This:
parallel --results mydir/ echo ::: foo
parallel --results {}.out echo ::: foo
parallel --results my.csv echo ::: foo
parallel --results my.tsv echo ::: foo
copies output to both the place given by --results and to stdout/stderr.
Personally I only use the output to stdout/stderr for debugging, so in
production I would always do:
parallel --results ... echo ::: foo >/dev/null 2>/dev/null
But that is really a waste of I/O, and if the amount of data is big,
this can be a considerable performance penalty.
So I am considering changing the behaviour, so that if you use
--results then it will not be copied to stdout/stderr.
In other words: --results will still save to the files, but the files
will not be read back and printed to stdout/stderr: Your terminal
would be silent.
What do you think? Please vote:
[ ] I don't care
[ ] Change it - I would prefer the new behaviour
[ ] Keep as is (please give an example why the current behaviour works
better for you)
/Ole
