Why is it necessary to use delete-delay explicitly? Conversely, why
does rsync makes two scans of the local disk, one for the transfers
and *another* for deletions with delete-after? Why doesn't it simply
keep files that are not present in the source list in memory when it
does the first scan and removes them after? Making a second disk scan
can be pretty expensive.

I also don't understand why it always writes the names of deleted
files on disk (after a certain size) with delete-delay. I've been
using it since going to version 3, and noticed that it writes a
constant flux of about 100KB/s for 10 min during a large scan! That's
really much more than I'd expect... Why can't we tell it to just hold
the names in memory like is done with the sender list?

This is related to a feature request: I'd like to have a way to make
rsync just list the files that should be removed, instead of
effectively deleting them. I need the list to remove them later
however. This could be another option; maybe the implementation could
be similar to what delete-delay already does?
--
Please use reply-all for most replies to avoid omitting the mailing list.
To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync
Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html