Prev: Passover Seder plates !!!
Next: Firefox P.O.S.
From: Nuno J. Silva on 11 Mar 2010 15:02
Torsten Mueller <dev-null(a)shared-files.de> writes:

> J G Miller <miller(a)yoyo.ORG> schrieb:
>
>> > Thanks! (You never end learning sed ...)

Thanks! Yet another sed feature I didn't know about. I think sometimes I
end up replacing \n to something else (using tr) to include \n on the
expression. This will surely make some of those command chains simpler.

>>
>> What annoys me most though, is people who resort to awk to change
>> things on one line. awk is for manipulating columns of data, not
>> just changing or selecting a single field or two on one line ;)
>
> Never used awk seriously. But indeed over the years I surprised a lot
> of people (especially from the Windows world ...) by demonstrating the
> features of sed upon a 1GB logfile. This ends up always in unbelieving
> eyes and a mutual whisper of "witchcraft" and so on. I really wonder
> that sed is still almost unknown even in the Unix world. Younger
> people have usually never heard of this.

I recall reading an instruction list to rename a project (on windows),
and the name had to be changed on a file. Of course, the instructions,
written for windows user, and (probably) by a windows user, asked people
to open the file in an editor and change some lines.

Fortunately sed was available.

--
Nuno J. Silva
gopher://sdf-eu.org/1/users/njsg
From: Chris F.A. Johnson on 11 Mar 2010 20:28
On 2010-03-11, Chris Davies wrote:
> J G Miller <miller(a)yoyo.org> wrote:
>> What annoys me most though, is people who resort to awk
>> to change things on one line. awk is for manipulating columns
>> of data, not just changing or selecting a single field or two on
>> one line ;)
>
> IMO awk instructions are often easier to read when processing data
> separated by whitespace. For example, I can write this:
>
> virsh list |
> awk '$3 == "running" {print $2}
>
> or this:
> virsh list |
> sed -n "/^[^{tab} ]*[{tab} ]\([^{tab} ]*\)[{tab} ]running.*/\1/p'
>
> or an equivalent in Perl, C, etc.
>
> One is definitely clearer (to me) than the other. On the other hand,
> the solution showing how sed can process a line /after/ a pattern match
> is definitely clearer than what I'd imagine writing in awk.

awk '

/pattern/ && (n == 0) { ++n; print; next }
(n == 1) { #do whatever
}
{ print }
' "$file"

--
Chris F.A. Johnson, <http://cfajohnson.com>
Author:
Pro Bash Programming: Scripting the GNU/Linux Shell (2009, Apress)
Shell Scripting Recipes: A Problem-Solution Approach (2005, Apress)
From: Chris Davies on 12 Mar 2010 06:21
J G Miller <miller(a)yoyo.org> wrote:
> virsh list is generating columns of data? yes?
> That is exactly the good use for which awk was designed.

Perhaps virsh was a bad example, but I prefer to choose something that
folks "out there" could potentially try.


> It is using awk on a single line of data that I was complaining about.

"Use" as in "search & edit", or "use" as in "search and print"? I
think the difference is best described by the two tools' names - one
is designed as a line-based pattern editor, the other is a more of
a scanning/reporting tool. Both can be used for parts of either job,
but not necessarily as easily.


> The point is awk is best used for extracting bits of information from
> columns of data eg telephone numbers from /etc/passwd, rather than
> sed which is best used to edit a field in situ.

Mostly agreed.


> As always, it depends on what you are trying to achieve and the
> type of data will determine which is the best tool to use, and
> the tool will be different for different types of jobs.

Definitely agreed. And for those with only hammers, everything's a
nail. Fortunately we have a plethora of tools from which to choose an
appropriate one :-)

Chris
First  |  Prev  | 
Pages: 1 2 3
Prev: Passover Seder plates !!!
Next: Firefox P.O.S.