Daichi Kawahata wrote:
> In my system SIGKILL is defined as 9, 2 is SIGINT,

I didn't really doubt that and my comment wasn't really directed towards
you but rather the author of this check. The problem here is also (or
especially) that the Configure check doesn't do, what the actual program
does. I'm pretty sure that gtk-gnutella doesn't use signal() with numerical
values - without even verifying this myself.
Consider SIGSTOP == 2. I didn't check it, but I guess that would Configure
forever.


> and by replacing
> signal() it seems Configure passes without error. Should the script
> is modified like that (but I don't know signal issues of other system)?

> -       signal(2, foo);
> -       kill(getpid(), 2);
> +       sigaction()(2, foo);

That's the certainly a creative way to use sigaction(). ;-) However,
I meant this check could be removed from Configure completely. gtk-gnutella
should use sigaction() instead because sigaction() isn't ambiguous like
signal().

-- 
Christian

This mail is 2xROT13 encrypted. You've just violated the YMCA.

Attachment: pgpQCLfHOSPj8.pgp
Description: PGP signature

Reply via email to