Re: paste for cygwin is broken when using MS style newlines.

[Bob Proulx]

>       I think paste can not handle the MS carridge returns... Take a look
> at the last four lines.(I tried the same files on solaris and got the same
> problem, but when I tried recreating them using solaris's echo the problems
> went away!)

Thank you for your report.  However, the problem is not in paste.  The
carriage returns are being printed and are being processed by your
terminal.

> E:\backup\tmp\old>echo 1 > out2
> E:\backup\tmp\old>echo 2 >> out2
> E:\backup\tmp\old>echo 3 >> out2
> E:\backup\tmp\old>echo applenuts > out1
> E:\backup\tmp\old>echo bakedgoods >> out1
> E:\backup\tmp\old>echo cannedfruits >> out1
[...]
> E:\backup\tmp\old>paste out1 out2
> applenut1
> bakedgoo2
> cannedfr3 ts

I was able to recreate the return, newline sequence using the standard
shell printf program which allows explicit control over newlines more
easily than using echo.  (I usually use echo myself unless I need that
extra level of control.)

  printf "applenuts\r\n" > out1
  printf "bakedgoods\r\n" >> out1
  printf "cannedfruits\r\n" >> out1
  printf "1\r\n" > out2
  printf "2\r\n" >> out2
  printf "3\r\n" >> out2

  paste out1 out2
  applenut1
  bakedgoo2s
  cannedfr3its

So it *appears* that I can recreate your problem.  However, those
return characters are being handled by paste.  Let's print all
characters and make invisible characters visible.  If 'cat -v' is not
available to you then you will need to use 'od -c' instead and deal
with the differences in output.

  paste out1 out2 | cat -v
  applenuts^M     1^M
  bakedgoods^M    2^M
  cannedfruits^M  3^M

Now it should be apparent what is happening.  Paste is adding a tab
character between the fields as it should.  But the return character
which is part of the input files are telling the terminal to move to
the start of the line.  The tab moves right from there.  The result is
that it overwrites part of the line.  This is working as it should
work because as the old adage says, it is doing exactly what you told
it to do and not want you wanted it to do.

You reported that using the Solaris echo worked.  On unix systems the
end of line convention is a single newline and not a return-newline
pair such as it is on MS systems.  Therefore the file did not contain
any return characters on the Solaris system.

Since the numbers are very small the tab skips over them and it
appears to work because you can't tell the difference.  If the out2
file containing the numbers instead contained wider fields it would be
apparent in that case too.

It is not clear to me if you have a problem with the MS-Windows
version of the tools.  Handling the end-of-line sequence as two
characters has always been troublesome.  I can't deduce if you have a
broken version there or are just copying files from MS to UNIX and
therefore mixing end-of-line conventions.  If you can do everything on
one host and it works then I think all of the tools are working fine.

If it fails to work on MS-Windows when run completely on that host
then I suspect that you have a broken version of the tools there.  You
should update to the current versions and try again.  However, because
MS-Windows has different conventions for binary and text files such as
this extra return character it makes it very difficult and sometimes
impossible to operate correctly on MS-Windows.

http://cygwin.com/
    Cygwin Home Page

http://cygwin.com/faq/
    Cygwin FAQ.

http://cygwin.com/bugs.html
    Cygwin Bug Report Guidelines. This is a good reference for general
    bug reporting.

Hope that helps.
Bob