The generator can skip a directory's contents altogether due to
--ignore-non-existing, a daemon exclude, or a mkdir failure. On a
--dry-run, the generator can also note the missingness of a directory
while still scanning its contents. These two scenarios were conflated
using a single set of missing_below/missing_dir variables in combination
with transient increments in dry_run; this caused at least three bugs.
Now recv_generator has separate variables for the two scenarios, called
skip_dir and dry_missing_dir, respectively. For simplicity, we take the
F_DEPTH instead of having separate *_below variables. We mark both
kinds of missing dirs with FLAG_MISSING_DIR. (dry_run > 1) iff the
*root* of the destination does not exist; it is no longer incremented
for missing subdirs. I added tests for the three fixed bugs in
missing.test.
the FLAG_SKIP_HLINK flag, which gets set on any hard-linked file that
the user wants to skip (e.g. via --ignore-existing, --append, etc.).
The code in hlink.c now deals with the skipped files instead of
triggering an assert() error.
- If a symlink/device/special-file changes its value without any
attribute changes, the itemized event no longer gets dropped.
- We put a 'c' into the checksum/change field now to indicate when
a symlink/device/special-file changes its value without changing
its type. This lets us properly interpret the --copy-links output
to know which items are getting copied without changes and which
are getting created with new content.
- Fixed the 'T' itemized output for a symlink when rsync tries to
set the right time but fails due to lack of OS/disk support.
preserving ownership, but we want to know what the UIDs are on the
files we're deleting. Changed the DEL_OWNED_BY_US flag to be
DEL_NO_UID_WRITE, which is only set when a file is owned by us and
we can't write to it. Fixed a glitch in the error handling of the
--delete-delay code where it might try to enable delete-after in
incremental-transfer mode. Made a simple (well tested) optimization
in the --delete-delay code and a few other readability changes.
tell the client if the server can set the times on a symlink (both
the server->client byte and the client->server use of -e). Make use
of this info to allow the proper output of the 't' flag when rsync
can set the time on a symlink (and we're talking protocol >= 30).
Added output of "[no] symtimes" info in the --version message.
Fixed the itemize.test so that it works when rsync believes that it
can set the time of a symlink, but it can't really do it.
- Changed itemized output to only report 'p' with -p or -E.
- Fixed a duplicate output of a link-dest file with -vv but no -i.
- Improved unchanged_attrs() to handle -E.
shared by multiple files: handle the case where one file has an
abbreviated item set correctly, but a following item does not.
Also extended testsuite/xattrs.test to verify that this works.
