Following the report of Henrik Bengtsson which was traced by Duncan
Murdoch to a bug in the regexp code from glibc-2.3.3 in R, I have updated
the code in R-devel to -2.3.5. This had that bug fixed and about 700
other changes, most of which seem to be bug fixes.
My tests on Linux (including of all CRAN packages) have detected no
differences (but then I was not able to reproduce Henrik's problem on
Linux, only on Windows). If anyone finds a difference that they think is
not a bug fix please let me know.
There are also quite a few new optimizations, especially in UTF-8 locales.
It seems somewhat clear that the codes available for multi-byte character
set regexps are not yet mature. You might want to remember that if you
use a GNU/Linux system which is not bang up-to-date (many are older than
even glinc-2.3.3).
--
Brian D. Ripley, ripley at stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595
