httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jim Jagielski <...@devsys.jaguNET.com>
Subject Re: [PATCH] Error logging in http_core.c: extra info, fseek()
Date Sun, 07 Mar 1999 22:34:34 GMT
Ben Laurie wrote:
> 
> Jim Jagielski wrote:
> > 
> > John Bley wrote:
> > >
> > > On Sun, 7 Mar 1999, Roy T. Fielding wrote:
> > >
> > > > >+              if (fseek(f, offset, SEEK_SET))
> > > > I'm all for adding more error detection stuff, but the comparison should
be
> > > >
> > > >    if (fseek(f, offset, SEEK_SET) == -1)
> > > >
> > > > since that is more reliable across platforms than a return of only zero
> > > > on success.  The same is true of other system calls (look at the man page
> > > > to see what is defined as "RETURN VALUES").
> > >
> > > Hmmm.. you learn something new everyday.  So "-1 for error" is more
> > > portable than "non-zero for error"?  OK, I'll make sure to follow that
> > > pattern.  Should I then assume that existing code which makes this
> > > same mistake should be corrected?
> > >
> > 
> > Nope... in fact, just the opposite. non-0 is, at least for fseek(), more
> > portable. Some systems (FreeBSD for example) specifically state -1
> > is an error, but others (as well as Ansi-K&R) just state non-zero.
> 
> You have to be kidding! fseek returns the current offset. -1 is an
> error. Non-zero is success for most cases.
> 

Nope. If fseek() is successful, it returns 0. I think you're thinking
of ftell().

-- 
===========================================================================
   Jim Jagielski   |||   jim@jaguNET.com   |||   http://www.jaguNET.com/
            "That's no ordinary rabbit... that's the most foul,
            cruel and bad-tempered rodent you ever laid eyes on"

Mime
View raw message