because some filesystems can't do it

static

AIX)

thanks to kalt@research.bell-labs.com for this fix.

interactions with ssh. The old ssh freezes have shown up again and
some debugging (with help from James Welborn) showed that the cause
was the read buffer hitting maximum size. I think this means that ssh
must be misbehaving about blocking IO.

This change gets rid of the freezes at the expense of memory
usage. Where it would have frozen it uses more memory instead.

for the 3 logical stages of rsync; generator, sender and receiver.

by a read during an io_flush() triggered during a readfd(). A simple
logic bug in the io code :(

specified. This makes things easier to debug.

socket is multiplexed. This allows clients sending new options which
the remote server doesn't understand to get a sensible error message.

a daemon. This prevents library functions (such as getopt) stuffing up
our protocol stream when errors are detected.

- defer the error message from the options parsing until after the
socket is multiplexed. This allows clients sending new options which
the remote server doesn't understand to get a sensible error message.

there is then assume that the : is from a filename, not a host:dir
separator. This allows you to copy files with a : in them. (suggestion
from pfeifer@wait.de)

when -p (preserve permissions) isn't set.

It works by taking the sending file permissions and masking them with
the umask to create the destination file permissions. (There is really
no "correct" way of doing this but at least we now behave like GNU cp
which fits the principle of least surprise.)

also fixed a race condition in copy_file()

destination tree

destination tree

to a file.

and report an appropriate error message

  verbose info not being line buffered to files.

- add a call to localtime() in open_log() in order to prime the C
  libraries timezone cache before the chroot(). This should fix the
  problem of rsyncd log entries being in GMT time.

ssh-friendly attempt)