Subject: Re: Query about return value of snprintf
To: None <,>
From: Michael Richardson <>
List: tech-userlevel
Date: 06/17/1999 18:05:57

>>>>> "Chris" == Chris G Demetriou <> writes:
    Chris> writes:
    >> The NetBSD man page says that we do:
    >> These functions return the number of characters printed (not including
    >> the trailing `\0' used to end output to strings).
    >> Which is what I expected.

    Chris> No, that's part of what it says.

    Chris> Further it says:

    Chris> snprintf() and vsnprintf() will write at most size-1 of the
    Chris> characters printed into the output string (the size'th character
    Chris> then gets the ter- minating `\0'); if the return value is greater
    Chris> than or equal to the size argument, the string was too short and
    Chris> some of the printed characters were discarded.

  Okay, I missed that part. I don't mind or object to it, I just didn't see
it. I just prefer that things are consistent.

    Chris> I consider the BSD behaviour much preferred to what you quote as
    Chris> the GNU libc behaviour.

  Me too. It is what I expect. It is also more useful since it tells me how
much space I *should* allocate, as you indicate. I am counting on this
behaviour actually, which is how I noticed this problem.

    Chris> If the caller wants to know that the output was truncated, they
    Chris> can do a simple comparison in either case.  However with the BSD
    Chris> return value, they can then go an allocate a right-sized buffer if
    Chris> their application needs it, and reformat the complete string.
    Chris> From what you've quoted, it doesn't sound like there's any good
    Chris> way to figure out the 'right size' of a right-sized buffer with
    Chris> the glibc implementation.

    >> I realize that this isn't POSIX, but are there any opinions as what is
    >> reasonable? I see reasons why the GNU version is useful, but I also
    >> see that it may break existing code.

    Chris> Actually, see above; i'd say that GNU version is strictly less
    Chris> useful than the BSD version, as it returns less information.

  As my compat snprintf() is a BSD4.3 version that I ported to StunOS years ago,
I'm just going to enable using it on Linux for now, but I thought I'd ask
some questions first.

   :!mcr!:            |  Solidum Systems Corporation,
   Michael Richardson |For a better connected world,where data flows faster<tm>
 Personal: PGP key available.
 Corporate: <A HREF=""></A>. 

Version: PGPfreeware 5.0i for non-commercial use
Comment: Processed by Mailcrypt 3.5.1, an Emacs/PGP interface
Charset: noconv