Re: Code reviews [Was: FWTK and smap/smapd]

[Darren Reed <darrenr () reed wattle id au>]

In some email I received from Joseph S D Yao, sie wrote:
> Oh, geez, shouldn't everyone have written [something like] this?

No...

Your parameters are wrong, for a start :-)
They were "char *buf" and "unsigned int buflen".

> register char *cp = buf;
> register int i, count = n;
>
> for (;;) {
>       i = write(fd, cp, count);
>       if (i < 0) {
>               perror("...");
>               fprintf(stderr, ...);
>               break;
>       }
>       count -= i;
>       if (count <= 0) /* yes, should never be <, optionally fuss */
>               break;
>       cp += i;
>       if (i == 0 {
>               sleep(1);
>               /* should probably count and break if too many */
>       }
> }

You're using a "for (;;)" where it isn't required - should be
"while (count != 0)", the "if (count <= 0) ..." should be nuked from orbit.

Hanlding write() returning 0 bytes, is, umm, bizarre ?  Carson never
said anything about non-blocking I/O but apparently it was implied.
That's a failure in the specification, I say :-)

Carson wanted specific handling of things like EINTR...

_______________________________________________
firewall-wizards mailing list
firewall-wizards () honor icsalabs com
http://honor.icsalabs.com/mailman/listinfo/firewall-wizards