[FFmpeg-cvslog] r17667 - trunk/configure

[Reimar Döffinger]

On Sat, Feb 28, 2009 at 08:56:19PM +0100, mru wrote:
> Author: mru
> Date: Sat Feb 28 20:56:19 2009
> New Revision: 17667
> 
> Log:
> configure: {enable,disable}_safe functions that sanitise arguments
> 
> Modified:
>    trunk/configure
> 
> Modified: trunk/configure
> ==============================================================================
> --- trunk/configure	Sat Feb 28 20:34:50 2009	(r17666)
> +++ trunk/configure	Sat Feb 28 20:56:19 2009	(r17667)
> @@ -325,6 +325,14 @@ disable_weak(){
>      set_weak no $*
>  }
>  
> +enable_safe(){
> +    enable $(echo "$*" | sed 's/[^A-Za-z0-9_ ]/_/g')
> +}
> +
> +disable_safe(){
> +    disable $(echo "$*" | sed 's/[^A-Za-z0-9_ ]/_/g')
> +}
> +
>  do_enable_deep(){
>      for var; do
>          enabled $var && continue
> @@ -566,9 +574,7 @@ check_header(){
>      log check_header "$@"
>      header=$1
>      shift
> -    var=`echo $header | sed 's/[^A-Za-z0-9_]/_/g'`
> -    disable $var
> -    check_cpp "$@" <<EOF && enable $var
> +    check_cpp "$@" <<EOF && enable_safe $header || disable_safe $header
>  #include <$header>
>  int x;
>  EOF

I suspect that this actually broke check_headers, if the check fails, disable_safe
will be run, which runs echo and the return value of check_headers is
thus the return value of that echo command (instead of the actual
check).
At least that's what happens on MinGW, though it might be a bug (I don't
really think so though)...