It's interesting that out of the three ways I posted to do it, the author chose the 'sed' method which would have been my last choice.
Someday I've got to sit down and really learn Perl regular expressions. It just won't stick in my mind, kind of like all that English grammar I couldn't get to stick in my mind back in school. It seems to me that regular expressions ranks right up there with gerunds and past participles. Ouch! :)
Yes, you are correct. I did not mean to delete *empty* lines and *should* have written:
# sed -e '/^__*$/d' filename
...THAT IS, *two* underscore characters preceeding the asterisk so that the match is *one* underscore anchored at the line's start, followed by *zero or more* underscore's anchored to a newline.
The extended regular expression:
^_+$
understood by 'perl', works nicely too with 'egrep' ['grep -E'] or with 'awk' but not with 'sed'.
James, I know :] GNU sed however /does/ recognize the + metacharacter. One more reason to learn perl? One more reason to replace all default commands with there superior GNU equivalent (tar springs to mind)
FWIW The regex bible from Jeffrey Friedl (http://www.oreilly.com/catalog/regex/) has nice comparison tables for the most common filter programs in what regex flavour they accept.
I know, and I'm proud to be among the first few to find the errors in the book. Please note the errata area on the web when reading the book. Some are very misleading, and the stuff dealt with is complicated enough as it is already.
