Findutils 4.2.9 is released on alpha.gnu.org

[James Youngman]

I'm happy to announce the availability of findutils version 4.2.9 at
ftp://alpha.gnu.org/gnu/findutils. 

The main focus of the findutils 4.2.9 release is better POSIX
compliance.  There are a number of changes to xargs introducing the
POSIX choice of option letter for some options that GNU xargs already
supported, (-I and -E for example), and some small changes to the
behaviour of find in some mildly unusual error conditions (symbolic
link loops, non-searchable starting points which are pruned).

The other significant change is that "_" is no longer used by default
as the logical end-of-input marker in xargs.  This is a SYSV behaviour
which is not mandatory for POSIX compliance, and I feel that the
(previous) default behaviour had too much chance of silently producing
an unexpected result.  The old behaviour can be restored by the use of
a command-line option.


The NEWS file for this release says:-

* Major changes in release 4.2.9

**  Functionality Changes
*** xargs no longer treats a line containing only an underscore as a
    logical end-of-file.  To obtain the behaviour that was previously
    the default, use "-E_".
*** xargs now supports the POSIX options -E, -I and -L.   These are synonyms 
    for the existing options -e, -i and -l, but the latter three are
    now deprecated.

**  Bug Fixes
*** xargs -n NUM now invokes a command as soon as it has NUM arguments.  
    Previously, it waited until NUM items had been read, and then
    invoked the command with NUM arguments, saving the remaining one
    for next time. 
*** If "find -L" discovers a symbolic link loop, an error message is issued.
*** If you specify a directory on the find command line, but -prune 
    is applied to it, find will no longer chdir() into it anyway.

**  Documentation improvements
*** The precise interpretation of the arguments to the -atime, -ctime 
    and similar tests in find has been documented more clearly.


The next thing that findutils needs is a better test suite.  This is
because I plan to implement the "-exec ... +" and "-execdir" options,
which will involve refactoring xargs to make use of some of its code
in find.  Since we don't want to break xargs while doing this, a more
comprehensive test suite would be a good idea.  If you are willing to
lend a hand in this area, please let me know.

Regards,
James Youngman