freebsd/gnu/usr.bin/grep/NEWS

Version 2.4.1

  - If the final byte of an input file is not a newline, grep now silently
    supplies one.

  - The new option --binary-files=TYPE makes grep assume that a binary input
    file is of type TYPE.
    --binary-files='binary' (the default) outputs a 1-line summary of matches.
    --binary-files='without-match' assumes binary files do not match.
    --binary-files='text' treats binary files as text
	(equivalent to the -a or --text option).

  - New option -I; equivalent to --binary-files='without-match'.

Version 2.4:

  - egrep is now equivalent to `grep -E' as required by POSIX,
    removing a longstanding source of confusion and incompatibility.
    `grep' is now more forgiving about stray `{'s, for backward
    compatibility with traditional egrep.

  - The lower bound of an interval is not optional.
    You must use an explicit zero, e.g. `x{0,10}' instead of `x{,10}'.
    (The old documentation incorrectly claimed that it was optional.)

  - The --revert-match option has been renamed to --invert-match.

  - The --fixed-regexp option has been renamed to --fixed-string.

  - New option -H or --with-filename.

  - New option --mmap.  By default, GNU grep now uses read instead of mmap.
    This is faster on some hosts, and is safer on all.

  - The new option -z or --null-data causes `grep' to treat a zero byte
    (the ASCII NUL character) as a line terminator in input data, and
    to treat newlines as ordinary data.

  - The new option -Z or --null causes `grep' to output a zero byte
    instead of the normal separator after a file name.

  - These two options can be used with commands like `find -print0',
    `perl -0', `sort -z', and `xargs -0' to process arbitrary file names,
    even those that contain newlines.

  - The environment variable GREP_OPTIONS specifies default options;
    e.g. GREP_OPTIONS='--directories=skip' reestablishes grep 2.1's
    behavior of silently skipping directories.

  - You can specify a matcher multiple times without error, e.g.
    `grep -E -E' or `fgrep -F'.  It is still an error to specify
    conflicting matchers.

  - -u and -U are now allowed on non-DOS hosts, and have no effect.

  - Modifications of the tests scripts to go around the "Broken Pipe"
    errors from bash. See Bash FAQ.

  - New option -r or --recursive or --directories=recurse.
    (This option was also in grep 2.3, but wasn't announced here.)

  - --without-included-regex disable, was causing bogus reports .i.e
    doing more harm then good.

Version 2.3:

  - When searching a binary file FOO, grep now just reports
    `Binary file FOO matches' instead of outputting binary data.
    This is typically more useful than the old behavior,
    and it is also more consistent with other utilities like `diff'.
    A file is considered to be binary if it contains a NUL (i.e. zero) byte.

    The new -a or --text option causes `grep' to assume that all
    input is text.  (This option has the same meaning as with `diff'.)
    Use it if you want binary data in your output.

  - `grep' now searches directories just like ordinary files; it no longer
    silently skips directories.  This is the traditional behavior of
    Unix text utilities (in particular, of traditional `grep').
    Hence `grep PATTERN DIRECTORY' should report
    `grep: DIRECTORY: Is a directory' on hosts where the operating system
    does not permit programs to read directories directly, and
    `grep: DIRECTORY: Binary file matches' (or nothing) otherwise.

    The new -d ACTION or --directories=ACTION option affects directory handling.
    `-d skip' causes `grep' to silently skip directories, as in grep 2.1;
    `-d read' (the default) causes `grep' to read directories if possible,
    as in earlier versions of grep.

  - The MS-DOS and Microsoft Windows ports now behave identically to the
    GNU and Unix ports with respect to binary files and directories.

Version 2.2:

Bug fix release.

  - Status error number fix.
  - Skipping directories removed.
  - Many typos fix.
  - -f /dev/null fix(not to consider as an empty pattern).
  - Checks for wctype/wchar.
  - -E was using the wrong matcher fix.
  - bug in regex char class fix
  - Fixes for DJGPP

Version 2.1:

This is a bug fix release(see Changelog) i.e. no new features.

  - More compliance to GNU standard.
  - Long options.
  - Internationalisation.
  - Use automake/autoconf.
  - Directory hierarchy change.
  - Sigvec with -e on Linux corrected.
  - Sigvec with -f on Linux corrected.
  - Sigvec with the mmap() corrected.
  - Bug in kwset corrected.
  - -q, -L and -l stop on first match.
  - New and improve regex.[ch] from Ulrich Drepper.
  - New and improve dfa.[ch] from Arnold Robbins.
  - Prototypes for over zealous C compiler.
  - Not scanning a file, if it's a directory
    (cause problems on Sun).
  - Ported to MS-DOS/MS-Windows with DJGPP tools.

See Changelog for the full story and proper credits.

Version 2.0:

The most important user visible change is that egrep and fgrep have
disappeared as separate programs into the single grep program mandated
by POSIX 1003.2.  New options -G, -E, and -F have been added,
selecting grep, egrep, and fgrep behavior respectively.  For
compatibility with historical practice, hard links named egrep and
fgrep are also provided.  See the manual page for details.

In addition, the regular expression facilities described in Posix
draft 11.2 are now supported, except for internationalization features
related to locale-dependent collating sequence information.

There is a new option, -L, which is like -l except it lists
files which don't contain matches.  The reason this option was
added is because '-l -v' doesn't do what you expect.

Performance has been improved; the amount of improvement is platform
dependent, but (for example) grep 2.0 typically runs at least 30% faster
than grep 1.6 on a DECstation using the MIPS compiler.  Where possible,
grep now uses mmap() for file input; on a Sun 4 running SunOS 4.1 this
may cut system time by as much as half, for a total reduction in running
time by nearly 50%.  On machines that don't use mmap(), the buffering
code has been rewritten to choose more favorable alignments and buffer
sizes for read().

Portability has been substantially cleaned up, and an automatic
configure script is now provided.

The internals have changed in ways too numerous to mention.
People brave enough to reuse the DFA matcher in other programs
will now have their bravery amply "rewarded", for the interface
to that file has been completely changed.  Some changes were
necessary to track the evolution of the regex package, and since
I was changing it anyway I decided to do a general cleanup.