What’s up today? (Part 1)

We can buy Sponsored article:


Privacy. Security. Freedom

3 Likes

$11,500 I wonder how many think it’s worth it.

Not me. Certainly not yet.

7 Likes

Someone just recommended the following to me… They use them to get attention in the mainstream media for their company . As well as pushing stories out to the media sometimes the general media will approach them for stories in topic areas - so they could get requests for crypto stories and push our story.

London, Brussels and Amsterdam based.

… Or maybe keep options like this for the stable Fleming release that also has GUI tools too. Not everyone can use a Linux terminal and perhaps we should wait till we have point and click applications.

4 Likes

This. From a media point of view there’s currently nothing to see, no customer stories, and the network has yet to prove itself. Going too early would likely do more harm than good.

16 Likes

Cannot emphasize the enough. We have some key points already from the testnet so far and we need to get these in place in the next few weeks. There is also a bunch of UX we can now focus on and get right that will help no end.

This testnet is slow, cumbersome and with a horrible UX. That was the side effect of reaching the goal of proving the network. The point so far are

We also have some smaller issues that we will try and attend to, but next iteration as soon as possible. I don’t want to wait weeks between these.

We need to be extremely focussed in the next few weeks and get the above done quickly. None of them are design work, just small fixes, but these are vital.

On another area upload speed may get some attention, right now we

  • Get Storecost
  • Pay for store
  • Get receipt
  • Actually store

For each chunk in a file, so a 1Gb is gonna be a thousand payments, 1000 receipts and 1000 stores, leading to 1000 credits to section wallets. That’s a lot :wink:

38 Likes

I suspect we can do many things,

For private data, perhaps what we have is the best we can do? For public data, we can pay for the data map. So the data map is packaged with the receipt and we pay once per file.

Going further may be troublesome, but we will see. It may be a lot of hassle for the network, but upload once and that is forever, so may not be too bad.

At the moment the stress of this is great though. The network should work 100% even with this massive stress test, until it does then I think we should perhaps not even optimise this part?

12 Likes

Would be nice to see CLI fail not hang… so, it’s more obvious connection to network is responsive or not. A progress indicator on uploads, would help later too.

7 Likes

Plenty of low hanging UX fruit.

Sorry, unintended pun.

8 Likes

Yes - as above, most important is that it’s reliable. Assurance first and optimisation/speed can wait. Time is not important.

8 Likes

How does the network handle requests?.. there’s obviously assurance about data and how nodes are aligned but what of requests to a node that disappears?.. are requests duplicated and handled as if they were short-term data… or is the onus on the client to repeat the request?.. I wonder if nodes that are churning are not responsive and nodes are dropping, if that accounts for some of the hanging requests… waiting for a reply that will now not be coming… or is there a fix to ensure that every request gets a response, expecting the client will see one response.

1 Like

Could this be fixed by the client doing the self encryption send the XOR addresses in batches (variable number) upto say 128 and the network charges for the batch and provides a signed “credit” that the client sends with the chunk store request and the section can verify the chunk has already been paid for by checking the “credit” is signed correctly.

This way there is only one payment for the batch and any file less than 128 chunks is paid for in one transaction.

This also bypasses the gaming that would be easily possible if simple pre-purchase of storage was done. This pre-payment only allows those particular chunks to be attempted to be stored and so no issue if done in 5 minutes, 5 hours or 5 days

7 Likes

Yes, there is a bit more to it though. This is certainly one area for us to consider optimising and batch payments are surely going to be the way. We pay at source for chunks stored at destination. The src section give SectionAuthority there, but at destination section they really should sign again for NetworkAuthority.

I think we need to define SectionAuthority as in a section and NetworkAuthority from the network but can be found in the local SectionChain.

What this means is any section can trust another, by getting ProofChains and that is fine in the dynamic live network. However when checking data in a Section it would be good that the section members (or ancestors) had signed. That saves us trying to get every SectionChain from every section (i.e. the whole Dag) when checking data for NetworkAutority in a section.

So a wee bit more subtle, but also batch signing is possible here. Good thing is there are a lot of ways to do this more effectively.

6 Likes

I got another one.

I was trying to sign in and i got the following error:

Passphrase: 
Password: 
Sending action request to authd to unlock the Safe...
Error: AuthdClientError: [Error] ClientError - Response not received: read error: connection closed: timed out

I’ve been able to produce it consistently on my machine. This may be because of network congestion

This problem always gets me thinking of GPU’s having very fast limited VRAM close to the cores, to always be able to feed the cores for 30-60s, so that there is no bottleneck, as it would be if the GPU would always need to ask the SSD or HDD for 1MB every second of texture shading.

What would be needed to get batch payment of chunks to work, would it be very difficult or time consuming to implement in existing code?

3 Likes

The fact that the testnet is up now—and working pretty ok for the most part—feels amazing. Now we just have to make it good!

11 Likes

:100: The hard bit’s done now, the proof is now happening. Test run 1 was very successful based on our feedback, nothing alarming at all. No OMG moments or similar. So we are in a good place here. We just need to keep the focus of the last few weeks for a few weeks more. Then it all changes and folk will see Safe the project and I think it will surprise a great many people.

We are small enough and tough enough to move at a pace then, we can do a lot fast and will need to, but this first step was always the dream.

29 Likes

@Dimitar @JimCollinson
" The permaweb was developed by Arweave with technology that is based on blockchains and distributed computing, the computer science innovation behind digital currencies such as Bitcoin."

1 Like

They are not our competitors, the hardware requirements to be a node are high. The struggle for dominance will be between projects that work on ordinary hardware. Therefore, the only competitor we can have is another Safe Network.


Privacy. Security. Freedom

2 Likes
1 Like