Paul Eggert <[email protected]> writes:
> * lib/filename.h: Add warning that IS_RELATIVE_FILE_NAME ("") is 1.
> ---
> lib/filename.h | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/lib/filename.h b/lib/filename.h
> index bc401b17a8..e37a373eb4 100644
> --- a/lib/filename.h
> +++ b/lib/filename.h
> @@ -49,7 +49,10 @@ extern "C" {
> any notion of "current directory".
> IS_RELATIVE_FILE_NAME(Filename)
> tests whether Filename may be concatenated
> - to a directory filename.
> + to a directory filename to yield the name
> + of a file relative to that directory.
> + Watch out: IS_RELATIVE_FILE_NAME ("") is 1
> + even though "" is not a relative file name.
> Note: On native Windows, OS/2, DOS, "c:" is neither an absolute nor a
> relative file name!
> IS_FILE_NAME_WITH_DIR(Filename) tests whether Filename contains a device
Just curious, did you run into a bug in your recent GNU tar changes
because of this behavior?
Collin
P.S. I'm suprised you wrote "1" instead of "true". I think you are the
only person I have seen use booleans in C macro definitions. :)
$ grep -F USE_LONGLONG_H src/factor.c
#ifndef USE_LONGLONG_H
# define USE_LONGLONG_H true
# define USE_LONGLONG_H false
#if USE_LONGLONG_H
