Slashdot Mirror
Free Optimizing C++ Compiler from Microsoft
FortranDragon writes "Microsoft has made the command line toolkit for Visual C++ available for a free download. You can use the toolkit to build applications and redistribute them if you want (though you should read the EULA for the details, as always). This is a nice boon for those that have to deal with cross-platform compatibility, especially since Microsoft has tried to make Visual C++ more conformant to the ISO C++ standard. Go forth and compile your favorite OSS or FS programs today. ;-)"
The platform SDK which includes the command line compiler, linker and debugger has allways been a free download (hasn't it). Also, the .net SDK which includes all the languages, libraries, and compilers has also allways been a free download, this is what sharp develop uses.
There's a decent free IDE available called Dev-C++ for windows, it comes with mingw32 ofcourse.
As usual, they won't let you view the license before you download it.
Karma: Segmentation fault (tried to dereference a null post)
This is absolutely awesome.
.NET framework SDK did not have optimization command-line options enabled.
While there were legal methods to obtain optimizing compilers before (you could order media-cost-only Windows 2003 DDK cdrom and with that you'd have vc++ 7.0 compilers), but with this kit there's no longer any problems.
While VS.NET 7.1/2003 IDE is quite well developed, I'm sure projects such as eclipse or dev-c++ can benefit from this.
Also notice the important thing about this kit is it includes optimizing compilers, free command-line tools which were available for a while from the Microsoft
Before this, there were ways to obtain the optimizing version of vc++ compilers, by ordering a recent Windows 2003 DDK (it included vs.net 7.0 compilers, if I remember correctly). But now with this kit,
Because you're a retard who doesn't know how to code C?
.. do you think that could have something to do with it ??
.. it strips out the HTML tags if you set it as Plain Text.
.. heh its a quiet night.
Or how about because I pressed Submit accidentally instead of Preview
#include <stdio.h>
not that I have any idea why im responding to an AC
Funtage Factor: Purple
like gcc, for free
No, not like GCC. MS's compiler is free in the sense that you do not have to pay anything for it. GCC is also available at no cost, but it is free in sense that it is an open product, developed using the principles of free speech.
MS provides the Debugging Tools for Windows as a free download. GUI and command line debuggers are included.
Major omission - no lib.exe for building .a files
.NET as without lib.exe it is pretty useless for larger projects.
Looks like this is just to encourage people to migrate to
Still, nice of them to release it
Asmo
Sad as it is, VC++ probably would compile that. One of the reasons MS wasn't ANSI standard was that it would do things like let the programmer use printf() without using or .
.c file to cpp and recompiled. Strange results....
One of the weirdest though is the time I renamed a
-B
just to clarify this Apple's dev tools ARE included, just NOT installed by DEFAULT.
...and now that I'm thinking of it most OSes of recent years, free AND proprietary seem to come with some sort of gcc based plus IDE dev tools any more anyways, e.g. BeOS, *Linux, *BSD, OSX, MacOS 7.5-9.x etc. On top of this didn't M$ used to have pretty decent pricing(surprisingly) for VC++/dev studio only anyways? (It's been a while since I've bothered to check.)
EVERY new machine SHOULD come with a separate developer's CD just as every boxed version of OSX should. The end user just has to apply some elbow grease in a) noticing this, and b) bothering to install it.
Also in the unlikely event that the developer CD is not present or there is an upgrade it is ALWAYS available(even to free online developer members) as a download. Hell even the old classic dev toolset(name escapes me at the moment) has been available free since, what?, 1994/5 or so?
The only commercial environments are things like Codewarrior(Metrowerks), Absoft FORTRAN, RealBASIC, etc. Codewarrior MIGHT be worth it if you are trying to make money as their compilers are proprietary and USUALLY optimize much better than gcc. (I don't think that anyone has done a recent set of benchmarks of Apple's latest improvements of gcc v. current Metrowerks compiler though...)
For OSX dev tools come free on the software update cds.
xcode, interface builder and more.
also see The Fink Project
groklaw, wired and slashdot. The holy trinity of work based time wasting.
The very first thing it does is try to configure .net on your machine.
Since my only Windows machine is 98, it couldn't complete the rest of the install.
Yes, it is a sacrificial test machine. I just wanted to see the EULA.
I don't have VC++ installed, so I'm afraid I cannot cite very specific examples. Most of my troubles came from pointers (big surprise there), and if I recall correctly calloc() would randomly fail to initialize allocated memory. I also recall being angry that I cannot sizeof() a row vector in a 2D array, but sizeof() on a 1D array works fine.
Most of my trouble likely came from being a n00b to the language, and coming from a Pascal background I don't quite like the way C does a few things things (like multi-dimensional arrays, and pass by reference).. but using VC++ made learning even harder, becuase stuff would just mysteriously fail to work post-compilation that worked fine on other compilers.
(Disclamier: never touched C++ with even a 40 ft stick, and was using an older version of VC++, I think it was 4)
DJ kRYPT's Free MP3s!
Dev C++ http://dcplusplus.sourceforge.net/
Windows is only $500 if your time is worthless.
The "learning edition" of MSVC++ prohibited you from commercially distributing binaries made with it (or perhaps even just distributing the binaries, I can't remember just now). It also displayed a pop-up window whenever you executed your code, reminding you of this.
However, considering it cost virtually nothing to get the thing, I don't think this was particularly unfair.
I haven't used MS VC++ for a few years (fortunately), but does it still get the scope of loop variables wrong? So that this:
{
for(int i=0;i!=num;++i)
{
}
for(int i=0;i!=num;++i)
{
}
}
would complain about i being declared twice? That used to really annoy me.
Large code size != bad code generation.
Considering a very common (and powerful) optimization of inlining. What you literally do is change function invocations to the code for that function.
Another common (but more advanced optimization) involves making an optimized duplicate of a particular code path.
What you're probably seeing is a difference between the included optimizations in GCC's -O2 verses MSVCP's -02. Try enabling individual optimizations and then comparing code size if that's what you're interested in. That's a more fair metric.
int func(int a);
func((b += 3, b));
I know knowledgable comments are looked down on at Slashdot, but the .NET SDK has been free from the start and included cl.exe, (the C++ compiler) but it was the non-optimizing compiler I think. What the .NET SDK didn't have was the full STL and a few other things that are now free.
The end-result of this is everyone gets a free compiler upgrade if they haven't moved to VC7.1 already. It means on my OSS project I can say "VC7.1 minimum," and expect everyone can follow suit without a problem. However I realize this isn't as fun as tin-foil explanations, so I'll let you get back to that.
The Mono group (http://www.go-mono.com) has been working on achieving platform independence with .NET and C#. Programming in C#, in many respects, has the feel that C++ is just getting ancient. C# constructs and methods, although not entirely mature, have learned from the best object oriented languages that we have right now.
.NET and Mono.
C# code for Microsoft's compiler (of an ECMA standard programming language), should run as advertised on a Mono CLR environment - once it becomes mature. Look at the many cross-platform (Windows/Linux) classes available for both
Since Mono is an open-source effort, Mono could be extended to MacOS X, FreeBSD, etc.
True portability is one source code - many OSes and architectures.
Ayup
I use a macro that handles this for me, with conditional directives to be empty on other platforms. Not that hard - really.
This is a few years old (as are most of the benchmarks i've seen) but you can try this http://www.willus.com/ccomp_benchmark.shtml?p11
Note that this is comparing MSVC 6 (current version is 7.1, and it's improved a lot). In this test, Intel beats everyone, MSVC comes in second with MingW third, however, both the Intel and MS compilers have improved massively in the last couple years. I don't know about MingW. At least part of MingW's success is that they use MS's runtime library, which means that a great deal of the code being run was compiled by MSVC.
If you need web hosting, you could do worse than here
However, considering it cost virtually nothing to get the thing, I don't think this was particularly unfair.
Not to be unfair, but I've got a fairer solution. Just season with a genetic optomizer.
I'm continually amazed at the number of folks that bash MS for not following standards and then ridicule them when they do try. I'm certainly not an MS evangelist (nor an apologist), but seriously, what do you expect them to do?
There has been some level of support for the Standard C++ Library (STL) since the 4.2 days. Granted it was broken and has been through 6.0. This is because they used an implementation written by PJ Plauger who worked for Plum Hall at the time. Plauger had a falling out with Plum Hall over the rights to said code. This is the legal battle that has caused problems since the 4.2 days. You can easily download the half dozen or so header file patches for this though. FYI - Plauger now has his own company "Dinkumware".
I have not used their latest compilers to know the current situation. Although there have always been standards compliance problems with MS compilers, at least they're getting better over time.
karma whoring, but anyway, here is the EULA:
END-USER LICENSE AGREEMENT FOR MICROSOFT SOFTWARE
MICROSOFT VISUAL C++ TOOLKIT 2003
IMPORTANT-READ CAREFULLY: This End-User License Agreement ("EULA") is a legal agreement between you (either an individual or a single entity) and Microsoft Corporation ("Microsoft") for the Microsoft software that accompanies this EULA, which includes computer software and may include associated media, printed materials including best practices, white papers, templates, "online" or electronic documentation, and Internet-based services ("Software"). An amendment or addendum to this EULA may accompany the Software. YOU AGREE TO BE BOUND BY THE TERMS OF THIS EULA BY INSTALLING, COPYING, OR OTHERWISE USING THE SOFTWARE. IF YOU DO NOT AGREE, DO NOT INSTALL, COPY, OR USE THE SOFTWARE.
MICROSOFT Software LICENSE
1. GRANTS OF LICENSE. Microsoft grants you the rights described in this EULA provided that you comply with all terms and conditions of this EULA.
1.1 General License Grant. Microsoft grants to you as an individual, a personal, nonexclusive license to make and use copies of the Software (i) for your internal use; (ii) for designing, developing, testing and demonstrating your software product(s); and (iii) for evaluation of the Software.
1.2 Documentation. You may make and use an unlimited number of copies of any documentation, provided that such copies shall be used only for personal internal purposes and are not to be republished or distributed (either in hard copy or electronic form) beyond your premises except as otherwise specifically provided herein.
2. ADDITIONAL LICENSE RIGHTS -- REDISTRIBUTABLES. In addition to the rights granted in Section 1, certain portions of the Software, as described in this Section 2, are provided to you with additional license rights. These additional license rights are conditioned upon your compliance with the distribution requirements and license restrictions described in Section 3.
2.1 Sample Code. Microsoft grants you the right to use and modify the source code version of those portions of the Software identified as "Samples" in REDIST.TXT or elsewhere in the Software ("Sample Code") for the sole purposes of designing, developing, and testing your software product(s), and to reproduce and distribute the Sample Code along with any modifications thereof, in object and/or source code form. For applicable redistribution requirements for Sample Code, see Section 3.1 below.
2.2 Redistributable Code-General. Microsoft grants you a nonexclusive, royalty-free right to reproduce and distribute the object code form of any portion of the Software listed in REDIST.TXT ("Redistributable Code"). For general redistribution requirements for Redistributable Code, see Section 3.1, below.
3. LICENSE RESTRICTIONS -- DISTRIBUTION REQUIREMENTS. If you choose to exercise your rights under Section 2, any redistribution by you is subject to your compliance with the following terms.
3.1 If you are authorized and choose to redistribute Sample Code or Redistributable Code (collectively, the "Redistributables") as described in Section 2, you agree: (i) except as otherwise noted in Section 2.1 (Sample Code) to distribute the Redistributables only in object code form and in conjunction with and as a part of a software application product developed by you that adds significant and primary functionality to the Redistributables ("Licensee Software"); (ii) that the Redistributables only operate in conjunction with Microsoft Windows platforms; (iii) to distribute the Licensee Software containing the Redistributables pursuant to an end user license agreement (which may be "break-the-seal", "click-wrap" or signed), with terms no less protective than those contained in this EULA; (iv) not to use Microsoft's name, logo, or trademarks to market the Licensee Software; (v) to display your own valid copyright notice which shall be sufficient to protect Microsoft's copyright in the Software; (vi) not to remov
Try not to let life get in the way of living.
I'll stick to BCC55 then...
This is basicaly the same EULA as the regular VC++ package. It restricts what you can do with Microsoft's code, but not what do with your own code. Get over it!
Check out section 3.1.i and 3.1.ii, which EXPLICITY DENY you from distributing your source-code.
You can't distribute THEIR source code. Do whatever you want with your own code.
6 is even better: "NO RENTAL/COMMERCIAL HOSTING.
No rental or commercial hosting OF THE COMPILER itself. Do whatever you want with your own code.
They are. You have to register, though. Make up a dummy account if you don't want Apple to have your real info. You can find development kits with IDEs for use with 10.2 and 10.3. I use Xcode quite a bit, although I'm also learning to work with vi and make.
(tig)
Ignorance and prejudice and fear
Walk hand in hand
Why not just develop for .NET? this way you can compile using either Microsofts compiler or either of the free open source compilers Portable.NET or Mono), and run it natively under either of the said runtimes.
And yes, this works. I have a project I'm working on now that I wrote in VS.NET2003 and ran with no problems using Portable.NET
I can compile with Portable.NET and it works under Windows. I can compile with Microsoft c# compiler and it runs under Linux. Just like Java, minus the insanely ugly and slow Swing widgets.
Example: You create a storage class (my example was a self-redistributing binary tree). The template class Storage<T> has a subclass Storage<T>::Element that holds three Element pointers (parent, left child, right child) and one const <T> pointer.
Of course, the Element class has a constructor, a destructor, a copy constructor, and an assignment operator defined. (Copy constructor and assignment op may not be formally required, but my experience says it's a very good idea to define them in anything that holds pointers).
As in:This code will produce an error. VC++ does not handle functions in named classes inside templates that are not declared inline.
(I may have missed some syntactic detail - It's been a long time since I wrote C++; I write C# these days. But the above annoyed me to no end. It works fine with gcc, which on the other hand can't handle when the functions are implemented inline, which should also work)
Side note: I absolutely love Microsoft's dev tools. This particular example is the exception to me, not the rule.
There are at least two easy answers to that question:
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.
.NET runs on Windows 98. You can get it here. The SDK is also available from that page.
Of course there is. You get pass by reference semantics by passing or returning pointers. It's not as clean as C++'s passing a reference, but that's just syntax, not semantics.
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.
Speaking of standards compliance...
there ISO
their ISO
it's not there fault.
it's not their fault.
I have quite successfully linked C++ libraries compiled with GCC, VC++, and the IntelCC compiler with no problems. It appears to be ok. Perhaps code that relies on internals of STL would fail, but at least pointers to STL objects actually port back and forth.
I agree that VC++ produces faster code than GCC. Don't really know how much, because we don't use GCC on Windows anymore because of this, except to make dependencies for the makefiles. Can't be too much today, as our GCC-compiled Linux versions seem to run at the same speed as the IntelCC compiled Windows version.
The IntelCC compiler produces code almost twice as fast as VC++6. I believe VC++7 is much better, however, but have not tested it.
VC7 is head and shoulders above VC6 for standards compliance, including templates. Our company is switching mainly just to get better templates compliance. We do not always upgrade; we skipped VC5, stayed on VC4.2 until VC6.
I'm not a compliance expert (I'm more of a perl hack), but the people at the company that are into advanced C++ code like the new compiler very much.
You can use POSIX via Windows Services for Unix.
Through my employer, I took a C++ course taught by the secretary of the ISO C++ committee. Most of use were using Microsoft Visual C++. Three of use used g++ (GNU) 2.95.3. Our instructor used Microsoft Visual C++, and a Greenhills compiler. For interest, we tried various programs to see which compilers required that we follow the standards and which would let us get away with non-standard code. The Microsoft compiler would let use get away with the most and sometime actually required that we don't follow the standard.
The system requirements list Win2000 on up and you complain that it won't install on Win98. Classic.
1.1 General License Grant. Microsoft grants to you as an individual, a personal, nonexclusive license to make and use copies of the Software (i) for your internal use; (ii) for designing, developing, testing and demonstrating your software product(s); and (iii) for evaluation of the Software.
As in, for demonstration only. If you're going to distribute binaries, don't compile them with this tool.
I [may] disapprove of what you say, but I will defend to the death your right to say it.
As it says on the downloads page: "Supported Operating Systems: Windows 2000, Windows Server 2003, Windows XP". The page also points out that it does try to install .NET (not that anyone reads the page...)
There's suprisingly little in it, the vast majority of it is standard boilerplate. Only the first clause relates to what you can do with it, and it seems to limit you to personal use only (ie, no selling the binaries you create), but it's not explicit. The rest of it deals with standard redistributables (the C/C++ runtimes/STL) and standard waivers of liability, etc.
Presumably because it's distributed/sold in Quebec, theres a bunch of French at the end which, in the interests of space, I have snipped.
END-USER LICENSE AGREEMENT FOR MICROSOFT SOFTWARE
MICROSOFT VISUAL C++ TOOLKIT 2003
IMPORTANT-READ CAREFULLY: This End-User License Agreement ("EULA") is a legal agreement between you (either an individual or a single entity) and Microsoft Corporation ("Microsoft") for the Microsoft software that accompanies this EULA, which includes computer software and may include associated media, printed materials including best practices, white papers, templates, "online" or electronic documentation, and Internet-based services ("Software"). An amendment or addendum to this EULA may accompany the Software. YOU AGREE TO BE BOUND BY THE TERMS OF THIS EULA BY INSTALLING, COPYING, OR OTHERWISE USING THE SOFTWARE. IF YOU DO NOT AGREE, DO NOT INSTALL, COPY, OR USE THE SOFTWARE.
MICROSOFT Software LICENSE
1. GRANTS OF LICENSE. Microsoft grants you the rights described in this EULA provided that you comply with all terms and conditions of this EULA.
1.1 General License Grant. Microsoft grants to you as an individual, a personal, nonexclusive license to make and use copies of the Software (i) for your internal use; (ii) for designing, developing, testing and demonstrating your software product(s); and (iii) for evaluation of the Software.
1.2 Documentation. You may make and use an unlimited number of copies of any documentation, provided that such copies shall be used only for personal internal purposes and are not to be republished or distributed (either in hard copy or electronic form) beyond your premises except as otherwise specifically provided herein.
2. ADDITIONAL LICENSE RIGHTS -- REDISTRIBUTABLES. In addition to the rights granted in Section 1, certain portions of the Software, as described in this Sectionÿ2, are provided to you with additional license rights. These additional license rights are conditioned upon your compliance with the distribution requirements and license restrictions described in Section 3.
2.1 Sample Code. Microsoft grants you the right to use and modify the source code version of those portions of the Software identified as "Samples" in REDIST.TXT or elsewhere in the Software ("Sample Code") for the sole purposes of designing, developing, and testing your software product(s), and to reproduce and distribute the Sample Code along with any modifications thereof, in object and/or source code form. For applicable redistribution requirements for Sample Code, see Section 3.1 below.
2.2 Redistributable Code-General. Microsoft grants you a nonexclusive, royalty-free right to reproduce and distribute the object code form of any portion of the Software listed in REDIST.TXT ("Redistributable Code"). For general redistribution requirements for Redistributable Code, see Section 3.1, below.
3. LICENSE RESTRICTIONS -- DISTRIBUTION REQUIREMENTS. If you choose to exercise your rights under Sectionÿ2, any redistribution by you is subject to your compliance with the following terms.
3.1 If you are authorized and choose to redistribute Sample Code or Redistributable Code (collectively, the "Redistributables") as described in Section 2, you agree: (i) except as otherwise noted in Section 2.1 (Sample Code) to distribute the Redistributables only in object code form and in conjunction with and as a part of a software application product developed by you that adds significant and primary functionality to the Redistributables ("Licensee Software"); (ii)ÿthat the Redi
This is not new. The .Net framework free d/l includes V# and VB command line compilers. In fact, I think that Visual Studio uses those under the covers. If you need SQL, you can use free MSDE ... yes restrictions, but 2GB db is enough for some work.
Not for standard compliant C++, bucko.
XML causes global warming.
Latest issue (May) of DDJ has an article on compiler performance.
For C code, VC++ 7.1 ranked 2nd (tied with Intel and Open Watcom) whereas GCC ranked 8th.
For C++ code GCC and VC++ 7.1 were tied for 2nd.
I'll leave it to you to buy the magazine if you want the details (they tested 9 x86/Win32 compilers all together).
It should be possible -- I've integrated the C compiler from the DDK with VS6, and that's basically the same compiler as what MS is giving away now. If you're happy with the VS6 IDE, this is a good way to keep the IDE you like, while getting the benefits of the latest compiler.
Select the Tools | Options menu and go to the Directories tab of the Options dialog. Add the directories for new C compiler to the top of the lists for Executable files, Include files, and Library files. I'd also recommend downloading the Platform SDK and using the Windows headers/libraires from there rather than the ones that came with VC6 (which date back to the NT 4.0 era).
Actually, looking at the compiler flags makes me wonder whether there's any point in taking those bench's seriously - the guy didn't even bother to use anything more than -mcpu=i686 (aka 'generate for pentium pro) for gcc, and the other compilers' switches are still close to minimal (even if they mean quite more than the gcc one). Makes one wonder about how competent the guy doing the benchmark was.
Intel's C++ compiler is also free for non-commercial use. It even runs under Linux.
Just wanted to note. If you install the dev tools on MacOS X, you get perl, python, ruby, c, c++, objective-c, java, php, bash, and probably more right from the start. You even get emacs and vim!
You also get a nice IDE called Xcode. I don't do much programming so I can't say how nice it really is, but it does the usual project management, debugging, etc that you would expect.
*twitch*
Actually implicit int return type is no longer legal in C99 so a good compiler should at least warn about it by now imho.
Ask and ye shall receive:
Microsoft Debugging Tools for Windows
That toolkit is part of the (also free) Platform SDK, but you can install it separately. It includes NTSD (command-line debugger) and WinDBG (GUI debugger), and KD (the kernel debugger). NTSD and WinDBG sit on top of the same user-mode debugging engine, "dbgeng.dll". They both include really fantastic on-line help which can teach you a lot about debugging in Windows. That said, they are not for the faint-of-heart (the Visual Studio debugger is much more user-friendly). KD.exe uses the kernel-mode debugger built into every NT kernel by default. Of course, you need a second machine to use KD and a serial cable; when broken into the NT kernel debugger, the debuggee is not in control.
(Incidentally, is there a kernel-mode debugger available for Linux? Last time I checked, Linus opposed the concept very strongly, and Linux did not have one available. He called it a "crutch." Sorry, Linus. Kernel-mode device driver writers *like* their debuggers. I have to say that this could be one of several factors impeding device driver development on Linux.)
There is no IDE, just command-line tools. The "build manager" referred to by the grandparent is msbuild.exe, a command-line tool that speaks a (N)Ant-like XML syntax.
As I understand it, VS.NET 2003 will use the same syntax, so you should be able to build a VS.NET project using the free SDK.
With MS's latest C++ compiler (and with newer Os's), you need "\r\n" to make a new line. This drove me nuts when I first started doing it. Now, it's not so bad.
That is weird, I do not have this problem. The I/O layer should translate newlines for you. All I ever use in Windows programs (console and GUI) is '\n' and it ouputs "\r\n" as well as reversing it on input. This is in both MinGW and Visual C++
24 beers in a case, 24 hours in a day. Coincidence? I think not!
Erm:
.o file)."
.file "foo.c"
.text
.p2align 4,,15
.type foo, @function
.size foo, .-foo
.section .note.GNU-stack,"",@progbits
.ident "GCC: (GNU) 3.3.3 (Debian 20040401)"
"-ffast-math on gcc only has it do what both Intel and MS are already doing - using the bloody fp unit instead of software emulation!!! (easy proof - do an asm dump of the generated
bash-2.05b$ more foo.c
double foo(double a, double b, double c, double d)
{
return (a+b)*(b+c)*(c+d)*(d+a);
}
bash-2.05b$ gcc -S -O3 -fno-fast-math foo.c
bash-2.05b$ more foo.s
.globl foo
foo:
pushl %ebp
movl %esp, %ebp
fldl 8(%ebp)
fldl 16(%ebp)
fldl 24(%ebp)
fld %st(2)
fadd %st(2), %st
fldl 32(%ebp)
fxch %st(3)
fadd %st(2), %st
fxch %st(2)
popl %ebp
fadd %st(3), %st
fxch %st(1)
fmulp %st, %st(2)
fxch %st(2)
faddp %st, %st(3)
fmulp %st, %st(1)
fmulp %st, %st(1)
ret
If that's not using the FPU unit, then I'm a Dutchman.
FP. (Not a Dutchman)
Also FatPhil on SoylentNews, id 863
OPTIONS THAT WERE IN VC6 BUT AREN'T IN VC7 cl.exe:NEW VC7 cl.exe OPTIONS:So as you see most of the differences are new switches for the VC.7 version of cl.exe. Those shouldn't cause any trouble, as VC6 just won't use them by default. The two options removed I don't think are very interesting to most poeple. Besides, the VC.7 cl.exe is happy to ignore them:So as another poster has said, it looks like it should just work if you put the new directories first in the MSVC6 bin/lib/include paths.
Cool! All the updates I care about for VC 6 without having to switch to a new IDE!
I wonder if it is possible to use the newer commandline compiler from within Visual Studio 6. From what I hear, the .NET IDE isn't so great for C/C++ work. And come on, a project file is now called a "solution" file?? That's just dumb.
I'm currently writing C++ code using Visual Studio.NET, and I haven't noticed any problems compared to Visual C++ 6. Some of the keyboard shortcut defaults are different, but you can revert to the VC++ 6 layout easily.
I think the main reason it might not seem as good for C++ work is that VS.NET adds considerably more features geared towards .NET development, while C++ appears to be neglected somewhat. Remember that VS.NET's IDE supports all their .NET languages, and C++.
For clarification, a solution can contain multiple projects. The old term they used was workspace. Yeah, I thought it was weird at first, but it's just terminology, plus a workspace sort of implies open windows and developer-specific settings to me. A single solution can be used by multiple developers (CVS works nicely).
And if you'd ever used their Alpha compiler you'd know how far behind the x86 compiler it was. Even internal MS developers had a lot of problems with it. The Itanium could be better supported perhaps, as they probably got some help from Intel and generally have their act together better now.
Heck, the CD I have is out of date and I have to go to the developer's tools page http://developer.apple.com/ to download the newest version anyway. Why bother giving me a CD? Just a pamphlet (nice and bio-degradeable and recycled paper if you could please, since I'm throwing it away anyway) giving me the URL?
Or a link from the OS? OH wait! There is one! Apple Menu, Mac OS X Software, Clink on "Developer" in the web site's menu.
Why were we talking about distributing CDs again? Developers will most likely be downloading the latest version of the tools anyway. (Have you not updated gcc? Visual C++?) Make the tools available, and they will code.
I am, and always will be, an idiot. Karma: Coma (mostly effected by
IIRC, Electric Fence detects heap buffer overflows (presumably using guard pages). MS has AppVerifier, which allows you to turn that on for arbitrary programs at runtime. It also has a bunch of other settings you can tweak, like causing the OS to report a different version. It's a great tool for testing obscure boundary cases.
The security checks metioned here are at code-gen time. They do things like insert cookies on the stack to detect stack buffer overflows.
Ok, I gave it a try, the new free compiler with VC6 IDE. It seems to compile and run ok. But debugging doesn't work.
/pdbtype flag. It still ignored it and linked ok, but when I went to debug, VC6 said the exe didn't have any debugging symbols. I've tried changing the debugging options in VC6 a few different ways (C7 compatible, no program database etc), but nothing I've tried makes the debugger happy.
The new link.exe complained that it doesn't know the
So it looks like this may not be something that will work as well as one might hope, unless anyone has any other ideas.
Loop vectorization does not have to do with hyperthreading. Loop vectorization takes code like this:
for(...iterate over buffer...)
for(...iterate over vector...)
do_scaler_op()
And turn it into code like this:
for(...iterate over buffer...)
do_vector_op(...on vector...)
This has nothing to do with hyperthreading. Hyperthreading deals with parallelism between threads, not parallelism within threads. Intel has a seperate facility for the compiler to generate seperate program threads (OpenMP), but the auto-vectorizer does not. The auto-vectorizer works at a much lower level, replacing several scaler ops with a smaller number of vector ops. GCC does not have the analysis technology necessary to do that.
A deep unwavering belief is a sure sign you're missing something...
So, no more need to install precompiled and tested binaries, you can now port the exact same code (where did you plan on getting it?) by hand yourself and find a whole new set of bugs. Yeah, brilliant. Oh, you say you weren't going to use the GNU code? Great, yet another weird variant of coreutils.
Installing Cygwin is hardly a chore. I click setup.exe, tell it what to get, and walk away.
If you don't like the restrictions that come with the POSIX-layer DLL, then install MinGW instead. Same tools, same source code, same GNU extensions that we're already used to, no cygwin1.dll.
You're either smoking crack or you haven't actually tried installing Cygwin in the last three years. I regularly run xterms under XP's own window manager, integrated clipboards, latest builds of just about everything. If you claim *nix stuff has been "missing," then you clearly haven't tried looking.
You cannot apply a technological solution to a sociological problem. (Edwards' Law)
The header file part was eaten by /. since it looks like an html tag. You need to use the html entities for the greater and less than signs.
grey wolf
LET FORTRAN DIE!
You guys are paranoid illiterates. It says you can't give out their code to their redistributables because you don't have the right to do that. I.e. the source code to sprintf if you linked in libcmt.lib or something of that nature.
Redistributables = Microsoft's binaries normally allowed to be given to end-users by YOU directly as part of your application. (C++ runtime DLL's, C runtime DLLs, MFC DLL's etc)
Did you read section 2.2 where it legally defined redistributable?
Assuming that "Redistributables" is defined to be the downloadable software development kit (as it is in licenses for similar MicroSoft SDKs) then I think you should not be worried.
My reading would be that the GPL is indeed "Excluded" according to the clause above. (Again, the term/clause is not new in MS SDK licenses.) But this just means you can't e.g. redistribute the SDK itself under the GPL.
The object files produced by a compiler are not considered to be derivative works of the compiler. So, you can distribute binaries compiled with this SDK under any license you desire. The analogy is that you can safely use Word to edit a document without Microsoft having a copyright claim on the resulting document.
I'm fairly sure this behavior is required by the C/C++ standard. The key is to open streams in ASCII mode, rather than binary mode. Because C/C++ are designed to be portable, the standard I/O library is required to convert the line endings (but not in binary mode). Unix's binary mode is equal to its ASCII mode (since it uses \n natively).
:) The Internet mainly uses \r\n as the line ending (although this made sense in the past, why they continue to stick with it in new protocols is beyond me, besides maybe the ability to easily telnet in and type in raw protocol commands by hand).
If the standard I/O library didn't understand line endings, then functions like fgets() could never work, and working with strings (where you expect to be able to match against \n) would be much harder.
Networking is different, since it uses raw I/O, and doesn't pass through stdio's translation layer. Also, you usually want the binary data on the network.
Here's an excruciatingly detailed explaination.
Note that fseek() with SEEK_CUR will almost always fail in text mode, since the number of bytes read may vary from the current offset in the file due to the conversion. Recording the current offset with ftell() followed by an fseek() with SEEK_SET will clear that right up.
Crazy but true. This has corrupted my data more than once. Here's an excruciatingly detailed explaination.