BOTH RIGHT!
>SB?
> sigset_t waitset;
Yes but I didn't discover this until I resolved the next question.
>Of possible use as a sample recipe for verifying include definitions:
>- - -
>#include <stdio.h>
>#include <sys/select.h>
>
>#define show(X) printf("sizeof '%s' is %d bytes\n", #X, sizeof(X))
>
>int main(int ac, char* av[])
>{
> show(sigset_t);
> return 0;
>}
>- - -
sigset_t isn't defined in signal.h, it's in select.h
Actually it isn't in there either, but select includes a handful of
others...I said I wasn't too good at this.
Now, back to "struct sigset_t". I shot myself in the foot, by googling for
"struct sigset_t". Is there a word for that? Look up "One small step for a
man" and "Mark Twain" and you will likely "discover" that he said it. In my
case I came up with a scrap of code from an IBM "public code" web site.
Obviously, in whatever universe they're in, sigset_t is a struct, just not
in our world. It's just the way API stuff is presented in the book, where
you manage pieces of data using system calls and you never get to know about
the contents. I was just expecting a struct.
Anyway, my resident guru said that signals are so nonstandard that I
shouldn't expect everything to be defined. He also went off to find a
better way. I think I am about to be told to just go read the clock...
>Also: you mentioned that this was an embedded app -- maybe the standard
>linux/posix stuff is not present?
The definition of embedded, especially where Linux is concerned, varies
considerably. In this case, it's a 2.2 regular kernel, and is built on a
Red Hat 6.2 system.
Thanks
Barry
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-lpsg
