Pipe vs redirect into process

Question

Looking for some bash-expert explanation. What is the exact difference between of the next

command1 | command2

e.g. classic pipe where the stdout of command1 is redirected to the stdin of command2, e.g.

• bash forks itself two times
• changes the file descriptors
• exec the comman1 and command2

and

command1 > >(command2)

where the result (and the bash actions) are the same...

At least i got the same from the

find . -print0 | xargs -0 -n1 -I% echo =%=

and

find . -print0 > >(xargs -0 -n1 -I% echo =%=)

Is the > >(command) only the longer way saying |? Or not, and I missed something?

Answer 1

To implement

command1 | command2

the shell creates a pipe in the parent process and attaches one end of it to the output (fd 1; it uses dup or dup2) of command1, and the other end to the input (fd 0) of command2.

To implement

command1 > >(command 2)

, the shell creates a FIFO. It attaches the command2's stdin to the FIFO (with open using the O_WRONLY flag, usually) and it passes the name of the FIFO as a positional argument to command1. You can easily see this by using echo >(true).

If you use the

> >(foo)

these forms are indeed very similar. However, the process subsititution mechanism is more powerful. For example you get do this kind of thing:

diff -u  <(curl 'http://www.a.example.com/')  <(curl 'http://www.b.example.com/')

You can't do that with pipes - you can't have two standard inputs.

Answer 2

At least part of a difference is discussed in my question.

An additional difference is control over which processes run in sub-shells:

$ declare -p f b
-bash: declare: f: not found
-bash: declare: b: not found
$ { f=foo; true; } | { b=bar; true; }
$ declare -p f b
-bash: declare: f: not found
-bash: declare: b: not found
$ { f=foo; true; } > >( { b=bar; true; })
$ declare -p f b
declare -- f="foo"
-bash: declare: b: not found