const question

[James Brown]

This may seem a little odd but I have a few simple questions. Basically, I'm
writing a really simple parser which understands C type declarations - the
only reason being, so that I understand C better. So here goes:

/* does this imply 'const int' ?? */
const c = 0;

/* is this legal C? it's allowed by my 'real' C compiler but I'm not sure */
const const const int c = 0;

/* same for this??? */
unsigned unsigned unsigned int u = 0;

thanks,
James

[Walter Roberson]

In article <Xd********************@pipex.net>,
James Brown <no*@home.netwrote:

>/* is this legal C? it's allowed by my 'real' C compiler but I'm not sure */
const const const int c = 0;

C89 3.5.3 Type Qualifiers

The same type qualifier shall not appear more than once in
the same specifier list or qualifer list, either directly or
via one or more typedefs.
--
Programming is what happens while you're busy making other plans.

[James Brown]

"Walter Roberson" <ro******@ibd.nrc-cnrc.gc.cawrote in message
news:ei**********@canopus.cc.umanitoba.ca...

In article <Xd********************@pipex.net>,
James Brown <no*@home.netwrote:

>>/* is this legal C? it's allowed by my 'real' C compiler but I'm not sure
*/
const const const int c = 0;

C89 3.5.3 Type Qualifiers

The same type qualifier shall not appear more than once in
the same specifier list or qualifer list, either directly or
via one or more typedefs.
--

ok thanks for clearing that one up....so if I'm understanding the above
paragraph correctly, the following is therefore invalid:

typedef const int cint;

const cint c; /* invalid as per C89 3.5.3 */

thanks,
James

[Jordan Abel]

2006-11-01 <Xd********************@pipex.net>,
James Brown wrote:

This may seem a little odd but I have a few simple questions. Basically, I'm
writing a really simple parser which understands C type declarations - the
only reason being, so that I understand C better. So here goes:

/* does this imply 'const int' ?? */
const c = 0;

This one wasn't answered. I believe the answer is yes for c89, no for
c99. 'int' is rarely strictly necessary for c89, actually, in that you
can usually write your program without it (but why would you want to?)

[Ian Collins]

Jordan Abel wrote:

2006-11-01 <Xd********************@pipex.net>,
James Brown wrote:

>>This may seem a little odd but I have a few simple questions. Basically, I'm
writing a really simple parser which understands C type declarations - the
only reason being, so that I understand C better. So here goes:

/* does this imply 'const int' ?? */
const c = 0;

This one wasn't answered. I believe the answer is yes for c89, no for
c99. 'int' is rarely strictly necessary for c89, actually, in that you
can usually write your program without it (but why would you want to?)

s/int/const/ ?

--
Ian Collins.

[Keith Thompson]

Ian Collins <ia******@hotmail.comwrites:

Jordan Abel wrote:

>2006-11-01 <Xd********************@pipex.net>,
James Brown wrote:

>>>This may seem a little odd but I have a few simple
questions. Basically, I'm writing a really simple parser which
understands C type declarations - the only reason being, so that I
understand C better. So here goes:

/* does this imply 'const int' ?? */
const c = 0;

This one wasn't answered. I believe the answer is yes for c89, no for
c99. 'int' is rarely strictly necessary for c89, actually, in that you
can usually write your program without it (but why would you want to?)

s/int/const/ ?

No, I think Jordan's point is that C89 supports "implicit int", but
C99 doesn't.

I *think* it's true that any legal C89/C90 program can be re-written
as an equivalent program that doesn't use the "int" keyword, but I
don't care enough to verify it.

--
Keith Thompson (The_Other_Keith) ks***@mib.org <http://www.ghoti.net/~kst>
San Diego Supercomputer Center <* <http://users.sdsc.edu/~kst>
We must do something. This is something. Therefore, we must do this.

[Ian Collins]

Keith Thompson wrote:

Ian Collins <ia******@hotmail.comwrites:

>>Jordan Abel wrote:

>>>2006-11-01 <Xd********************@pipex.net>,
James Brown wrote:
This may seem a little odd but I have a few simple
questions. Basically, I'm writing a really simple parser which
understands C type declarations - the only reason being, so that I
understand C better. So here goes:

/* does this imply 'const int' ?? */
const c = 0;
This one wasn't answered. I believe the answer is yes for c89, no for
c99. 'int' is rarely strictly necessary for c89, actually, in that you
can usually write your program without it (but why would you want to?)

s/int/const/ ?

No, I think Jordan's point is that C89 supports "implicit int", but
C99 doesn't.

I *think* it's true that any legal C89/C90 program can be re-written
as an equivalent program that doesn't use the "int" keyword, but I
don't care enough to verify it.

Good point, it just wouldn't occur to me to use "implicit int".

--
Ian Collins.

[Frederick Gotham]

James Brown:

>C89 3.5.3 Type Qualifiers

The same type qualifier shall not appear more than once in
the same specifier list or qualifer list, either directly or
via one or more typedefs.

That's bullshit about the typedef's! Why is the Standard so?

(It's fine in C++.)

--

Frederick Gotham

[Walter Roberson]

In article <nv*******************@news.indigo.ie>,
Frederick Gotham <fg*******@SPAM.comwrote:

>James Brown {included material that I quoted from the C89 standard}

>>C89 3.5.3 Type Qualifiers

>> The same type qualifier shall not appear more than once in
the same specifier list or qualifer list, either directly or
via one or more typedefs.

>That's bullshit about the typedef's! Why is the Standard so?

If your typedef is so transparent that you don't already know the
attribute you would qualify with, then when you qualify it
you run the risk of logical contradiction. If you already know
enough about the typedef to know there will be no logical contradiction
then you already know whether the attribute would be redundant or not.

--
All is vanity. -- Ecclesiastes

[Frederick Gotham]

Walter Roberson:

>>That's bullshit about the typedef's! Why is the Standard so?

If your typedef is so transparent that you don't already know the
attribute you would qualify with, then when you qualify it
you run the risk of logical contradiction. If you already know
enough about the typedef to know there will be no logical contradiction
then you already know whether the attribute would be redundant or not.

Because C doesn't have templates, there would be no need to allow multiple
consts via typedef's.

--

Frederick Gotham

[Jordan Abel]

2006-11-01 <ln************@nuthaus.mib.org>,
Keith Thompson wrote:

Ian Collins <ia******@hotmail.comwrites:

>Jordan Abel wrote:

>>2006-11-01 <Xd********************@pipex.net>,
James Brown wrote:

This may seem a little odd but I have a few simple
questions. Basically, I'm writing a really simple parser which
understands C type declarations - the only reason being, so that I
understand C better. So here goes:

/* does this imply 'const int' ?? */
const c = 0;
This one wasn't answered. I believe the answer is yes for c89, no for
c99. 'int' is rarely strictly necessary for c89, actually, in that you
can usually write your program without it (but why would you want to?)

s/int/const/ ?

No, I think Jordan's point is that C89 supports "implicit int", but
C99 doesn't.

I *think* it's true that any legal C89/C90 program can be re-written
as an equivalent program that doesn't use the "int" keyword, but I
don't care enough to verify it.

There are cases where you can't do it if you want prototypes. And cases
where you need prototypes to be equivalent. Other than that, I think
you're probably right. If you don't want binary-compatibility... well,
is it possible anymore to declare a variadic function without
prototypes?

eh - well, int wasn't necessary in K&R as far as I can see.

[Keith Thompson]

Jordan Abel <ra****@random.yi.orgwrites:

2006-11-01 <ln************@nuthaus.mib.org>,
Keith Thompson wrote:

[...]

>I *think* it's true that any legal C89/C90 program can be re-written
as an equivalent program that doesn't use the "int" keyword, but I
don't care enough to verify it.

There are cases where you can't do it if you want prototypes. And cases
where you need prototypes to be equivalent. Other than that, I think
you're probably right. If you don't want binary-compatibility... well,
is it possible anymore to declare a variadic function without
prototypes?

eh - well, int wasn't necessary in K&R as far as I can see.

I don't think the "int" keyword is necessary even for prototypes.
Even in C99, the type usually referred to as "int" can be called
"signed".

--
Keith Thompson (The_Other_Keith) ks***@mib.org <http://www.ghoti.net/~kst>
San Diego Supercomputer Center <* <http://users.sdsc.edu/~kst>
We must do something. This is something. Therefore, we must do this.

[Nick Keighley]

Frederick Gotham wrote:

James Brown:

C89 3.5.3 Type Qualifiers

The same type qualifier shall not appear more than once in
the same specifier list or qualifer list, either directly or
via one or more typedefs.

That's bullshit about the typedef's! Why is the Standard so?

(It's fine in C++.)

can't you make your point in a more civil fashion?
--
Nick Keighley

[Clever Monkey]

Frederick Gotham wrote:

Walter Roberson:

>>That's bullshit about the typedef's! Why is the Standard so?

If your typedef is so transparent that you don't already know the
attribute you would qualify with, then when you qualify it
you run the risk of logical contradiction. If you already know
enough about the typedef to know there will be no logical contradiction
then you already know whether the attribute would be redundant or not.

Because C doesn't have templates, there would be no need to allow multiple
consts via typedef's.

Oh gods, please no. If templating ever comes to C, please oh please let
it be an unrequired extension of some sort.