Re: [Duplicity-talk] librsync protocol: version 30?

[Chris Poole]

I emailed Martin Pool (developer of librsync) to find out. His response:

> librsync uses the rsync algorithm but not the same binary protocol as rsync.

So I still don't really know what hash function is used, if at all, to
generate the diffs (and thus what the probability of a collision is),
so when I get time I'll dive into the source code and see if I can
figure it out, but it seems like it might not be an issue.


Chris

On Thu, Sep 1, 2011 at 4:57 PM,  <address@hidden> wrote:
> could somebody enlighten us here? ken?
>
> ede/duply.net
>
> On 01.09.2011 16:54, Chris Poole wrote:
>> On Thu, Sep 1, 2011 at 3:36 PM,  <address@hidden> wrote:
>>> isn't the protocol not only important for rsync<->rsync communication?
>>>
>>> as far as i understand, duplicity is using the rolling checksum from 
>>> librsync
>>> only and this algorithm shouldn't be affected by this issue.
>>
>> I was under the impression that it used both the rolling checksum and some 
>> kind
>> of 'overall' checksum function, here a hash, to generate any diffs, but I 
>> could
>> be very wrong.
>>
>> Chris
>>
>> _______________________________________________
>> Duplicity-talk mailing list
>> address@hidden
>> https://lists.nongnu.org/mailman/listinfo/duplicity-talk
>
> _______________________________________________
> Duplicity-talk mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/duplicity-talk
>