problem with select( )

[huskers]

When i am compiling on linux box with the following code:

long n;
n=select((1L<< 16) + 8L*BITS(int),&fds,NULL,NULL,NULL);

I am getting error: parse error before &quot;int&quot;....am i doing something wrong...the same code works fine on AIX.

thanks.

 

[Salem]

> (1L<< 16) + 8L*BITS(int),
1. (int) looks like it should be a cast, but casts prefix the expression, not postfix them
So perhaps
(1L<< 16) + 8L*(int)BITS,

2. The first parameter is supposed to be the number of fds to check, but this is one massive number - (1<<16) is like over 65000.
Does your 'fds' structure have this many entries?

--

http://www.catb.org/~esr/faqs/smart-questions.html

 

[azimuth0]

To me, BITS looks like a macro should probably be defined as:

#define BITS(type)    (sizeof(type) * 8)

However, that doesn't look like something appropriate to put in the first value of a select() call. You're supposed to put the maximum fd in any of the FD_SETs you pass to it, plus one.

-

http://www.sanctuary.org/~azimuth/

 

[gasingh]

From the code snippette, it seems Azimuth0 is right in the definition od BITS.

I think the problem is in the first argument only.
(1L<< 16) + 8L*BITS(int)

This is a huge number. I dont think that any OS allows so many open FDs. HP-Unix allows 2K. And this number is more than 64K.

Maybe you can post some more related code for finding details about the error.