On Saturday 07 March 2015 23:41:17 Sven Eckelmann wrote:
> The dataset stores the source and the alfred_tlv of the send data next to
> the actual data buffer. The source and alfred_tlv::type was used to find an
> entry in the dataset hashtable when reading/updating the buffer. The
> alfred_tlv::length is always updated when the size of the associated buffer
> gets modified.
> 
> But the alfred_tlv::version is only written when receiving the first data
> for a dataset. But this initial (maybe outdated) version is always
> propagated when a client asks over the unix socket or the dataset is
> exchanged between alfred servers. This has to also be updated when new data
> is added to the dataset to avoid clients to interpret the data using the
> wrong data format version.
> 

Applied in revision 17d1102.

Thanks!
    Simon