3 ms·

> Why not use > file? I'm assuming sponge buffers everything in memory so you don't get concurrent modification bugs.

• masklinn 6y ago
  Yes, if you try to sed a file back to itself for instance you'll get a truncated file (or maybe a no-op?). In fact, that's exactly the example sponge's manpage uses.
  • lloeki 6y ago
    That redirection will make the shell truncate the file long before sed has a chance to open it.
  • moonchild 6y ago
    It's a truncated file: $ cat test moo $ cat < test > test $ cat test $
    • Izkata 6y ago
      I was wondering about tee for that: $ cat test moo $ cat < test | tee test > /dev/null $ cat test moo ...seems to work. Am I just getting lucky with a race condition?
      • notimetorelax 6y ago
        Possible, since you don’t pipe into the file but instead ask tee to write into it.
      • Erwin 6y ago
        Yes, check this, with f.txt containing 99999 numbers: $ seq 1 99999 > f.txt $ cat < f.txt | tee f.txt > /dev/null $ wc -l f.txt 23696 f.tx
      • moonchild 6y ago
        > Am I just getting lucky with a race condition? Probably this is buffering, either from dd or from the kernel. $ dd if=/dev/urandom bs=1M count=1 of=test.bin 2>/dev/null $ cp test.bin test2.bin $ cat < test2.bin | tee test2.bin >/dev/null $ diff test.bin test2.bin Binary files test.bin and test2.bin differ $ wc -c test{2,}.bin 131072 test2.bin 1048576 test.bin Notice the file got truncated at exactly 128k; a nice round size for a write cache.
      • kasabali 6y ago
        I think using tac | tac is a safe bet since must be reading all file before by definition.