Looking At Changes In the Newest GCC

cyberpead writes "With GCC 4 comes a new optimization framework (and new intermediate code representation), new target and language support, and a variety of new attributes and options. Get to know the major new features and their benefits in this article."

They've optimized it so much...

[Anonymous Coward]

That they were able to reduce it down to just one C. Now it's just GC.

Re:They've optimized it so much...

[HTH NE1]

That they were able to reduce it down to just one C. Now it's just GC.

Of course. It's a GNU Compiler.

Re:

[Impy the Impiuos Imp]

Oh, GCC!

I thought they were making yet another Enterprise, the Enterprise-F. I'm like, "enuf is enuf!"

Re:

[An ominous Cow art]

GCC-1701-F?

Umm, gcc 4 was released 2 years ago

[vlad_petric]

Granted, 4.3.2 is pretty cool, but AFAIK it's not revolutionary wrt earlier 4.* releases

Re:Umm, gcc 4 was released 2 years ago

[Anonymous Coward]

But most developers still do not know the important differences between 4.x and 3.x other than the superficial ones like changes in headers that need to be included(ie stuff that breaks their code). For example, few seem to be aware that GCC does profile guided optimization now with -fprofile-generate and -fprofile-use switches(not even mentioned in the article).

Re:Umm, gcc 4 was released 2 years ago

[sxeraverx]

One of the things I personally am the most excited about is the ability to do function-level optimization in 4.4. Last year, I had a project that required me to have compiled code be as fast as possible, but you could only submit one source file and no Makefiles, which would be compiled with no arguments, optimizations, etc.. With this, I could throw the optimizations straight into the code, instead of having to compile with optimizations, taking the assembly, throwing that into a wrapper C file, and hoping the code was tested on the same architecture.

Re:

[Bert64]

Didn't know gcc did profile feedback, used to make use of that feature on sun's compiler a few years back...

Re:

[Anonymous Coward]

It's been available since 2.95, about 8 years ago.

There have been lots of improvement on the use of that information since then, but the flags have existed for a long time

Re:Umm, gcc 4 was released 2 years ago

[Anonymous Coward]

You missed one of the largest improvements in the four series was time travel. This article was actually published two years ago.

The Slashdot team are now frantically searching for the wormhole in their office. This is also why there are so many dupes, articles keep popping down GCC invoked wormholes.

Really, Stallman is just messing with us for modding-up comments like this one [slashdot.org].

gcc 4 is "new"?

[Anonymous Coward]

I've been using it for a year and a half now.

Re:gcc 4 is "new"?

[Anonymous Coward]

Wouldn't it be more logical to start supporting C++ 98 first?

Re:

[Anonymous Coward]

That deserves "sad but true" rather than "funny"... VC supports C89, but nothing more recent.

Re:

[mabinogi]

Not only that, but Microsoft have no plans whatsoever to add support for any C99 features unless they happen to accidentally do so when adding C++ features.

Re:

[XDirtypunkX]

As it has done since VC2003 thanks to the hard work of Herb Sutter? DUN DUN DUN.

Re:

[LingNoi]

You gotta be joking..

Re:gcc 4 is "new"?

[TheRaven64]

The article is about 4.x, and covers everything including the latest 4.3.x series. It's very light on details, however, so unless you know next to nothing about GCC you are unlikely to learn anything.

Re:

[psmears]

And it entirely fails to mention the 4.4 series [gnu.org]...

Wow...scary

[Anonymous Coward]

Wow! With all of that, you'd think I was truly enamored with GCC. Let's just say that when I'm developing software with GCC and my wife walks into the room, I feel a little uncomfortable.

That's a little creepy.

GCC?

[tacarat]

I read that as CCG at first... It sounded pretty good too.

Just taking his time I guess

[Anonymous Coward]

TFA says GCC 4.0 was released 2 years ago... yet it's titled "Getting to know GCC 4". I guess the author took his time to "get acquainted". Best not to rush into those sort of things...

Re:

[aesiamun]

He was too busy getting busy with gcc 3.x when his wife wasn't in the room :)

Re:Just taking his time I guess

[Doug Neal]

Seriously! From TFA:

With all of that, you'd think I was truly enamored with GCC. Let's just say that when I'm developing software with GCC and my wife walks into the room, I feel a little uncomfortable.

Yeah, GCC 4 has more backends, the little slut.

FINALLY

[kwabbles]

I've been stuck on gcc 3.4.3 for a few years now. Fortran 95 here I come!!

Re:FINALLY

[pablomme]

Are you implying that you're using Fortran 77?

That's just... gross!

Re:

[kwabbles]

Well up until today I haven't had a decent compiler for 95 - so what was I to do? Use FTN95? That's even more grosserest.

Re:

[pablomme]

Well up until today I haven't had a decent compiler for 95

Other than gfortran: g95, NAG's, PathScale's, Intel's, Absoft's, Sun's, Lahey's, Portland Group's, Compaq's...

For things other than PCs, there's IBM's, Cray's, Hitachi's, Fujitsu's...

You have no excuse! But please tell me it's not that you actually like Fortran 77... :)

Re:

[kwabbles]

Allright dammit, you got me. Who the hell uses PURE and ELEMENTAL anyways?

Re:

[pablomme]

Me. Plus modules, derived types, pointers, allocatable arrays, recursive subroutines, array sections...

Fortran 9x is likeable, even if it's still somewhat limited. I'd like to see Fortran 2003 become fully supported by major compilers in the near future -- but what I can't wait for is the co-array features of Fortran 2008, so that I can ditch MPI once and for all!

Fortran 77 is the reason why people laugh at Fortran.

Re:

[Silpheed]

"Fortran 77 is the reason why people laugh at Fortran."

Get off my lawn you lazy kids.

Re:

[coats]

Actually, the reason they laugh is that they're still stuck in 1960's-vintage FORTRAN IV, and aren't willing even to acknowledge the progress made for FORTRAN-77.

As an aside, going from "FORTRAN" to "Fortran" was a Fortran-90 change...

Re:

[pablomme]

To me Fortran 77 is equally laughable. Starting right from the punchcard-oriented fixed format of the source files. It feels old and uncomfortable, to say the least.

Re:

[elbarney]

Are you saying that in 4.x ifortran .gt. 77 ??????

Re:

[Random Walk]

Sure would be nice to have them. I suppose the gcc maintainers would never admit to someone from the openbsd group knowing what they're doing, though.

That belongs into (g)libc, not into the compiler. It's Ulrich Drepper who is blocking it (see this thread [redhat.com] on the glibc mailing list).

Re:

[setagllib]

Just copy them into your source tree. BSD license lets you do that, unlike most of the glibc (LGPL) extensions.

Let me know when GCC4 is available for MingW

[jonwil]

Big deal about all this GCC4 stuff, let me know when GCC 4.x becomes available for MingW as an official build (or better yet, when the GCC community stops treating Windows builds of GCC as second class citizens)

Party like its April 2005

[toofishes]

How the hell did this make front page news like 3 and a half years too late? Firehose fail.

One target language still missing: XML

[mma]

There is still one major target language missing: XML. Hopefully gccxml will one day be merged into the main source tree.

GCC to .NET?

[joeljkp]

Theoretical question here:

The diagram in the article shows GCC as a bunch of language front-ends that translate to an intermediate language, then a compiler for that intermediate language that produces the machine code.

Isn't this exactly the architecture of the .NET framework? Language compilers for VB, C#, C++ that compile to intermediate code, then interpreted directly instead of to a static machine-code file?

In other words, could a back-end be developed to change GCC into a universal runtime to replace J

Re: GCC to .NET?

[gidds]

Hey, wow! What a great idea!

Coz what with the JVM, CLR, Parrot, p-code machine, LLVM, Rubinius, SWF, Lua, Squeak, Dis, Waba, Z-machine, and a whole host of others, a new VM is just what we need!!!

...

Er, sorry about that. My sarcasm chip seems to have overloaded. But several existing VMs already let you run code written in many different languages. And issues of static (before-the-event) compilation can be rather different from dynamic (while it's running) compilation; the intermediate code may be w