Re: -a is not -e ?! wrong entry in "man bash" ?

[Chet Ramey]

> Machine Type: i686-pc-linux-gnu
> 
> Bash Version: 2.05b
> Patch Level: 0
> Release Status: release
> 
> Description:
> 
> We have found a very strange entry in the man-page:
> line 1748:
> ~       -a file
> ~              True if file exists.

I should never have made this a unary primary, ksh compatibility aside.

> line 1756:
> ~       -e file
> ~              True if file exists.
> 
> - -a and -e should do the same according to the man-page, but they dont!

Yes, they do.  Really.

> if [ -a /bin/bash ]  is TRUE if /bin/bash exists.
> if [ ! -a /bin/bash ]  should be FALSE if /bin/bash exists, BUT it
> returns TRUE.
> 
> if [ -e /bin/bash ] is TRUE if /bin/bash exists.
> if [ ! -e /bin/bash ] is FLASE if /bin/bash exists.
> 
> So -a is not -e as the man-page says. So what does -a really do ? Is
> this a bash bug or a man-page "bug" ?

Neither.  This question is answered, for the most part, by item E1 in
the Bash FAQ.  The short answer is:

test/[ works off the number of arguments
binary primaries have higher precedence than ! for the three-argument case
-a and -o are considered binary primaries for the three-argument case

A more complete answer is in the FAQ.

> Fix:
> ~ Never use -a :)

If you mean as a unary primary, I would not disagree.

-- 
``The lyf so short, the craft so long to lerne.'' - Chaucer
( ``Discere est Dolere'' -- chet )

Chet Ramey, ITS, CWRU    chet@po.CWRU.Edu    http://cnswww.cns.cwru.edu/~chet/