bug so bad that it shouldn't have been released.

[blackiwid]

No no no no you don't understand. A regression is something getting worse. This is literal breakage. Like breakage. A bug so bad that it shouldn't have been released.

so GCC now stands for kernel-only-compiler?

I guess there are a million other programms that just compile fine with that compiler, so its a regression. The kernel is not just the biggest project out there, so if something breaks its the kernel or nothing.

and this we are now big blabla... where are the 20 fulltime google developers or better 200 for the gcc project that is so important...

redhat ist by far not the biggest player thats business depends on linux, but if u not only count the kenrel devs but on all the other projects I am pretty shure they give back most to the community. or pays most of the community if u want call it like that.

[jimbohale]

so GCC now stands for kernel-only-compiler?

I guess there are a million other programms that just compile fine with that compiler, so its a regression. The kernel is not just the biggest project out there, so if something breaks its the kernel or nothing.

and this we are now big blabla... where are the 20 fulltime google developers or better 200 for the gcc project that is so important...

redhat ist by far not the biggest player thats business depends on linux, but if u not only count the kenrel devs but on all the other projects I am pretty shure they give back most to the community. or pays most of the community if u want call it like that.

If there is a bug that breaks something so horribly, it is a breakage. It breaks the kernel and who knows what else, probably more things than you know. That mindset is one of the many reasons the Linux desktop isn't taken very seriously by the population, you underestimate the severity of bugs.

[blackiwid]

If there is a bug that breaks something so horribly, it is a breakage. It breaks the kernel and who knows what else, probably more things than you know. That mindset is one of the many reasons the Linux desktop isn't taken very seriously by the population, you underestimate the severity of bugs.

Of course Desktopusers use the gcc to compile their kernels... ubuntu does something wrong I guess.

[jimbohale]

Of course Desktopusers use the gcc to compile their kernels... ubuntu does something wrong I guess.

Who the FUCK said it was bad for desktop users? It's bad for enterprise who wind up paying for most of the kernel development. Technically it's bad for everyone because it's a bug that breaks software completely. It should have never been released, and if you think it's even REMOTELY acceptable, please let me know which projects you have been involved in, so I can get as far away from using them as possible, thanks. You think that releasing a half-broken product is fine? Do you actually that's OK? Like, calling software with known bugs stable? What??????? This isn't the 1990s where it was a 'hey I'll contribute my time on this project' it's people with salaries releasing broken software.

[blackiwid]

Who the FUCK said it was bad for desktop users? It's bad for enterprise who wind up paying for most of the kernel development. Technically it's bad for everyone because it's a bug that breaks software completely. It should have never been released, and if you think it's even REMOTELY acceptable, please let me know which projects you have been involved in, so I can get as far away from using them as possible, thanks. You think that releasing a half-broken product is fine? Do you actually that's OK? Like, calling software with known bugs stable? What??????? This isn't the 1990s where it was a 'hey I'll contribute my time on this project' it's people with salaries releasing broken software.

Do u want a fish from me?

Haha, how can a or some bugs in a C Compiler bring people into not taking Linux Desktop as serious, its just retarded what u are saying, so linux as server is serious because there compilers dont matter at all, but for desktop where the users all day compile kernels its more important?

Ohh my gosh.

U btw claimed that it broke probably many other projects, can u proove that?

As far as I understand this its not even breaks kernel compililation in general, it depends on your kernel compile config or did I got this wrong, I dont care to much because I dont compile kernels since 10 years or so. I compile not even much other software. Because I use most of the time interpreter languages. So I am not hyper-interested into that issue.

Every bug sucks every regression sucks, but its not the first happend and not the last.

its opensource, if can do it better, do it. Or use windows if u think thats better, nobody forces u to use that non-serious Desktop.

Either nobody send a usable bug report, or nobody got payd for companies that cared about that part, or unpaid people didnt care too. So if u care for that part, do it yourself or hire somebody to do it. Its that simple. Thats what opensource is.

Its easy for Linus to troll around having 100x more people working on his kernel from many companies its easier to get not that big regressions.

Its like the openssl bugs, when google paid zero develoopers and a few guys drive internet security u get what u pay for.

[rabcor]

What a steaming pile of crap... I never fancied clang seeing it's affiliation with apple... but fuck me sideways, I was hoping the only thing it actually had over GCC would be the compilation speed, not the quality

Guess I have no choice now but to deal with it and just use clang, fuck! I don't want to use apple developed software... But I guess I can relax since Google, ARM, Sony and Intel are also involved.

Still, sometimes GCC is the better option, use GCC to compile programs that were made to be compiled with GCC I guess would be the wisdom here... Use clang for anything it can be used for otherwise, but rewriting the kernel to compile on clang, or implementing clang compatibility with gcc to the level of being able to compile the stock kernel at full performance would both be cumbersome, but I guess the former would be the way to go (i.e. kernel clang compatibility, if it's to be compiled with clang) there was such a modified kernel somewhere too.

[ultimA]

Come on people, all software have bugs, doesn't matter if it is GCC, CLang, Linux or MacOS kernels, open-source or closed commercial software, and the truth is, most of those bugs manifest themself way more often than a miscompilation in any compiler. Still, are you calling those an utter crap?

So what happened now? An older compiler bug manifested itself in a single version, on a single project, in some non-standard configuration. Is that really that bad? I guess from now, whenever the Linux kernel crashes on a specific computer in some configuration, you people should send Linus a letter about Linux being utter crap. Or if not just shut up and quit making a drama.

Some people on this thread make me laugh (or cry).

[blackiwid]

Come on people, all software have bugs, doesn't matter if it is GCC, CLang, Linux or MacOS kernels, open-source or closed commercial software, and the truth is, most of those bugs manifest themself way more often than a miscompilation in any compiler. Still, are you calling those an utter crap?

So what happened now? An older compiler bug manifested itself in a single version, on a single project, in some non-standard configuration. Is that really that bad? I guess from now, whenever the Linux kernel crashes on a specific computer in some configuration, you people should send Linus a letter about Linux being utter crap. Or if not just shut up and quit making a drama.

Some people on this thread make me laugh (or cry).

+1 thats what I tried to say too more or less