WHEN WILL I LEARN RSYNC?

So I want to synchronise two similar directories, maybe with some 
weird exceptions. My options are to manually go through and compare 
file modifications times (assuming system clocks have been correct 
and there aren't things unpacked from old archives etc.) in a file 
manager, selecting and copying as I go. Or I can simply "rm -r" 
everything that I intend to synchronise in the target directory, 
then copy everything over from the source ignoring the painfull 
inefficiency of copying over a majority of files that were there 
before. Or I can let the computer do it all for me with this handy 
tool called Rsync.

Oh no, I chose the last option! Didn't I know that every time I use 
that command I forget one of the alphabet of parameters that it 
takes to do what I want, or omit/mangle an exclusion wildcard, or 
get confused about whether to use a trailing slash on one of the 
paths. At this point I know most of the the mistakes that I can 
make, BUT I KEEP ON MAKING THEM. Yes I know that "-n" (dry-run 
mode) sometimes helps, but often it doesn't. Yes it prints a vague 
list of things that it's going to do, but on a large job that's 
miles long and I usually miss the part where it suddenly decides 
to replace my entire OS with one file named "bob".

What I need is some sort of Rsync-preview file system view where I 
can browse around a preview of what things will look like after I 
make all my fatal mistakes, before committing to actually doing the 
job. It would need to know about existing files and timestamps of 
synced files (with/without -t) to present a clear view, so a little 
more involved than just processing the output of "rsync -n".

Web searches for something like that don't return anything except 
old forum threads where people have "-n" yelled at them repeatedly. 
Maybe that's all _they_ need, but having just discovered that my 
upgraded icli OS SD card is _still_ not working after I spent over 
an hour fixing my most recent (post-"rsync -n") rsync mistakes last 
night, I know I need more. I don't think I've ever used Rsync to 
syncronise my testing and 'production' SD cards during an OS 
upgrade before, because I know I always stuff it up, but this time 
I thought "oh come on, I'm not that stupid, with Rsync I can avoid 
needlessly wearing out the flash by rewriting deleted files...". 
YES I AM THAT DAMN STUPID! NEEDLESSLY REWRITE THAT 500MB, WEAR OUT 
THE FLASH, JUST DON'T TRY TO USE RSYNC!

Once a fully tested Rsync command is safely locked away in a script 
that runs with no manual interaction, it's a wonderful tool which I 
admire. But I really need a better way of safely crafting those 
command arguments.

An alternative idea to the Rsync preview interface would be some 
sort of virtual file system which clones a real file 
system/directory but just supplies the directory listings and 
doesn't actually store files, like a pretend bind mount that 
doesn't actually touch files at the mounted location. It would be 
less efficient because Rsync would need to read the files thinking 
it's really copying them, but it would be very useful for testing 
other programs and scripts too. Who hasn't created a complex base
set of test files to make copies of then experiment with
rearranging them using a program/script they're writing and mid-way 
accidentally get confused about which directory they were in and 
run it on the base set instead of the test set? Or is that just me 
too?

Yes Git does similar-ish things, but it wouldn't be very efficient 
for testing a multi-GB Rsync run. Plus I always stuff up Git 
commands as well, doubly so if I'm trying to do anything even 
slightly unusual.

Anyway having wasted my morning internet time grumbling over all 
that, I'll put off the upgrade for possibly another month and go 
fiddle with my car instead. Another complicated thing that I never 
quite understand...

 - The Free Thinker