I am a little confused. Am using DMC 8.44 (beta) which in which literal strings are taken as constant. Except when they aren't...

Given the following

void foo (char *);

main ()

{ char *p = "hello";
int f= 1;
foo("hello");
foo(p);
foo (f ? "a" : "b");
foo (f ? p :  "b");
}


foo (f ? "a" : "b");
^
test.cpp(9) : Error: need explicit cast for function parameter 1 to get
from: char const *
to  : char *
foo (f ? p :  "b");
^
test.cpp(10) : Error: need explicit cast for function parameter 1 to get
from: char const *
to  : char *
--- errorlevel 1

It seems to me that the first of the function calls should be an error. Except - I guess - that this would break so much existing code as to be useless. But then if we are to allow that, then why not the tertiaries?

Cheers,
John.

"John Jameson" <John_member@pathlink.com> wrote in message news:dc2joe$1l01$1@digitaldaemon.com...
> I am a little confused. Am using DMC 8.44 (beta) which in which literal
strings
> are taken as constant. Except when they aren't...
>
> Given the following
>
> void foo (char *);
>
> main ()
>
> { char *p = "hello";
> int f= 1;
> foo("hello");
> foo(p);
> foo (f ? "a" : "b");
> foo (f ? p :  "b");
> }
>
>
> foo (f ? "a" : "b");
> ^
> test.cpp(9) : Error: need explicit cast for function parameter 1 to get
> from: char const *
> to  : char *
> foo (f ? p :  "b");
> ^
> test.cpp(10) : Error: need explicit cast for function parameter 1 to get
> from: char const *
> to  : char *
> --- errorlevel 1
>
> It seems to me that the first of the function calls should be an error.
Except -
> I guess - that this would break so much existing code as to be useless.
But then
> if we are to allow that, then why not the tertiaries?

I thought about that case when implementing it, but could find no language in the C++ Standard which supported the implicit conversion for the ?: expressions - it's only for "string literals". Perhaps my interpretation is wrong, but I thought I'd try it this way.

Well, I guess I should be grateful that the mighty standards committee at least decided to allow an exception for simple char *arguments :-)

Anyway, am pleased to report that we have now ported all the core sections of Proteus from SC 7.5 to DM 8.44 beta with no sign of new bugs or problems so far.

thanks once again for putting the -Ai switch in.

Cheers,
john


In article <dcet0r$2srr$1@digitaldaemon.com>, Walter says...
>
>
>"John Jameson" <John_member@pathlink.com> wrote in message news:dc2joe$1l01$1@digitaldaemon.com...
>> I am a little confused. Am using DMC 8.44 (beta) which in which literal
>strings
>> are taken as constant. Except when they aren't...
>>
>> Given the following
>>
>> void foo (char *);
>>
>> main ()
>>
>> { char *p = "hello";
>> int f= 1;
>> foo("hello");
>> foo(p);
>> foo (f ? "a" : "b");
>> foo (f ? p :  "b");
>> }
>>
>>
>> foo (f ? "a" : "b");
>> ^
>> test.cpp(9) : Error: need explicit cast for function parameter 1 to get
>> from: char const *
>> to  : char *
>> foo (f ? p :  "b");
>> ^
>> test.cpp(10) : Error: need explicit cast for function parameter 1 to get
>> from: char const *
>> to  : char *
>> --- errorlevel 1
>>
>> It seems to me that the first of the function calls should be an error.
>Except -
>> I guess - that this would break so much existing code as to be useless.
>But then
>> if we are to allow that, then why not the tertiaries?
>
>I thought about that case when implementing it, but could find no language in the C++ Standard which supported the implicit conversion for the ?: expressions - it's only for "string literals". Perhaps my interpretation is wrong, but I thought I'd try it this way.
>
>