This is a placeholder for the Deliverable 4.2 due for publication in March 2022.
Commitment
- Keep track of R&D during the DREAM project in order to write the final report.
- Anticipate next research stages.
Outcome
Complete final milestone. Final report available at https://dream.public.cat/pub/research-development-prospective
DREAM Research & Development Report and Prospective
Outcomes
Difficulties
Expectations
Developers Expectations
Funders Expectations
Human Factors
The COVID-19 Crisis
Remote Work
Life Contingencies
During the year 2020, out of seven team members, 5 moved from one country to another, and 3 had children. This might be a sign that DREAM provided a safe space for people to engage in life-changing behavior. 
Future Work
Community Activation
In order to pursue DREAM work in good conditions, we need to activate the OCaml and academic communities. The former will enable finding new collaborators to implement missing functionality and reinforce the long-term viability of the project. The latter will inform us about the validity of our approach, help evaluate, test, visualize, and eventually formally prove the reference implementation.
Research Directions
The state we’ve reached after one year of research and development, although commendable, remains under our initial expectations. We have made good progress on certain aspects and have left others aside.
CRDTs
Conflic-free Replicated Data Types remain an active research area and our use of commutative observe-removed sets in DMC propose a tangential use-case for CRDTs, that we nicknamed “slow CRDTs”. We need extensive testing of the commutativity of the data structure, especially in case of several remote groups working on similar data and merging their data sets eventually, to understand the properties of redundant operations and how their affect the portability and optimization of the graph.
ERIS
The Encoding for Robust Immutable Storage (ERIS) forms the foundation for DMC research and is being pursued within the scope of the openEngiadina project. We expect a 1.0 release for complete functionality by June 2022. It will facilitate DMC development by providing a strong standard to distribute encrypted blocks.
HDTs
Header-Data Triples promise a binary format to store searchable RDF triples. Using this in DMC would allow to search through the contents without needing to decode and decompress stored data. This research has been conducted by an international team who is now focusing on C++ and Java implementations. An OCaml implementation could be useful to integrate this feature with DMC.
P2P
A lot of work is missing from the Peer-to-Peer aspect of DREAM as the associated P2Pcollab implementation did not reach completion as we speak, especially with regard to peer discovery and dissemination protocols that are still in flux. Therefore the current work done on the Router for the UPSYCLE Network, although instrumental in the design, remains useless as it is. Further collaboration with P2P projects beyond P2Pcollab, including Irde.st and Retroshare aim to provide a diversity of outcomes for DMC.
Usability
Finally, once the programs reach a certain maturity, it is important to provide academic and associate use-cases with usable systems to make proper use of DMC and other DREAM components. As much as we’d like to develop a strong user base as early as possible, it currently appears reasonable to keep DREAM outcomes into an academic environment for the time being. Once we reach Release Candidates, we can engage with users from the cooperative economy with confidence.