Today, we are posting the last dev update of 2017. The team will be working intermittently over the next couple of weeks due to the holidays The next dev update will be on January 11, 2018.
Here are some of the main things to highlight this week:
The front-end team is releasing a few updates, including a new patch version of the SAFE Browser, a new version of Peruse (the custom browser that will eventually replace SAFE Browser) and a new version of the Web Hosting Manager that fixes a minor issue reported by @bzee.
The Routing team has updated the Dev Forum post that contains the proposed implementation of Data Chains.
The Crust team has published a roadmap of what they are planning to work on next.
Community Engagement Program
As we have been mentioning recently, marketing has been getting much more focus of late. A big part of our strategy moving forward is to engage much more with you guys – the SAFE Network community – to make the most of your skills, passion, and hard work for all our benefit. The Community Engagement Program, which we are launching today, is the first part of this process.
We have created a short video (with the help of @whiteoutmashups) and an infographic to help you understand how to participate in the CEP program and we hope to see many community members getting involved in the near future.
Initially we will be running one project only to help refine the process, we plan to increase the number of concurrent projects in the near future. The first CEP we ran back in 2016 produced the SAFE Browser and we are now inviting the community to submit proposals for a Safecoin animated video. We hope that this will raise the profile of the SAFE Network, it’s unique features and use cases once live. This should be a high-level video that acts as an introduction for newcomers who have little or no previous knowledge of the SAFE Network. The full Request for Proposal can be found here and we welcome questions and participations in the thread. Once you and/or your team have an idea, please submit your proposal in the #community:proposals category within the next 2 weeks. We look forward to seeing what you come up with!
As the development and marketing plans continue to roll out we are also placing a lot of emphasis on growing the team. With the impending opening of the Indian office in late January, we are looking to fill 7 roles initially to increase the capacity and range of skills within the front-end team. We are currently looking for two Java developers, two .NET developers, two web/UI developers and one designer with the Java and .NET developers also working on mobile.
Closer to home in the Ayr office we continue to look for a UI/UX designer, two Rust Engineers (preferably local), a QA/Release Manager, Customer Support Manager and another administrator. Some of these roles have proven to be difficult to fill, but we have involved a number of recruitment agencies to try and speed things up while still maintaining the quality of candidates.
We are starting to see progress on some of the roles so we are optimistic we will see some movement soon. It would be easy to rush and take candidates that we have doubts about, however, we have found in the past that this can cost more in the long run so are pushing the recruiters hard to get the best people possible.
SAFE Authenticator & API
- Fix the
safeMutableDataEntries.forEachfunction which was incorrectly returning the ‘key’ as an object
- Allow MutableData handles to be removed from the safe-app plugin’s Map through a ‘free’ function
- Minor fix in the DOM API documentation example for the safeCryptoSignKeyPair.getSecSignKey function
A new version of the
@maidsafe/safe-node-app npm package (v0.6.0) has been published which incorporates a couple of minor fixes and it upgrades its safe_app library dependency to the latest version (v0.5.0). We have also made a minor reorganisation in the tests as well as increasing the test coverage to 90%+.
As a consequence of this, new patch versions of both the Web Hosting Manager (v0.4.4) and the email application (v0.4.3) are being released, mainly to upgrade their dependency with
safe-node-app to v0.6.0.
The new version of the Web Hosting Manager also incorporates a fix for an issue recently reported by the community (thanks @bzee!) in relation to a convention for service names (more details can be found in this discussion on the forum).
This is the very first time we are able to generate 100% of the released packages automatically from our CI processes, which has not only saved us considerable time and energy but also gave us a good knowledge base to be able to automate such processes for other artifacts in the future.
Peruse has been updated with a number of enhancements and increased test coverage, as well as a larger behind the scenes refactor of the Redux store to enable sync across many window instances. This paved the way for History and Bookmarks pages in the browser, which are now enabled.
This is all available in the latest early-stage release, which you can find here, alongside a more detailed list of changes. This version is very close to being functionally equal to our Beaker Browser fork. We now need to make sure it is equally stable, as well as start putting some focus and effort on the UI/UX.
Beyond that, now that we have History/Bookmarks in place, we’ve been refactoring the save/retrieve state implementation for the new Peruse browser.
The JNI bindings have been integrated with the Java API. However, we are expanding the test cases right now and then we’ll update the documentation for the same. If anyone wants to try it out in the meantime, the JAR files for desktop can be built by following the README. The Android AAR files are not yet ready. The tests are not in place to assure that the Java APIs are stable. Considerable progress has been made this past week so expect tests and documentation to be updated soon.
Once the Java documentation is updated, we will be getting the C# API updated to be compatible with the latest native libs and have the mobile applications using these APIs also updated and published for Android and iOS.
SAFE Client Libs
We are releasing new versions of SAFE Client Libs: safe_app 0.5.0, safe_authenticator 0.5.0, ffi_utils 0.4.0, and safe_core 0.28.0. This release includes all of the improvements and bug fixes that were made to SAFE Client Libs over the last 3 months. Major changes in this release include: new cryptography functions in SAFE App, multiple improvements in the test suite (that helped us to catch and fix many obscure bugs), support for config files in SAFE Core (allowing to use in-memory mock storage and unlimited mock mutations), improved documentation (which is available on docs.rs), and many refactorings of the FFI functions. Overall, with this release, SAFE Client Libs has become much more stable and mature.
We are also continuing to work closely with the front-end and mobile teams to wrap up the auto-generated Java bindings. This week we have discovered some minor bugs (mostly related to naming conventions) that should be fixed very soon, so there should be more updates on this next week.
Routing & Crust
The last year has been full of hard work, but it is finally starting to visibly pay off.
Throughout this year, we have been focused on the design work. We were exploring consensus algorithms like PBFT or Tangaroa, and constructing our own, which resulted in the creation of the Ewok simulator. We had multiple brainstorming sessions and lengthy discussions, which were the source of most of our current ideas and gave us a lot of insight. All this work made us learn a lot about the minute details of how the network will function and what we want it to do. We were able to get a closer look at multiple designs and this allowed us to choose the one we think will work best.
Going back to present events - this last week was again full of thinking about the approaches to merging. After fleshing out all of the details we were finally able to come up with a proposal that we think will work. We also prepared a similar approach to splits. The full description is available in this Dev Forum post, but here is a short one:
Both merges and splits will begin with reaching consensus about the fact that such an event needs to happen. Next, in the case of merges, we need to additionally reach consensus about some details of the sibling section. Afterwards we start voting on blocks that will gradually change our current set of elders into the target one. During this part, since we are dealing with two sections most of the time, all peers that were elders or are supposed to be elders will vote - this way we can ensure that the relevant blocks will get a quorum of votes regardless of the order in which they accumulate (which can be different for different peers). Once we reach the set of elders we initially calculated as the target, we consider the merge/split complete.
In Crust, we have successfully integrated uTP into the
dev branch. We have specced out our test plan and feature roadmap which will enable us to validate multi-protocol, randomised-port networking in various real life scenarios.
We had to make some changes to
p2p as it was using a singleton object for NAT traversal configuration. Though nice for user friendliness it soon became an issue for testing. Rendezvous server integration into Crust was also pending. This week both issues were fixed: hard-coded contacts in
Crust config are used as rendezvous servers and
p2p uses a special structure (
P2p) for NAT traversal information.
uTP was already merged and direct connections seemed to work properly. Now that we are using the rendezvous servers we have a few ways to test peer-to-peer connections. We did however realize that it was difficult to integrate
UDP rendezvous server into
uTP crate changes were required. These were promptly carried out and
Crust now has the ability to rendezvous over
UDP - an essential part of NAT traversal using
UDP and subsequently feeding those hole-punched sockets to
uTP for permanent connection-oriented communication.
We will also soon start with Crust integration into Routing. Routing
master will be branched off and since Crust, in the new
dev branch, exposes a compatibility layer (because there’s no immediate plan on refactoring Routing yet to work with Tokio event loops) the integration should hopefully not be hard - only the errors and some other types have potentially changed. Once this is done, we will spin up a few droplets and test for any regressions since we pretty much know what to expect from Routing’s
master - Alpha-2 is running on it after all.