Efficient pipeline

[tombruijn]

Try to keep tmp file creation to a minimum. Pipe the actual output from one command to the next in order to create more efficient commands and not hog the available space on the machine.

See also this issue on the Backup issue tracker: backup/backup#594

[epcim]

+1

how about to use FIFO pipes (https://github.com/shurizzle/ruby-fifo). IE avoid complex rewrite of core, just write an resource to pipe cmd's to fifo pipe (in the archive method (when tar is being created)) and if "pipeline" is used start asynchronously syncer/uploader method which would read FIFO pipe until ( pipe is empty AND the "archive" process is finished).

This would allow some flexibility like, any CLI command in model that would be sending data to stdout would be possible to use.