Em Sun, Apr 15, 2012 at 04:26:27PM +0200, Paul van Tilburg escreveu: > On Thu, Apr 12, 2012 at 08:54:55PM -0300, rhatto wrote: > > Em Mon, Apr 09, 2012 at 02:01:45PM +0200, Paul van Tilburg escreveu: > > > What it IMO doesn't solve, is the fact that the handler gave a syntax > > > error and probably returned and error code, but backupninja intepreted > > > this as "backup succesful". This worries me a bit. > > I meant this differently. It just seems that if you make an error, e.g. > you set: > > rsync_options = --non-existing-bla --syntax-error-coming up > > and rsync bails out, or maybe segvs or whatever, backupninja happily > reports via email: SUCCES.
Yes, that's another issue: https://labs.riseup.net/code/issues/3892 > > I think that the only way to avoid that is to validate the metadata. > > > > The following commit tries to do it: > > https://git.sarava.org/?p=backupninja.git;a=commitdiff;h=e22107cf0954f29215052becf848bc28b47ffbe0 > > > > Could you test to see if it works? > > I am reluctant to apply this... why also first $created but \$created in > the second hunk? The second hunk refers to code to be executed in the remote side, so it has to be escaped to not evaluate at the local side. Maybe would be better to wait me to sort out the most important rsync handler issues and then give you a single patch to try so I don't bother you too much with my development as I already did. :/ -- rhatto at riseup.net pubkey 64E39FCA / keys.indymedia.org
signature.asc
Description: Digital signature