On Sun, 2010-03-28 at 20:08 +0200, Martin Langhoff wrote: > On Sun, Mar 28, 2010 at 2:29 PM, Bernie Innocenti <ber...@codewiz.org> wrote: > > On Sun, 2010-03-28 at 14:14 +0200, Tomeu Vizoso wrote: > >> I'm a bit confused, is the problem reliable full restore with limited > >> local space or the DS having the xapian DB open while the copy > >> happens? Or both? > > > > Both, but somehow we managed to solve them in a quick & dirty way: > > Doesn't copy-to-journal skip all the ugly?
Oh, well, indeed. Though we need to figure out how to glue it to the rsync back-end. The easy part is binding rsync to Python: this was already done in duplicity by wrapping librsync with simple C and Python code which we could steal. Then it starts to get hard: the files are probably going to come down from the xs in some random order. librsync provides a sophisticated callback interface, but there seems to be also a whole-file interface. This approach also opens other interesting possibilities such as restoring individual files or merging the backup with the current journal contents. However, it looks like it would take a couple of months of effort to get something like this done. For the time being, we could keep using the plain rsync and maybe just add a stop/resume interface to the datastore? -- // Bernie Innocenti - http://codewiz.org/ \X/ Sugar Labs - http://sugarlabs.org/ _______________________________________________ Sugar-devel mailing list Sugar-devel@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel