The majority of the development team have been working and testing Vaults in the last week and are reporting back very positive findings even under aggressive churn. Another large scale test was started tonight and we shall update the community on our findings from this test later in the week. We hope to include the whole community in the next testing phase, which should begin very shortly.
##Crust : Vinicius / Andrew / Spandan
Because of the phenomenal success rate of TCP hole punching, we have decided to park uTP / UDP for just now and concentrate on TCP only for immediate iterations. Crust was built using blocking I/O because that is what we had when we started with Rust. However, blocking I/Os don’t scale much as is evident in the number of threads in our design. So Crust right now is undergoing a rewrite using async frameworks and will be made available in future releases. MIO is now stable and supported on at least three major platforms - Linux, Windows and OS/X. So we are in the process of brainstorming the best possible way to do async I/O (plain even-loop, co-routines, etc) and will code the library accordingly.
##Routing
All development resources have been reallocated to other libraries this week.
##Vaults : Andreas / Fraser / Qi / David
Most of the Dev capacity is currently focused on Vaults. The team rewrote the refresh algorithm so that a joining node does not cause a flood of large messages and instead downloads the data it needs to store gradually over a longer period of time.
The guys then moved on to testing mixed networks with droplets as well as nodes behind routers; problems with asymmetric internet connections became apparent. We are now working on a few measures at several layers to deal with connections that have much lower upstream than downstream bandwidth, as is the case with most ISPs.
##Client : Krishna / Spandan / Shankar
The focus from the Client team continues to be on the log visualiser and roadmap from last week. @krishna_kumar was working last week to get the log visualiser running against the actual Vaults. Today we hope to be testing this in a scalable hosting platform to allow the log visualiser to handle large volumes of data. The log visualiser will be ready for the internal dev team to use tomorrow.
@shankar and @scott have been working on the roadmap changes that were suggested by the community. Also, changes to the website that will be needed to support up and coming releases have been designed and implemented.
The plan is to get back into the RFC discussions, followed by prioritising and scheduling for RFC work from this week.
##RFC : lightyear / Ross
Community contributor @lightyear made a proposal to simplify and improve the RFC process and then raised the PRs to implement the proposed changes. The previous structure of folders and status progression of RFCs made it difficult to link to an RFC and for that link to consistently work and not be dead. So @lightyear proposed we look at the:
“Python PEP Process (which the rust RFC process is largely inspired by), how about adding a top-level field status
instead like they do (and maybe one that mentions the release version once it was published)? This way, each RFC could stay in one place and be linked to.”
The team liked the idea and @lightyear raised the PRs (also raised a tidy up one as well) and the change is now implemented; he has also added a page where you can easily browse RFCs by status. A great example of community contribution - thank you @lightyear.
##UI Design: Scott
Last week, @scott worked on the design of a number of updates to the MaidSafe website and also on a network status dashboard.
@scott is now working on designing the changes to the MaidSafe website which were detailed and scoped out in previous weeks.
Thanks as always for your support, here is a link to the weekly transcript.