So, first update post test network, SAFE Launcher and SAFE Demo App release and to say it has been a whirlwind would be quite an understatement - the response has been amazing! Since Friday the testnet has seen:
Total downloads of SAFE Launcher from Github directly: 1126
Total client sessions: 8119 (includes registered and unregistered clients)
Total get requests made: 350842
We’ve had a lot of very useful feedback and for that we are grateful to everyone that has taken time to provide it. Apologies for not acknowledging everyone yet, it’s great that we have such an engaged community, we are currently collating the feedback and are using it to help shape the next iterations to be released.
Work has begun to improve the logging framework of the test networks during these rolling release phases; this will provide the team with even more useful data to analyse. We will be collecting data and sharing it - only during the test phase - the full list of what we are planning on collecting has yet to be finalised. For full transparency we shall outline this very clearly before every roll-out; if you have privacy concerns it may be better waiting until this phase has been completed.
With regards to the MVP status, the team are still working on the one last outstanding TODO.
- NAT Traversal integration into Crust (allows for people behind different types of routers to connect with each other and not just rely on direct connections between nodes in the network);
##Crust - uTP / API : Vinicius / Andrew / David
The main focus for the team is still on closing out the tasks associated with the outstanding TODO (above), with uTP testing and bug fixing ongoing. The Crust guys have also been working on enabling UDP rendezvous connect. This is a particular connection type where both ends try to connect at the same time (approx) and forces hole punching to hopefully succeed. In essence this is a large part of getting vaults to work from everyone’s home and office, without manually port mapping a router or buying a direct internet connection. The initial implementation of TCP hole punching is now also ready for testing.
There is a paper that has shown a small change to these processes can give 100% nat_traversal over symmetric Nat devices. Thls will be added in parallel and if successful allows a ton of logic and code to be removed from Routing, where we jump through hoops to get nodes connected, using a huge number of network messages. So great news and all with very little effort.
##Routing - Vault Integration : Andreas / Brian / David
Most of the work in Routing in the past few days has been preparation for last week’s test release and some last minute fixes. Routing now also indicates to the upper layers if it gets disconnected from the network.
Next, we will focus on stabilisation and on proper handling of joining and leaving nodes, as well as some cleanup within the code.
##Vaults : Adam / Fraser / Qi
The droplet testnet shows vaults running reasonably stable in handling requests from sample apps in launcher. The current priority of vaults is to test and further implement churn handling capability. CI tests have been heavily updated to provide a certain standard of churn simulation as part of the testing environment. The debugging work that has been carried out so far has spotted a couple of bugs.
##Client - Launcher : Krishna / Spandan / Shankar
The current client focus is on fixing the issues with the SAFE Launcher and SAFE demo application. @ustulation (welcome back Spandan!) has resumed work and he is trying to get the network callbacks propagated to the upper layers. We expect that this will bring more stability to the Client libraries. @krishna_kumar and @Shankar have been improving the SAFE Launcher and demo application by fixing the known limitations and identified issues.
We expect that the callbacks will be integrated into the safe_core library within the next couple of days. Once that is completed, the Launcher will have to integrate the changes being made in the safe_core library for listening to the network event and act accordingly. We hope that this change will bring more stability to the applications and also resolve a few issues which are related to the network event handling.
##Development roadmap : Scott / Shankar
During the main website update we found that our (then) current implementation of the roadmap didn’t adequately communicate the features we were working on. We began to think of how we could improve on this and one area of focus was making sure people new to the project could follow the roadmap to some degree. However, it was also key that the design did not only cater for those new to the project, it was important that dedicated followers could also drill down and get much more information and detail. To do this we designed an interactive approach inspired by Gantt charts, where users could navigate down into features to see more detail. Work on implementing this design began over Christmas and New Year and a first draft of this was completed in January, though it has taken a bit of a back seat during February when we concentrated efforts on the testnet. This week we are focusing again on reviewing and fixing these issues.
##UI Design : Scott
Last week a lot of work was done in frontend on the Launcher and demo app UIs. Now, following the test release, work has begun on revising some of the Launcher UX, based on user feedback and observations and this work will continue this week and as the test network progresses.
##Customer Support : Ross
As with the network itself, customer support i.e. how we at MaidSafe respond to feedback from the community, will evolve and improve as the platform matures. So bugs, requests for technical assistance, feature requests, etc need to be collected, understood and dealt with efficiently, or if required converted into tasks for the Dev team. We are currently trialling JIRA Service Desk to do this; it seemed like a natural choice as we already use JIRA Core, so integrating both systems was already done. We have identified a couple of issues that we are currently trying to work around, however the main one is that folk need a login and even then can only see issues they have raised themselves from the customer portal, rather than the ability to view all issues. So we may have to rethink this strategy to make this much more transparent and community focused. We would like to thank everyone in the community for taking time to ask and answer questions, folks helping each other out is beautiful to watch. A special mention to all the forum mods who are doing a great job especially with the increased activity: your efforts are very much appreciated by everyone here at MaidSafe.
On Friday the roll-out of the SAFE Network testing began, the genie is out of the bottle and it is all very real now. Our focus is still very much on delivering the MVP and allowing folks to run vaults from wherever they wish, the fun has just begun…
Thanks again for the ongoing support, here is a link to the transcript.