Re: socket api question - is it possible for recv() to receive0byte from server?

Vernon Hamberg wrote:

As you say, functions that set errno do not change it if they end
successfully. That is why it is good practice in C to set it to 0
before any call to such a function. ...

I used to think this was the right way to use errno until a recent post
by Scott Klement (maybe here, maybe on another forum) convinced me
otherwise.  I may even have been a (the?) source of the misleading
information about errno in the Redbook; I don't recall.

For the couple of cases like strtod() where errno is the only way to
determine if an error occurred, and if so, those cases can be handled
differently.  (And from reading Simon's post, even strtod() has a
non-errno way of checking for error.)

But aside from right or wrong, I think Scott's way is just plain nicer. 
It doesn't interrupt the normal flow with the assignment to errno, and
the checking for an error is directly related to the call that is being
checked.

   parm = something;
   errno = 0;                    // this line gets in the way
   retval = somefunction(parm);
   if errno <> 0;                // check seems unrelated to the call
     --- handle the error
   endif;

vs

   parm = something;
   retval = somefunction(parm);
   if retval indicates failure; 
     --- handle the error, possibly referring to errno
   endif;

A programming rule of thumb: "If it looks nicer, it's better."  (This
rule of thumb, more than most, should be taken with a huge grain of
salt.  Still, I find it useful, along with a related rule "If it looks
icky, it's got bugs.")