Apparently @riddim had already found a workaround I didn’t think of, using a name as input for the key derivation algorithm and setting the Pointer version counter to the maximum value to make it immutable. Thanks @riddim!
Actually I think using a graphEntry, putting into it the address you want to point to and just writing that to the network might be better than burning a pointer/Scratchpad because the data type is immutable by design
Edit:
That graphEntry solution probably wasn’t possible with transactions because those were designed for the network currency… But the now existing graphEntry is technology agnostic and can be used for random use cases
And if the GraphEntry gets an additional output (a key not derived from the initial graphEntry but one just the owner knows) names could even be sold again… By giving that additional output priority if a graphEntry exists there… Obviously comes with the issue of every owner change adding an additional indirection… So not super cool…but it would work
It would be possible via a proposal I put up a while back:
Since Autonomi is already using libp2p, it would be able to add the feature and get some free support from the libp2p team.
Hey Guys. Been waiting to post something until I had something to show. I finally got around to making a video describing what I’m doing on Colony for content discovery. You can see it on my impossible futures page here: Impossible Futures
At a high level, the idea is that users post file metadata into a publicly readable scratchpad (or collection of them) and share the addresses to these scratchpads with their friends. Colony then recursively traverses through these metadata scratchpads to build up an index of everything it finds. Just a few hops and you have access to everything that people have indexed in this way. The more people use this strategy, whether or not they use Colony, the more resiliant and expansive the index becomes without having to have a server doing the indexing or a single page multiple people need to write to.
you made a base cap for your app!
well - then you need to link your friends scratchpads too for this to work - right?
…maybe first watching the video …
edit: i See you’re doing exactly this
Gotta have the merch ready .
Very nice! This is the kind of thing that autonomi will benefit from massively. Supercool!
If I were you, I’d just focus on Autonomi, no Docker/Web/Server crap If you want a web-tech GUI, think about Tauri perhaps? Also with Tauri 2 it’s already possible to release app for Android (iOS is planned also I think) .
Web API talking to local dweb REST server is also going to be a nice option.
Standard html and JavaScript, simple to create and use. I’m building a demo app right now using Scratchpads and it is a really good developer experience.
Agreed, MVP is stand alone app and that’s the core focus. Local AI agent for search will be the next feature to integrate into the core app, before a different GUI frontend. Its kind of ironic that I’m looking at doing a server/client thing at all, but I self host a lot of docker containers on my machine at home and kind of like it, but its one of those “I wish there was a simpler way”. If I can make a mobile app, that connects to a server instance through some REST API calls, then I think that’s all I really need. I don’t think fully mobile is a good long term solution as the datasets will become pretty large and the compute power to do search will be too much for a handheld device.
I didn’t stumble upon Tauri 2 when I was looking for GUIs, this looks really promising. My software engineer friend who will help me if I get to the next phase is a Javascript guy (wasn’t excited about rust) so this may be perfect. Thank you for the suggestion!
When I get there I definitely plan to look at what you’ve got going for dweb. Thank you for your development here, this is one of the crown jewels of community development we have on Autonomi.
That’s good to hear about scratchpads. My whole concept relies on that datatype working well