I just had an email exchange with Phil Hazel, author of PCRE and pcregrep, asking him what GNU grep now lacks that pcregrep offers, with a view to making pcregrep redundant. He replied:
A quick glance at "man grep" (2.5.4 on a Gentoo Linux system) shows up the following: 1. It says grep -P is "highly experimental". 2. GNU grep lacks --include-dir. 3. GNU grep with -P doesn't have a means of setting certain PCRE-specific options such as -M, --match-limit, --recursion-limit. -M (multiline matching) is a rather different way of matching the text of a file. 4. There's no mention of UTF-8 support in the GNU grep man page. 5. There are some trivialities such as pcregrep's synonym of --word-regex for --word-regexp (but of course those are minor issues). 6. There are surely other issues that I missed in a quick compare. In fact, I see that I have already documented some differences in the pcregrep man page (including those mentioned above). It says this: However, the --file-offsets, --include-dir, --line-offsets, --locale, --match-limit, -M, --multiline, -N, --newline, --recur- sion-limit, -u, and --utf-8 options are specific to pcregrep, as is the use of the --only-matching option with a capturing parentheses number. Would it be worth adding this to the GNU grep TODO list? It would be nice to make pcregrep redundant, as it's essentially a GNU grep clone that uses PCRE, unlike some other grep implementations that have more major differences. -- http://rrt.sc3d.org
