What's the deal with LLVM? Is it gonna replace GCC in our lifetime?
>>52118065
It will easily replace GCC in 4 or 5 years
>>52118065
Better do. GCC is a bugged pile of shit written by freetards with no actual clue of what they were doing.
>>52118065
One can always hope it will. Are you doing your part and using it yourself?
>>52118065
how would this nonfree abomination ever replace glorious GCC?
>>52118065
No because it's developed by retards and the documentation goes out of date every 3 months and new documentation only appears once a year.
Also, clang executables are slow as balls compared to gcc's and clang doesn't even support proper C++ completely.
>>52118456
https://www.youtube.com/watch?v=UtwaK-s9QRI
>>52118681
>reads good answer
>HAHAHA THIS IS NOT A USEFUL ANSWER LOL WHAT A FOOL LOL WE NEED A REAL ANSWER
>>52118720
>applefags
>>52118720
>everyone laughs at stallman and his retarded policy of purposely gimping gcc
>autist on /g/ claims this is fine and makes no counter argument while sperging out
Just another day on /g/ - Autism
>>52118065
It's not going to replace GCC, but it has already spurred the GCC developers to improve a lot to keep feature parity and a performance lead. Having two major Free compilers around encourages competition, which is awesome.
>>52118758
If GCC dies
I die
>>52118772
>It's not going to replace GCC
It absolutely is. If you compare their codebases you'll see why.
GCC's source is an absolute clusterfuck of spaghetti code that's extremely difficult to maintain and improve and discourages developers from contributing.
Clang's source on the other hand is neat and tidy and beautiful and the advantages show. Developers are flocking to Clang in droves and it's improving very, very rapidly.
This is on top of issues with political ideologies that prevent many really useful features on GCC. Clang doesn't have any of those issues and is quickly dominating the associated areas of use.
It's going to get to the point very soon where Clang is the only sane choice for several use cases, and just as good as GCC for the rest. When that happens, people are going to stop bothering with GCC at all and it will become a legacy compiler used to build old and abandoned software projects that never made the switch.
>>52118880
Oven if why would people compile propriety software on a open source compiler
>>52118065
Ken Thompson's compilers are going to replace gcc and LLVM. Why have layers and layers of VMs and translation layers when you can have one suite of simple and portable compilers that target different architectures? Easy cross-compilation is the only way forward.
>>52118880
>It absolutely is. If you compare their codebases you'll see why.
If you compare their codebases, you'll see why nobody is ever going to use clang for anything serious.
The GNU project will only ever use GCC or a comparable compiler under a free software license. As long as GNU exists GCC or something better will exist in some form.
LLVM will probably take over everything else because companies like the permissive license, being able to take and not give anything back is pretty nice.
>>52118065
Clang is already better than gcc.
People only use gcc at this point for compatibility and for gcc specific lock in features
>>52118065
No, it will never replace GCC on most Linux distros unless it adopts GPLv2
The freetards take the software they ship with their distro pretty seriously
>>52119064
LLVM is under a free licence though, the UIUC license... It's compatible with GPL, it's just not copyleft like GPL is.
>>52119091
>better
Implying
>inb4 cherry-picked benchmarks
You should know better than trusting compiler benchmarks, see icc for example.
>People only use gcc at this point for compatibility and for gcc specific lock in features
GNU extensions are now considered lock-in, good joke anon.
>>52118287
but gcc was written by Stallman himse... oh, right
>>52119180
That's the old GCC, Stallman has little to do with the new GCC (EGCS).
Its funny how GNU was supposed to be the hero's of free dumb and instead turned into monsters that locked software into their horrid licenses and filed lawsuits against even the minor of issues.
>>52119163
That's what I meant to say. Yeah the next GCC will have to be copyleft like its predecessors. rms doesn't want permissively licensed software because it can benefit proprietary software.
>>52119194
Stallman hasn't had much to do with anything in years.
>>52119208
The shilling and FUD is hard with this one
>>52119214
RMS doesn't get a say in what distros use. Debian is probably the most free-software-dick-sucking distro in popular use, and they have no problem with building their packages with clang when it's deemed ready for that.
Debian has an entire subdomain dedicated to tracking their packages incompatibilities with clang so they can eventually make the switch completely.
>>52119248
What they do internally matters little, they will never ship Debian with LLVM as the default compiler suite as long as it remains under a non-copyleft license.
In the end, what matters is which compiler they ship with their distro, not what the project does internally.
>>52118456
>nonfree abomination
>nonfree
It's FSF approved, based on the MIT/X11 license and the 3-clause BSD license. Naturally, it isn't copyleft.
Likely.
And it will be glorious.
Even if it didn't, it has already done enough. It has forced the lazy GCC cucks to step it up in the last few years.
>>52119331
As long as it's not recommended by the FSF, it will not be used by Gahnoo-fanbois.
>>52119286
"In the next few years, coupled with better static analysis tools, clang might replace gcc/g++ as the C/C++ compiler used by default in Linux and BSD distributions"
- Sylvestre Ledru, one of the major debian devs.
>>52118681
Whoa, he went full Adolf there for a second.
>>52119331
no copyleft shows it clearly supports nonfree software.
>>52119352
See >>52119342
It's not a matter of functionality, it's a debate about die-hard freetardist politics.
This is a perfect example of what I mean: >>52119375
>>52118065
It's open sores and even under a license approved by the FSF, so it's still shit. We need a corporate driven closed source compiler if we want this to be a success.
>>52119398
Meanwhile the people you're claiming to know the behavior of are in full support of using and shipping clang in the future.
You don't know what you're talking about.
>>52119398
Free Software is cancer, I can't wait till the FSF goes under.
>>52119423
Dude, these are the extremists that stopped supporting WINE and kicked the maintainer off the Debian team because he made a joke about women...
Also, you claim that they are currently porting, but you base this entirely on a statement in a blog post saying that 14.5% of their packages now compiles with clang (against 8.8% for the last stable release).
It literally says in the first line:
>Recently, I have been working on a side project for Debian. The goal was to rebuild of the Debian archive (the distribution) with clang, a new C/C++ compiler.
It's a one-man side project. Period.
>>52119481
>Also, you claim that they are currently porting, but you base this entirely on a statement in a blog post saying that 14.5% of their packages now compiles with clang (against 8.8% for the last stable release).
>22319 packages have been rebuild. Among them, 1320 (5.9 %) failed
http://clang.debian.net/
It's gotten pretty good. Most of them don't build because clang is stricter than gcc.
>>52119481
>14.5% of their packages now compiles with clang
12% of their packages *don't* compile with clang, and only because clang prints more errors than gcc with -Wall and they're using -Werror to turn all the warnings into errors. Clang compiles almost every package perfectly fine, there's just some that don't pass the automated builds because they're made specifically for gcc.
>>52119560
Oh, I read that the wrong way. Yeah, you're right. But still, it is a side-project, I would be quite surprised if the more aggressive "freetards" (in lack of a better word) would be convinced to ship LLVM as the default compiler suite.
On a side-note, has anyone gotten lldb-server to run on Linux yet?
>>52119562
Yeah, I read it wrong.
>>52119179
Sure thing gnushill
I still don't understand the difference between LLVM and Clang
>>52120125
LLVM is the actual compiler framework. Clang is a frontend for LLVM for compiling C/C++
>>52120125
LLVM is the compiler infrastructure
Clang is a front-end to LLVM
>>52119091
>clang can't even compile compiler-neutral C++ correctly
>better
>>52121243
Source
>>52121299
The Linux kernel. Ever heard of it, fucktard?
>>52121383
>The Linux kernel. Ever heard of it, fucktard?
It can mostly compile it now http://llvm.linuxfoundation.org/index.php/Main_Page
>>52121441
>mostly
Operative word you fucking retard.
>>52121441
>It can mostly compile it now
Doesn't matter.
Linux kernel is a hackjob that relies on non-standard custom GNU extension behavior.
Getting CLANG/LLVM to build Linux kernel and software has quite literally meant building hacky crap in to LLVM.
At least people like FreeBSD fixed their massive kernel and entire userland yet Linux kernel alone relies on so many cheap hacks thousands of people cannot fix it.
>>52121602
>has quite literally meant building hacky crap in to LLVM
They're mostly patching the kernel, not LLVM.
>>52121667
The LLVM source code has had a bunch of GNU hacks disguised as extensions inserted. It's disgusting.
>>52121706
They're removing VLAIS from the kernel, for example... Obviously, as opposed to adding it as a GNUBloat extension.
>>52121499
>Operative word you fucking retard.
It doesn't support nested function and VLAIS you fucking retard. VLAIS is not part of the C11 spec.
>>52119222
>>52118681
>>52118758
>>52119180
Microsoft shills hard at work
>>52121941
>>52119222
I like Stallman, but he said himself that he doesn't program professionally anymore.
>>52121383
>Linux kernel
>compiler-neutral
>C++
>>52118880
>clang
But clang is not LLVM.
>>52123290
>But clang is not LLVM.
When people say Clang they are also obviously referring to LLVM by extension, you dumbfuck autistic pedant.
>>52123308
No.
>>52123411
>No.
Non-autists have the ability to infer meaning through context.
gcc will never die you know why? Because the minute clang stops being foss, if gcc is better nothing will change, if clang is better, the gcc team will just fork the last free version of the code and continue development from there.
Stallman has no say in the gcc/emacs development so that talk about Stallman wanting to gimp it so it stays free is pointless.
Stallman is lost to politics
>>52118287
Daily reminder that clang and LLVM was written by "freetards" by your stupid definition.
>>52118065
i switched to clang when freebsd switched.
>>>>http://www.phoronix.com/scan.php?page=news_item&px=RMS-Emacs-Gud-LLVM
>>52124089
Hi dumbass. You'll notice I didn't define what a freetard is. Besides I don't have anything against freetards. I also said it was written by freetards who have no clue what they're doing. Try to read the full sentence next time.
Daily reminder next time you should shut the fuck up.
>>52124231
Nice damage control, faggot. You wouldn't mention the word "freetard" if you didn't mean anything by eat. You don't just bark a random offensive word for no reason.
Now, go back enjoying your shitty botnet OS.
>>52124268
>this projection
>offensive word
>thinks he knows what OS I use
That's /g/ for you. Also seems like you had trouble understanding my previous post's last sentence.
>>52124396
>thinks he knows what OS I use
Are you implying he was wrong?
>>52124396
Just an ordinary shithead
>Gets caught on his stupid logic.
>Turns on damage control and starts throwing random insults.
>Finally tries to stop the discussion by saying shut up.
>>52124449
He's implying that an unwarranted assumption was made.
