What is branch in sed

[Dan Track]

On 8/23/06, Paul Howarth <paul at city-fan.org> wrote:
> Dan Track wrote:
> > On 8/23/06, Cameron Simpson <cs at zip.com.au> wrote:
> >> On 23Aug2006 11:52, Dan Track <dan.track at gmail.com> wrote:
> >> | I've searched and read documents on branching in sed (b) but I still
> >> | can't get my head round it.
> >> |
> >> | Could someone please explain to me how branching works, an example
> >> | would be nice.
> >>
> >> It's like goto. Here's an infinite loop:
> >>
> >>         :foo
> >>         b foo
> >>
> >> Cheers,
> >> --
> >
> > Hi,
> >
> > Thanks for the info. With that in mind I have the following example
> > I'm having trouble unerstanding. Would you be able to help me out with
> > understanding it? What does the $b in context to the whole file and
> > executing the lines of code on the file. Also what does $!N mean? And
> > do I really need P;D?
> >
> >
> > sed -e '$b
> > /^Target\[[^]]*\][.0-9]*:.*@[0-9.]*$/ {
> > $!N
> > s/^\(Target\[[^]]*\][.0-9]*:.*@\)[0-9.]*\nDirectory\[[^]]*\]:
> > \(.*\)$/\1\2.example.com/
> > P;D
> > }' text_files/mrtg.cfg > text_files/mrtg.cfg.changed
>
> The sed script quoted is heavily based on an example in the "sed FAQ"
> (http://sed.sourceforge.net/sedfaq4.html#s4.23.2); reading that document
>   is quite useful for people learning sed I think.
>
> Paul.
>

Hi Paul,

Thanks for the info. As a small query I'm kind of understanding what b
means, but what does "$b" mean and how does it affect the script. I
thought what preceded b was the matching clause which dictates when a
jump should be performed. So how does the "$" come into play?

Thanks
Dan