grep: option to use GNU regex matching instead of POSIX one.
 This fixes problems with NULs in files being scanned, but
 costs +800 bytes. The same can be done to sed (TODO).

diff --git a/Config.in b/Config.in
index 5ad35ce..c2005c7 100644
--- a/Config.in
+++ b/Config.in
@@ -21,6 +21,15 @@
 	  Select this only if you plan to use busybox on full-blown
 	  desktop machine with common Linux distro, not on an embedded box.
 
+config EXTRA_COMPAT
+	bool "Provide compatible behavior for rare corner cases (bigger code)"
+	default n
+	help
+	  This option makes grep, sed etc handle rare corner cases
+	  (embedded NUL bytes and such). This makes code bigger and uses
+	  some GNU extensions in libc. You probably only need this option
+	  if you plan to run busybox on desktop.
+
 config FEATURE_ASSUME_UNICODE
 	bool "Assume that 1:1 char/glyph correspondence is not true"
 	default n