Re: [Duplicity-talk] Error messages regarding backends

[Kenneth Loafman]

On Mon, Mar 5, 2012 at 6:56 AM, <address@hidden> wrote:

On 05.03.2012 13:27, Timo Schoeler wrote:
> On 03/05/2012 01:25 PM, thus Kenneth Loafman spake:
>> On Mon, Mar 5, 2012 at 12:41 AM, Timo Schoeler
>> <address@hidden>wrote:
>>
>>> Hi,
>>>
>>> I just fetched 0.6.18 source and built an RPM for RHEL 5 and 6,
>>> both x86 and x86_64. For all archs I get
>>>
>>> Import of duplicity.backends.giobackend Failed: No module named
>>> gio Import of duplicity.backends.sshbackend Failed: No module named
>>> paramiko
>>>
>>> when issuing my backup jobs. This didn't happen before -- is that
>>> a regression? Did the build process change?
>>
>>
>> No, not a regression, just exposed the message to normal logging
>> level as it should be.  It was previously 'hidden' and you'd have to
>> run with verbosity to see it.
>>
>> ...Ken
>
> Hi, thanks for the clarification.
>

we should probably issue these with debug level info _only_ if the backend matches the target. else users will probably come around more often to complain duplicity issues errors although this one is not even remotely in their gift basket.

just checked
http://bazaar.launchpad.net/~duplicity-team/duplicity/0.6-series/view/head:/duplicity/backend.py
and see that there is no easy solution to it. to find out what protocol a backend supports we need it to be initialized. the mentioned backends do not initialize properly because of the missing components.

how could we make the backends initialize properly, but having them tell that there is an error later?

I don't know.  I do know that it's a catch-22.  If we hide the error, then it causes confusion, if we expose the error it causes confusion.

...Ken