Bill Studenmund wrote:
> On Mon, 8 Feb 1999, Christoph Badura wrote:
> 
> > dolecek@ics.muni.cz (Jaromir Dolecek) writes:
> > >One more shot -- for object files, the grepped line is useless anyway
> > >(you are not going to edit the object files manually, are you ?),
> > 
> > Oh, it's useless?  Surreal!  Obviously, if I couldn't get the information
> > I am looking for by greping the object files I wouldn't grep them in the
> > first place!
> > 
> > >so information whether the string has been found is enough even for you 
> > >I guess :)
> > 
> > Eh?
> 
> As I understand it, for a "binary" file, grep will say "I found it" for
> the file rather than showing you the "line" on which it found the string.
> 
> So it'll still tell you which .o contains foutine name X, it just won't
> show you. :-)

I have no argument with Bill's note, he simply collected the right
thread of conversation to knock me off the fence I was sitting on.

It seems that using grep as a filtering process in a pipeline would
be severely hampered by using this "binary file" interpretation.
Consider the following case:

	grep foo ./* | grep bar | ...

If we are concerned about not breaking scripts that use the assorted
options to tar (the pax vs. GNU Tar debate), how about scripts using
the above construct, or its moral equivalent...

/sjd