UNIX for Dummies Questions & Answers

Hi all

I tried this on an old version of sed on NCR Unix MP-RAS:

Code:

sed -e "s/$/nnn/" file1 >file2

This file (file1):

Code:

the cat sat on the mat.
the cat sat on the mat.

the cat sat on the mat.

becomes this (file2):

Code:

the cat sat on the mat.nnn
the cat sat on the mat.nnn
nnn
the cat sat on the mat.nnn

Now, I tried to port this sed script to GNU sed v.4.1.5 on RHEL5. Using the same script:

Code:

sed -e "s/$/nnn/" file1 >file2

my output looks like this:

Code:

nnn cat sat on the mat.
nnn cat sat on the mat.
nnn
nnn cat sat on the mat.

I tried awk instead of sed...

Code:

awk '{print $0 "nnn"} file1 >file2

but get the same results from both platforms (MP-RAS appends to end of line, RHEL overtypes first 3 chars of each line).


Any ideas why? Any ideas how to fix - I am running out of ideas.

TIA for any insight...

BRgds
JG

Can it be you transferred the input file from a windows platform and forgot to convert to unix file format?

Hi Scrutinizer

It would appear not.

While it is true that the data has been copied from Windows, `cat'ing and `vi'ing the file shows up no ctrl-M characters as line ends.

Thanks for the input though, much appreciated.

BRgds
JG

---------- Post updated at 12:06 PM ---------- Previous update was at 11:59 AM ----------

Scrutinizer

Oops - my bad!

You were right. Even though VI / CAT showed no ctrl-m chars (they DO show up under MP-RAS when it occurs), the file was in fact encoded as DOS (x'0d0a' line ends). a quick dos2unix fixed it for me!

Thanks - I thought I was going nuts.

BRgds
JG

It would explain the nnn at the beginning of the line, which can happen because of spurious carriage return characters. What happens if you create a new test file by hand with vi and then run your sed statement? Same thing? gsed should have no trouble with your script.

-- Our posts crossed --