I need to generate UUID with `uuidgen` (not through any function call, but external command, because processes which will read the file cannot call any APIs or external commands directly).

My idea is to create an `incron` job which regenerates content of file on read (`IN_ACCESS` operation).
So that whenever the file, monitored by `incrond` is read, a script overwrites its content with new UUID:

    #!/bin/sh
    /usr/bin/uuidgen > file_to_be_monitored_and_read

My questions:

 - Is there another way to create such a device like file?
 - What problems can arise if multiple processes try to read from the file in about the same time? Will `incrond` and script be able to regenerate the content quickly enough?