J
Jim Showalter
Does a hosted implementation require any particular libraries in order to be
considered a standard C compiler?
considered a standard C compiler?
J
jacob navia
Jim said:
I suppose that the standard C library (at least) should be present.
In any case a compiler without any library is not very useful...
jacob
R
Richard Heathfield
Jim Showalter said:
A hosted implementation must provide all the standard library functions, but
the Standard does not say in what form they must be provided.
A hosted implementation must provide all the standard library functions, but
the Standard does not say in what form they must be provided.
P
pete
Jim said:
Yes.
C89, hosted:
<assert.h> <locale.h> <stddef.h>
<ctype.h> <math.h> <stdio.h>
<errno.h> <setjmp.h> <stdlib.h>
<float.h> <signal.h> <string.h>
<limits.h> <stdarg.h> <time.h>
C89, freestanding:
<float.h> <limits.h> <stdarg.h>
<stddef.h>
C99, hosted:
<assert.h> <complex.h> <ctype.h>
<errno.h> <fenv.h> <float.h>
<inttypes.h> <iso646.h> <limits.h>
<locale.h> <math.h> <setjmp.h>
<signal.h> <stdarg.h> <stdbool.h>
<stddef.h> <stdint.h> <stdio.h>
<stdlib.h> <string.h> <tgmath.h>
<time.h> <wchar.h> <wctype.h>
C99, freestanding
<float.h> <iso646.h> <limits.h>
<stdarg.h> <stdbool.h> <stddef.h>
<stdint.h>
P
pete
Richard said:
Yes.
This directive in a C file: #include <float.h>
only means that the compiler has to compile the C file
as though float.h was pasted in there.
If that can be done without a float.h file being present,
then there doesn't need to be a float.h file
in the hosted implementation.
J
Jim Showalter
jacob said:
That's what I thought, too. But I've installed the latest versions of MinGW
along with gcc-core, and cannot find even a mention of libc or glibc.
K
Keith Thompson
Jim Showalter said:
The standard says nothing about "libc" or "glibc". As long as a
program that includes standard headers and calls standard library
functions works as specified in the standard, it doesn't matter how
the implementation does it.
<OT>gcc typically uses whatever runtime library is provided by the
underlying operating system. I don't know what MinGW does.</OT>
If you're curious about the details of how your implementation
provides the standard library headers and fucntions, either check your
implementation's documentation or ask in a system-specific newsgroup.
R
Richard Heathfield
pete said:
[etc]
Those are not libraries. They're headers.
<snip>
[etc]
Those are not libraries. They're headers.
<snip>
R
Richard Heathfield
pete said:
What has <float.h> to do with libraries?
What has <float.h> to do with libraries?
M
Malcolm
Yes. the standard library.Jim Showalter said:
The problem is that most hosted implementations run on computers with
graphical user interfaces, and there are no standard library functions for
communicating with a GUI. Some evil systems also take away streams such as
stderr and stdout, which makes life very difficult.
P
pete
Richard said:
Parts of the standard library,
are sometimes refered to as "libraries",
such as "the math library",
which is mentioned in the FAQ.
http://c-faq.com/fp/libm.html
J
jacob navia
Jim said:
mingw doesn't have an own version of the standard C
Library and uses the implementation provided with the
operating system: crtdll.dll.
This rather ancient implementation will provide you with
C89 semantics with a compiler that can possible emit
C99 code, so take care as not to ask for anything
recent what standards is concerned.
At the same time, even if the compiler accepts and emits
code for long doubles, mingw doesn't have any library support
for long doubles so better not use them.
The same with long longs, since the printf version of crtdll.dll
doesn't understand any long long semantics. Maybe it understands
long long ll;
printf("%I64d\n",ll);
but I am not sure about it.
Another possibility is that you make mingw link with the
C library of the microsoft compiler system, but I do
not know how to do that or if it is fair to do this (use
Microsoft's libraries without buying a copy of the
Microsoft compiler); See your lawyer for additional
information
jacob
R
Richard Heathfield
pete said:
If you read it more carefully, you'll see that the question addresses the
/difference/ between headers and libraries. When it's talking about "the
math library", it is talking about "actually linking with the math
*library*", not "linking with the <math.h> header". You don't link with
headers.
If you read it more carefully, you'll see that the question addresses the
/difference/ between headers and libraries. When it's talking about "the
math library", it is talking about "actually linking with the math
*library*", not "linking with the <math.h> header". You don't link with
headers.
T
those who know me have no need of my name
in comp.lang.c i read:
[re: platform provided standard c library]
eh? decoder ring needed.
that would be sad then, since long double has been part of the language for
a long time [c89 3.1.2.5].
[re: platform provided standard c library]
eh? decoder ring needed.
that would be sad then, since long double has been part of the language for
a long time [c89 3.1.2.5].
J
jacob navia
This means that the printf provided by crtdll.dll is strictlythose said:
C89 and doesn't provide any way of printing either
long doubles, long longs or any normal printf directives.
OK?
Yes, but microsoft implemented long double as a
synonym for double. Since gcc has real long doubles
the library CRTDLL.DLL does NOT follow and you are stuck
R
Richard Heathfield
jacob navia said:
That's a contradiction. If it's strictly C89, it will provide a way to print
long doubles.
Only in some of their implementations. I have a couple of Microsoft
compilers with 80-bit long doubles, compared to just 64 bits for double.
That's a contradiction. If it's strictly C89, it will provide a way to print
long doubles.
Only in some of their implementations. I have a couple of Microsoft
compilers with 80-bit long doubles, compared to just 64 bits for double.
J
jacob navia
Richard Heathfield a écrit :
#include <stdio.h>
int main(int argc, char *argv[])
{
long double d = 1e600L;
printf("%Le\n",d);
return 0;
}
This will FAIL under mingw.
I posted you this message in another thread but you continue
to post nonsense.
Yes, very ancient implementations have real long doubles. Modern ones
have not.
Why confuse people? It is your hobby or what?
#include <stdio.h>
int main(int argc, char *argv[])
{
long double d = 1e600L;
printf("%Le\n",d);
return 0;
}
This will FAIL under mingw.
I posted you this message in another thread but you continue
to post nonsense.
Yes, very ancient implementations have real long doubles. Modern ones
have not.
Why confuse people? It is your hobby or what?
R
Richard Heathfield
jacob navia said:
You claimed:
(a) mingw uses the C runtime library provided in crtdll.dll
(b) the printf provided by that library is strictly C89
(c) the printf provided by that library doesn't provide a way to print long
doubles
It's possible that my C89 draft is wrong, I suppose, but anyway it says:
an optional L specifying that a following e , E , f , g , or G
conversion specifier applies to a long double argument.
Was that removed from the Standard at the last moment?
If so, then all that you have said makes sense. But if not, then I don't
understand what point you are making, as you appear to be contradicting
yourself.
I am sorry if you find the facts confusing.
You claimed:
(a) mingw uses the C runtime library provided in crtdll.dll
(b) the printf provided by that library is strictly C89
(c) the printf provided by that library doesn't provide a way to print long
doubles
It's possible that my C89 draft is wrong, I suppose, but anyway it says:
an optional L specifying that a following e , E , f , g , or G
conversion specifier applies to a long double argument.
Was that removed from the Standard at the last moment?
If so, then all that you have said makes sense. But if not, then I don't
understand what point you are making, as you appear to be contradicting
yourself.
I am sorry if you find the facts confusing.
M
Michael Mair
jacob said:
Yes, but the problem is that mingw plus MS libs is _not_ a
conforming implementation.
long doubles, doubles, and floats can have exactly the same size
and representation.
The above code relies on implementation limits rather than
guaranteed standard limits (LDBL_MAX is guaranteed to be at least
1E+37).
The obvious fix is to not use mingw in this combination as it is
broken (or keep in mind that it is broken w.r.t. long double I/O).
Cheers
Michael
J
jacob navia
Michael said:
This is exactly what I wanted to convey. Same problems appear with
long long.