switch expression not integral ???

[bayxarea-usenet@yahoo.com]

I am getting the following error during compiling:

'switch expression not integral'

I am new to the switch command - here is a snip of my code where I have
used this.. .. what is the problem here? Thanks!

------------- snip --------------------------

for (ii = 0; ii < n; ii++)
{
if (!UF_OBJ_ask_na me(members[ii], obj_name))
{
sprintf(msg,"\n Object #%d has an object tag = %d and is named
%s,",ii,member s[ii],obj_name);
WRITE(msg);

UF_CALL(UF_DRF_ ask_draft_aid_t ext_info(member s[ii], &nn,
&text_info)) ;

for (iii=0; iii<nn; iii++)
{
WRITE3F(text_in fo[iii].origin);
for (jj=0; jj<text_info[iii].num_lines; jj++)
{
sprintf(msg, " %s\n", text_info[iii].text[jj].string);
UF_UI_write_lis ting_window( msg );
}
}
UF_CALL(UF_DRF_ free_text(nn, &text_info)) ;
switch (obj_name)
{
case "NOTE TEXT CP SL":
uc5566(members[ii], 1, 6, "J");
break;
case "NOTE TEXT CP COC":
uc5566(members[ii], 1, 6, "O");
break;
case "NOTE TEXT CP PQP":
uc5566(members[ii], 1, 6, "H");
break;
case "NOTE TEXT CP FRD":
uc5566(members[ii], 1, 6, "N);
break;
case "NOTE TEXT CP SC":
uc5566(members[ii], 1, 6, "P);
break;
default:
break;
}

}
}

[Ulrich Eckhardt]

Re: switch expression not integral ???

bayxarea-usenet@yahoo.co m wrote:[color=blue]
> I am getting the following error during compiling:
>
> 'switch expression not integral'[/color]

The argument to switch must be integral. It can't be a floating point
value, a structure or a pointer (including char pointers aka strings).
[color=blue]
> I am new to the switch command - here is a snip of my code where I have
> used this.. .. what is the problem here? Thanks![/color]

Please reduce your examples to the smallest piece of code that demonstrates
your problem.

Uli

[Michael Mair]

Re: switch expression not integral ???

bayxarea-usenet@yahoo.co m wrote:[color=blue]
> I am getting the following error during compiling:
>
> 'switch expression not integral'
>
> I am new to the switch command - here is a snip of my code where I have
> used this.. .. what is the problem here? Thanks![/color]

Well, switch can only be applied to integer data types, i.e
char, (un)signed char, (un)signed short, (un)signed int,
(un)signed long (, size_t, ..., C99: (un)signed long long,
the int*_t types, ptrdiff_t, ...) but not to a pointer
-- at least I guess that obj_name is either of type char *
or of type array of char.

[color=blue]
> ------------- snip --------------------------
>
> for (ii = 0; ii < n; ii++)
> {
> if (!UF_OBJ_ask_na me(members[ii], obj_name))
> {
> sprintf(msg,"\n Object #%d has an object tag = %d and is named
> %s,",ii,member s[ii],obj_name);
> WRITE(msg);
>
> UF_CALL(UF_DRF_ ask_draft_aid_t ext_info(member s[ii], &nn,
> &text_info)) ;
>
> for (iii=0; iii<nn; iii++)
> {
> WRITE3F(text_in fo[iii].origin);
> for (jj=0; jj<text_info[iii].num_lines; jj++)
> {
> sprintf(msg, " %s\n", text_info[iii].text[jj].string);
> UF_UI_write_lis ting_window( msg );
> }
> }
> UF_CALL(UF_DRF_ free_text(nn, &text_info)) ;
> switch (obj_name)
> {
> case "NOTE TEXT CP SL":
> uc5566(members[ii], 1, 6, "J");
> break;
> case "NOTE TEXT CP COC":
> uc5566(members[ii], 1, 6, "O");
> break;
> case "NOTE TEXT CP PQP":
> uc5566(members[ii], 1, 6, "H");
> break;
> case "NOTE TEXT CP FRD":
> uc5566(members[ii], 1, 6, "N);
> break;
> case "NOTE TEXT CP SC":
> uc5566(members[ii], 1, 6, "P);
> break;
> default:
> break;
> }
>
> }
> }[/color]

A workaround: Apply switch to the first significant
character...
switch (obj_name[13]) {
case 'S':
/* Either SC or SL */
break;
case 'C':
....
}

Two things for future requests:
* Please give us a minimal example which we can directly
feed to a compiler.
* Indent your code. Not with tabs but with spaces.


Cheers
Michael
--
E-Mail: Mine is an /at/ gmx /dot/ de address.

[Andrey Tarasevich]

Re: switch expression not integral ???

bayxarea-usenet@yahoo.co m wrote:[color=blue]
> I am getting the following error during compiling:
>
> 'switch expression not integral'
>
> I am new to the switch command - here is a snip of my code where I have
> used this.. .. what is the problem here? Thanks![/color]

The problem here is just what the compiler said: 'switch expression not
integral'. In C language switch argument as well as case labels must be
integral expressions. You seem to be trying to use pointer expressions
('char*') instead. This will not work. If you want to build a multiway
dispatcher based on string argument, one way to go is to use 'if' ladder

if (strcmp(obj_nam e, "NOTE TEXT CP SL") == 0)
uc5566(members[ii], 1, 6, "J");
else if (strcmp(obj_nam e, "NOTE TEXT CP COC") == 0)
uc5566(members[ii], 1, 6, "O");
else if (strcmp(obj_nam e, "NOTE TEXT CP PQP") == 0)
uc5566(members[ii], 1, 6, "H");
/* ... and so on */

Switch is not applicable here. At least, not immediately applicable.

--
Best regards,
Andrey Tarasevich

[bayxarea-usenet@yahoo.com]

Re: switch expression not integral ???

Thanks for all the replies - I did not realize it was only a test for
integers - didn't realize that integral meant integer actually.

I will post a more usable snippet in the future...

Thanks again..

John

[Mike Wahler]

Re: switch expression not integral ???


<bayxarea-usenet@yahoo.co m> wrote in message
news:1102974969 .522711.34080@f 14g2000cwb.goog legroups.com...[color=blue]
> Thanks for all the replies - I did not realize it was only a test for
> integers[/color]

Actually, the type of the 'case' expression is even more restricted
than that. The expression must be a constant integer expression.

int main()
{
int x = 42;
int y = 99;

switch(x)
{
case y: /* ERROR: not constant expression */
break;
case 99: /* OK: constant expression */
break;
default:
break;
}

return 0;
}
[color=blue]
>- didn't realize that integral meant integer actually.[/color]

Now you know. :-)

-Mike

[Sniper1]

Re: switch expression not integral ???

In article <1102974969.522 711.34080@f14g2 000cwb.googlegr oups.com>,
bayxarea-usenet@yahoo.co m says...[color=blue]
> Thanks for all the replies - I did not realize it was only a test for
> integers - didn't realize that integral meant integer actually.[/color]

Where you thinking it meant the area under the switch curve? :-)