Hi, the issue is not so much about grepping through the content of a "real" directory. As the directory content is binary encoded, that is almost never useful.
I don't mind having skip behaviour on systems like GNU/Linux where directories always really are directories (if that statement is even true). In the Hurd, a directory might be a conventional directory. However, it might also be a strange non-standard thingie. That thingie could allow directory operations on it like a conventional directory. However it could also allow file operations on it like a conventional file. For example, an XML file could be accessed simply by "emacs foo.xml", but it could at the same time implement a directory hierarchy for "ls foo.xml" that represents the structure of the XML content. I think that on the GNU/Hurd, the read behaviour is more favorable. It allows for scenarios like the above, without adding a lot of noise to greps output in case you have a match with a directory content. On systems where opening a directory always fails, the error output of grep tends to annoy me, though. On such systems I would prefer skip to be the default behaviour. How to differentiate these systems I don't know, maybe a "mkdir foo; cat foo"? Or just using host_os Thanks, Marcus -- `Rhubarb is no Egyptian god.' GNU http://www.gnu.org [EMAIL PROTECTED] Marcus Brinkmann The Hurd http://www.gnu.org/software/hurd/ [EMAIL PROTECTED] http://www.marcus-brinkmann.de/ _______________________________________________ Help-hurd mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/help-hurd