Wayne Davison [Sat, 3 Nov 2007 16:57:17 +0000 (16:57 +0000)]
- Renamed updating_basis -> updating_basis_or_equiv.
- Set updating_basis_or_equiv if we're inplace-updating a
file using the backup file as the basis.
Wayne Davison [Wed, 31 Oct 2007 14:02:51 +0000 (14:02 +0000)]
- In --iconv mode, we can't lower the ndx_end value because dirs
aren't sorted to the end.
- Made output_flist() output from "files" instead of "sorted".
Wayne Davison [Wed, 31 Oct 2007 04:43:32 +0000 (04:43 +0000)]
- Fixed a problem with merging dot dirs with non-dot-dir args
in inc-recursive mode.
- Clean up a relative name to remove interior dot dirs and extra
slashes.
- Fixed the error output about /../ in a -R path after a /./ cut-off.
- Changed the starting ndx value to 1 so that a dot-dir flist can
use 0 to refer to its parent (".") directory.
Wayne Davison [Mon, 29 Oct 2007 20:43:34 +0000 (20:43 +0000)]
My version of Matt's cleanup patch from bug 5051. This makes
--copy-dest use a temp file when not in in-place mode, and has
various improvments for the code. I have also "#if 0"ed the code
in the receiver that makes missing directories to see if we can
figure out if it is needed (and if so, what for).
Wayne Davison [Tue, 16 Oct 2007 15:10:12 +0000 (15:10 +0000)]
- Send an 'i' in the -e arg to the server if we can support inc_recurse.
- Got rid of some unneeded blocking_io value twiddling.
- Send --use-qsort to the server if it was specified.
Wayne Davison [Tue, 16 Oct 2007 15:10:09 +0000 (15:10 +0000)]
- Moved the arg-checking relating into set_allow_inc_recurse() and
call it when the server is in setup_protocol(). The function sets
allow_inc_recurse to 0 if some options won't allow us to support
an incremental-recursive transfer.
- The server now checks for an 'i' in the -e option from the client
and zeros out allow_inc_recurse if not found.
- The server reports its inc_recurse determination back to the client.
- The client sets inc_recurse based on the value it gets from the server.
Wayne Davison [Tue, 16 Oct 2007 04:08:18 +0000 (04:08 +0000)]
- Send a flag from the client to the server to indicate if the protocol
is going to run in inc_recurse mode or not.
- Verify that the options we received (as a server or as a batch-reader)
are compatible with the requested inc_recurse mode.
Wayne Davison [Mon, 8 Oct 2007 18:02:11 +0000 (18:02 +0000)]
- Use the new *_CONTENT_DIR flag names (was *_XFER_DIR).
- Check the FLAG_IMPLIED_DIR flag to find implied dirs.
- Expect flagged implied dirs for all protocol 30 transfers.
Wayne Davison [Mon, 8 Oct 2007 18:02:07 +0000 (18:02 +0000)]
- Mark implied dirs with FLAG_IMPLIED_DIR.
- Use the new *_CONTENT_DIR flag names (was *_XFER_DIR).
- Send implied dirs with XMIT_TOP_DIR | XMIT_NO_CONTENT_DIR.
- All protocol 30 transfers send flagged implied dirs.
Wayne Davison [Sun, 7 Oct 2007 20:43:49 +0000 (20:43 +0000)]
Reposition the call to setup_iconv() so that it gets called only once
is a local_server copy, and so that arg-parsing code doesn't interfere
with the sending of a remote charset value.
Wayne Davison [Sun, 7 Oct 2007 07:18:49 +0000 (07:18 +0000)]
Improved on the list-only fix and on the other similar checks that want
to ignore implied dirs that are actually present in the file list
when inc_recurse is active and --no-implied-dirs was specified.
Wayne Davison [Sun, 7 Oct 2007 07:18:45 +0000 (07:18 +0000)]
- Simplified the code in recv_file_entry() for protocol 30.
- Simplified the flag setting in send1extra since we know that
recurse is set.
- Make sure that FLAG_XFER_DIR is unset for a non-dot-dir when
recurse is not set.
- The --relative code wasn't setting is_dot_dir enough.