Understanding Unix process substitution behavior

Question

awk '{print FILENAME, $0}' <(ls)  # output:   /dev/fd/4 file
awk '{print FILENAME, $0}' < <(ls) # output:   - file

In the above one-liners, the first one generates file descriptor and then the filename where as the second one generates the hypen (-) character and then the filename. Why this behavior?

Answer 1

You can see it this way:

awk '{print FILENAME, $0}' <(ls)
# is the same as
awk '{print FILENAME, $0}' output_of_ls_command

awk will read a tmp file (created by bash, let's name it as output_of_ls_command(it's /dev/fd/4 in your case))

---

awk '{print FILENAME, $0}' < <(ls)
# is the same as
awk '{print FILENAME, $0}' < output_of_ls_command

awk will read stdin (bash read the tmp file, and send the content to awk, FILENAME is -)