[Dump-announce] (no subject)

[Tim W. <sou...@wo...>]

Hi everyone,

I've pushed 0.4b53 today. This fixes most of the outstanding tickets for 
dump, some very old :-)

The interactive restore is much improved, as well as proper quoting and 
handling spaces in filenames, it also no longer appends a space while 
tab-completing paths, only once you tab-complete the filename.

Let me know if you spot any more annoyances or things it's doing wrong, 
it's hard to make sure everything has been tested and I might have 
missed some things.

One change that might be useful to some, the restriction on not doing 
incremental dumps of subdirectories is now relaxed.

And there's now the option to dump via a bind mount - but note that the 
entire device is dumped, not just the data from the bind point. Don't 
rely on this, but also don't complain about it :-)

--use-blockdev-for-dumpdates has been added to restore the old behaviour 
of the dumpdates file containing the device rather than the parameter 
passed to dump. If this is used then incremental dumps of subdirectories 
are still not allowed.

Note for anybody building the code - by default it now builds with 
-Werror. There's --disable-werror available if you're building with 
different build options or on older compilers.

I also cleaned up some of the more obvious typos in the manpages.

Enjoy.

Tim.

p.s. my changes I needed in order to test using fuse2fs haven't yet made 
it into the official sources, but I've started a project here:
https://github.com/tjwoodall/dump/
to try and get some sort of automatic testing done. This is very much a 
work-in-progress but for anyone who wants to run all the tests, it 
should give hints as to what is necessary.