MaidSafe Dev Update - 28th June 2016

Hello everyone,

We are working in several parallel tracks right now. This week we will share the next iteration of binaries and invite the community to partake in TEST 5. We were aiming for today, as per usual, but there are a couple of “niggle like” loose ends to clean up. We anticipate these will resolved and we will be ready to release TEST 5 tomorrow, assuming all of our tests are ok. There were a few large changes here, so we want to make sure it’s not a wasted test.

The purpose of this network will be to test specific aspects of SAFE, one of which includes improved data persistence. This involves disabling the ability to run multiple vaults per LAN, as well as re-implementing immutable data caching (this was turned off in TEST 3). These tests are very important and have been incredibly valuable, we really plead with the community to not attempt to disable the switches we have set here. This is of huge importance. We are still chasing down the mechanism to handle very low quality nodes, both in terms of measuring these and also having them stopped or relocated. We appreciate any input on these matters as we continue the testing.

Another track, as you know is the Community Engagement Project. We had hoped to publish the proposal request for the SAFE browser today, but time has beaten us. We hope to publish this proposal this week though.

We are also considering some further funding options in relation to MaidSafe itself, which hopefully we will share in the next few weeks / months. We cannot publicise all negotiations, but will endeavour, as always, to keep you as informed as is feasible. It is fair to say though, that there are some options that we feel will again benefit everyone. We are also trying to ensure the current exchange issues in China are being co-ordinated with tech support from the Omni team. The issues experienced are more process driven as opposed deep underlying technical issues.

The client apps, including launcher, are also seeing larger improvements in the background. These improvements are all aimed at providing much more efficient and clear feedback to users. The community really has helped to show a clear path here and allow us to effect the designs required.

##Crust, Core: Andrew, Spandan (tl) & Vinicius.

Core now does error recovery for composite operations. For example, if Operation A required 2 PUTs to succeed and due to some reason the 2nd PUT failed, the user is informed (correctly) that operation A failed. This could be because the network was disrupted between the two operations. Now when the user retries A, the 1st PUT itself will fail because it had succeeded the last time. So the user is deadlocked and can never perform operation A again without changing some parameter. Now with the recovery mechanism in place Core will check if the PUT failure is due to the data being already present and if so it will further check if we are the owner. If all conditions are satisfied the code will declare that the current PUT in that composite operation has succeeded and will proceed to the next PUT and finally return success for the entire operation A.

Core also provides facility to extract stats on number of GETs/PUTs/POSTs/DELETEs performed.

The team have also completed other refactors relating to Clippy errors and general code hygiene.

##Routing, Vaults: Adam, Andreas (tl), Fraser & Qi.

We have reimplemented caching, made the chunk store directory configurable and improved the reproducibility of our unit tests. We are also continuing our spring cleaning: reorganising our code and updating some neglected parts of the documentation to get everything into a better shape for future extensions.

##Client : Krishna (tl), Scott & Shankar.

The tasks which were planned as a part of implementing v0.5 of launcher have now been completed and we are currently testing these updates now. The NFS API has been refactored with support for streaming. Encryption and base64 conversions between an app and launcher have been removed along with a few other minor updates. The demo application has also been updated to work with this latest launcher version.

We have not made any UI or UX improvements in this version because @scott is working on a new UI (mockups will be shared to the community soon). Since there is a new UI planned, spending time improving UX on the present version would not be the most efficient or effective use of limited resources within the current Client team. The new UI will introduce new features such as launcher analytics and user statistics.

For this coming week, we will release the v0.5 of launcher and continue working on the new UI for launcher.


We hope you enjoy TEST 5 this week and thanks again for your continued support!

53 Likes

Your report is excellent as always, it seems that incremental improvement is gaining momentum and the final product will be worth waiting for, well done.

13 Likes

Great stuff as always Ross. Enjoy your time off! you’ve earnt it :smiley:

8 Likes

how did you read it that fast?! :open_mouth: *reading* :nerd:

ps: very cool =) i’m looking forward to test5 =) my server is ready to rumble :smiley:

6 Likes

Always good to see these updates :smiley:

What does that suggest?

3 Likes

I didn´t expect to see you @Ross. Thought you already took your time off?

5 Likes

It is the mechanism for the apps writing to launcher, not network streaming yet :wink:

10 Likes

What are those issues?

2 Likes

Withdrawals, Nick has posted on that thread about them.

3 Likes

This is the thread David mentioned: Yuanbao maid/cny pair withdrawal issue. Cheers.

5 Likes

Good progress as usual! Seems the networking layers are mostly done and that once these data persistence tests are proven successful that not much more will follow until we start sending each other messages? Hope I’m not getting too ahead but I think I see a light in the far off distance. :sunny:

4 Likes

Still around for a few days yet :smiley:

15 Likes

Great to read this as always – Tuesday mornings have become a highlight of my week. :relieved: And imo: take all the time you need. We’ve recently seen what happens with rushed, improperly tested code (DOH!), which just highlights the importance of the care you guys and gals are taking to make sure the foundations are solid.

11 Likes

Excellent progress it seems. Can’t wait for test 5 :slight_smile: Hope these well written updates will continue in Ross his absence. Enjoy @Ross and hope to welcome you back here after your time off! :beach:

3 Likes

Workaholic :stuck_out_tongue: thanks Maidsafe team for another great update…

Love to start fooling around with test5 :stuck_out_tongue:

P.s. @Maidsafe Mozilla will almost release am preview of it’s Servo browser built with Rust.

6 Likes

Woot test net 5 tomorrow. When do you think the new Launcher 0.5 will be out? Good to see you @Ross.

2 Likes

Looking forward to the next test net, I hear tfl are interested…

Yup, I’d already found it. Thanks anyway!! :slight_smile:

1 Like

I thought you left, ross?

I guess he meant 30th June

3 Likes