|
From: Jeremy F. <je...@go...> - 2004-01-05 16:59:46
|
On Mon, 2004-01-05 at 03:37, Doug Rabson wrote: > I think the key problem was the use of '\1' but I didn't realise that > until I started messing around with -E (which selects an alternative > regexp expression syntax). It should be possible to come up with > something that works on both versions of sed. Well, both versions of sed support a more complete regex syntax than normal sed, but they seem to do it incompatibly. GNU sed uses a lot more \'s, and BSD sed uses -E. We could just weaken the regexps until they're representable in standard sed regex syntax, but that seems like giving up. J |