It is possible to have written the command like this:

Code:

cat $(egrep -l "pattern" *)

except that the result of the egrep might be more arguments than can be handled on a single command line. By piping the output to xargs, xargs will execute the command that you give it (cat in this case) with n number of arguments such that n does not exceed the limit from a command line perspective. It may invoke the command multiple times to accomplish its task.

With this said, I realise that my original command would be flawed as the '*' given to egrep might expand into too many arguments, causing an error, and thus the xargs in this case would do no good. This would probably be better:

Code:

ls | xargs egrep -l "system.*log|log.*system" | xargs cat > result

The ls command generates a list of filenames which is given to egrep via the first xargs. The result of the searches is then given to cat via the second xargs.

When writing a script it is usually best to assume the worst case and use xargs in this situation. If you're willing to live with the errors, and are pretty sure that there will not be more files in a directory than the max allowable args on a command line, then the first form (without xargs) is easiest and straightforward.