I have over 25 Years of various Tech experience and would love to be part of this emancipating project, however when I go download the CLI so I can upload some test data I am getting a 404 Not found on Github
Is there a mirror somewhere to download the CLI or is there a reason it is not currently available?
How familiar are you with rust? If you can install rust, the best way forward would be to clone safe_network and build it on your own box
cargo build --release
Are you using Windows or linux? Both will work but most of us are using some form of linux, predominantly Ubuntu/Debian so Qs from Ubuntu users will generally tend to get the snappiest answers.
There is a lot can be learned from running a local network on your own box, once you have the basics installed
safe node run-baby-fleming will set up a basic minimal network where you can put and get files and directories and experiment with keys, wallets and DBCs to get the hang of the basics.
Never forget āThe only stupid question was the one you felt embarassed to askā
He doesnāt need to build to install the latest release unless thereās a bug. Iām busy doing human things so canāt check but I suggest @Independent_Champion:
find setup instructions on this forum, such as for a community test
use those to install the CLI (because maybe GitHub README is out of date?)
run a local test network (see āsafe run baby flemingā)
Shhhh!!! I know that, you know thatā¦ @Independent_Champion says he has 25yrs exp, Iām trying to find folk that will be happy soon to dive into building with --statemap and other such goodies - weāll show him the dead easy stuff if he cant handle the fairly straightforward stuff </cunning plan>
I am not familiar with Rust but have been reading up about it and find it fascinating and I am willing to learn, my ambition from many years ago is to deploy cheap and energy efficient data centres in Africa, most of my tech experience from early 2000ās onwards is based on running my own Satellite Telecommunications business mainly focussing on emerging markets, I have experience running Ubuntu as an Apache web server from many years back so getting back on Ubuntu wonāt be a problem but the quest was always to have these cheap indpendent data centres in Africa, Middle East and South America connected to both satellite and Independent local LAN links. I believe this could be the alternative answer im looking for, as I understand it the āfarmsā can now be a bunch of peer to peer mini PCās instead of these monster data centres that currently make up the internet, so I have ordered a cheap Mini PC with Ubuntu pre installed and I am going to give it a go, I hope to make a kind of off the shelf farming PC form a commercial point of view.
The main reason for me doing this was to improve the latency of the internet for these markets who mostly have to connect to Europe with a round trip ping of near 150ms for even a fibre connection, my question is that will the Safenetwork know who is the nearest node geographically and can you run a seperate āofflineā Safe Network via the Local host like you can with Apache Web Server?
Theres a few on here interested in mesh networks, that might fit well
Lots of discussion in the forum archives on that
No, it will know its closest neighbours in XOR space only and form a section with them initially. Others can answer this better for groups of nodes on āislandsā @neo?
Yes baby-fleming is a minimal SAFE network on localhost BUT there is no web browser at the moment. There was a web browser and there will be again soon but the development priority is to get a basic functional network launched. We will have a pretty GUI again soon though, search through posts by @JimCollinson to see examples. Im not on the dev team but I think its a case of If the client code stands still long enough then the browser code can catch up very quickly.
Not using the current design and unlikely to change
āofflineā Safe Network can be run like @Southside suggests.
But also Iād expect when the software is past launch then you can setup your own āglobalā network by using a different ID for the network. This would be like the safe network in that you have to have your initial seed nodes (which can be retired later on) and nodes can join using *your* network ID.
This would be a separate running network and would not be talking to the Safe network at all. Then I am sure that a small modification would be able to restrict the IP addresses to set ranges (your island) so that only your island can join and not some joe blogs from across the world.
Of course if your āislandā is a intra-net then that IP restriction would not be needed
This is a little dissapointing, because this random lottery nature of selecting the node available may connect them to a node that is on the other side of the planet and just by the laws of physics of distance, even at the speed of light the response and latency will make the connection slow, as someone who is at eternal war with latency on Satellite Networks for a living I can tell you those milliseconds make all the difference.
This is great news, it will be a way of mitigating the latency between nodes as well I suppose by building mini networks covering a close nit geographical area, I canāt wait to get my Ubuntu box and try it out.
Yeah I knew this would not fill you with joy. It got me thinking though - would sufficient churn eventually make a section tend to choose new members with lower latency and hence (effective) geographical closeness?
As alluded above would there be an advantage to starting a local SAFE network, building it up to a few sections and then connecting to the main worldwide network?
This isnāt possible and if it were, would probably provide a way to attack the network.
I havenāt kept up with the latest architecture so these questions should be re-examined. Earlier designs were intended to mitigate latency (for example nodes used to compete to be first to deliver each chunk) but Iām not sure if the latest also does so.
@JPL do you know if the Primer is up to date in this area and if it could help understand latency?
Further random out loud thinkingā¦ @Josh is ATM testing from both NYC and LON datacentres to find out where the no-load spikes are coming from. All testnets so far (please correct if I am wrong) have started with a bunch of seed nodes in a single datacentre and then a few nodes from random geographical locations - but mostly Florida and Europe. When we get a critical mass of testers (and budget) we should aim to have only a couple of seed nodes in the datacentre(s) and punters worldwide with suitably specced machines to fill out the first section(s) - or launch with a couple of nodes each in (almost) every AWS or DO region.
Will it be cripplingly slow?
Will we still see these no-load spikes?
How soon would it take for a popular download to be cached locally?
How would we know other than suddenly our favourite cock(erel) pic arrives a few ms faster than previously?