По сравнению с 4.1.3, вышедшей в январе (никто о её выходе не постил):
2005-01-25 Paolo Bonzini <bonzini@gnu.org>
* sed/regexp.c: Fix building on GCC 2.95 and earlier.
По сравнению с 4.1.2, вышедшей в августе:
2004-12-26 Paolo Bonzini <bonzini@gnu.org>
Do not use leftmost-longest matching for addresses.
* NEWS: Add a note about this.
* testsuite/recall2.good, testsuite/recall2.inp,
testsuite/recall2.sed: New test.
* testsuite/Makefile.am, testsuite/Makefile.tests: Add the
recall2 test.
* sed/sed.h (struct regex): New.
(struct addr, struct subst, compile_regex, match_regex,
release_regex): Use it instead of regex_t.
* sed/compile.c (compile_program): Update for new meaning of
the third parameter of compile_regex.
* sed/execute.c (do_subst): Pass less conservative value to
the regsize parameter of match_regex.
* sed/regexp.c (compile_regex_1): New, extracted out of
compile_regex. The third parameter, needed_sub, now includes
\0 (so 10 means that \0 .. \9 are needed). Pass RE_NO_SUB
if needed_sub is zero.
(compile_regex): Accept a struct regex instead of a regex_t.
Save the regular expression's text.
(match_regex): Accept a struct regex instead of a regex_t.
Recompile the pattern if it was compiled with RE_NO_SUB.
(release_regex): Accept a struct regex instead of a regex_t.
* doc/Makefile.am: Generate sed.texi correctly when
building outside srcdir.
2004-12-26 Paolo Bonzini <bonzini@gnu.org>
* BUGS: Add section about [a-z] matching uppercase characters,
and other locale issues.
* doc/sed-in.texi [!PERL]: Likewise.
2004-11-15 Paolo Bonzini <bonzini@gnu.org>
* sed/execute.c (str_append_modified): Copy the first character
when using \l or \u in a multi-byte configuration. Use
WCRTOMB instead of wcrtomb.
* sed/sed.h (WCRTOMB): New.
2004-11-11 Paolo Bonzini <bonzini@gnu.org>
* tst-rxspecer.c: Do not mix instructions and
declarations.
* basicdefs.h: Include locale.h before #defining
gettext, to avoid breakage under Solaris.
* sed/sed.c: do not include locale.h.
интересно, почему только в одном текстовом
редакторе кроме собственного механизма поиска-
замены есть возможность выполнить sed-программу
на выбранном участке текста?
emacs и vim в расчет не беру -- там все можно сделать,
но остальные-то что