Re: vfprintf, scanf portability

[Bruno Haible]

Paul Eggert wrote:
> >      Be aware that @code{vfprintf} is not always available.

Already 5 years ago, all gnulib target platforms had vfprintf:

$ ./show-portability vfprintf
libc                aix-4.3.2
libc                aix-5.1.0
libc                aix-5.2.0
libc                aix-5.3.0a
libc                aix-6.1.0
libc                aix-7.1.0
libc                beos
libc                cygwin-1.5
libcygwin           cygwin-1.5
libc                cygwin-1.7
libcygwin           cygwin-1.7
libc                freebsd-5.2.1
libc                freebsd-6.0
libc                freebsd-6.4
libc                glibc-2.3.6
libc                hpux-10.20
libc                hpux-11.00
libc                hpux-11.11
libc                hpux-11.23
libc                hpux-11.31
libc                interix-3.5
libc                irix-5.3
libc                irix-6.5
libc                macosx-10.3
libc                macosx-10.4
libc                macosx-10.5
libcrtdll           mingw
libmsvcr70          mingw
libmsvcr71          mingw
libmsvcrt           mingw
libc                netbsd-3.0
libc                netbsd-5.0
libc                netbsd-5.0.1
libc                openbsd-3.8
libc                osf1-4.0d
libc                osf1-5.1a
libc                pips
libc                solaris-2.10
libc                solaris-2.11_2010_11
libc                solaris-2.4
libc                solaris-2.5.1
libc                solaris-2.6
libc                solaris-2.7
libc                solaris-2.8
libc                solaris-2.9

> That's a bit dated, but not entirely irrelevant.  See, for example,
> 
> http://buildlogs.pld-linux.org/index.php?dist=th&arch=athlon&ok=1&name=gcc&id=78974776-f08b-4b31-a977-863793c73c22&action=text

This log is inconsistent: In line 872 it shows

  checking for strrchr... checking for vfprintf... yes
  yes

and then in line 893:

  checking for vfprintf... no

> http://buildd.emdebian.org/~zumbi/toolchain/squeeze/i386/logs/i386-sparc-gcc-4.4.log

Likewise inconsistent: In line 3099

  checking for vfprintf... no

and in line 3346

  checking whether gcc supports -Wwrite-strings... checking for vfprintf... yes

I guess it was a user error in both cases, such as running several configures
in parallel in the same directory, or botched configure.ac files.

In summary, vfprintf exists everywhere. But like fprintf and printf, it has a
number of portability problems.

Bruno
-- 
In memoriam Diego de Enzinas <http://en.wikipedia.org/wiki/Diego_de_Enzinas>