IRC channel logs
2024-03-21.log
back to list of logs
<nanomonkey>Sorry Zarutian_iPad for not leaving correctly...got pulled away unexpectedly. <nanomonkey>dthompson, I was able to replace iota, and the file compiles cleanly, but I now get the following error at the browser console: "Uncaught (in promise) RuntimeError: unreachable executed" <nanomonkey>Is that a known failure mode, do I need a more recent reflect.js file, or have I borked something in my js file within Scheme.load_main, or that gets passed to it. <Zarutian_iPad>so cryptography is mainly about the confidentiality, authenticity, and integrity of data. <Zarutian_iPad>ocaps are more about causal access/influence to objects that live in particular places <nanomonkey>Yes, thanks I think I understand, I'm just trying to wrap my head around how it is typically enforced. What makes an object reference unforgeable, and how can state change transaction be transmitted over a decentralized environment. I need to delve into the code on serialization, I'm guessing. <Zarutian_iPad>inside an runtime on one machine, the same mechanism that ensures memory safety <Zarutian_iPad>between machines though you can only use either big unguessable bitstrings (swissnums) or small integers that are used as an index into an list of references the connection/context they are used on/in has (c-llist) <nanomonkey>I'm not sure I'm familiar with c-list*, and DDG isn't being much help. I assume swissnumbers are content addressable hashes or similar, which once you know about, are hard to revoke. Unless there is a lookup table of sorts, but that index would also need to be transmitted to the group that has access. <nanomonkey>Anywho...this is all curious. Trying to get a sense of gotchas/shortcomings and how the capability model affects how work is done. <Zarutian_iPad>c-list is just for a capabilities list. Think of it as an array of references like in java or js <nanomonkey>I still need to add the ability to scroll around and zoom in further, or enter in coordinates. But it's at a functional state. <nanomonkey>Also, thanks Zarutian_iPad, I've read that article. I was specifically asking about how Spritely Goblins was implementing capabilities, as there seems to be some variation on how capabilities are accomplished based upon the scope. In the case of a distributed network, where you don't have any guarentees on the hardware that the vats are running on, it would seem that you need to incorporate some encryption. <nanomonkey>There are some definite performance tunings I should add, and the mentioned ability to interact with it to render at different locations...but yeah. <cwebber>really impressed by how compactt this code is <nanomonkey>scheme makes things easy...but I mostly just used the psuedocode from wikepedia, and optimized the scaling calculations to the 320x240 pixels <nanomonkey>No problem, thank you both for Hoot and Goblins. You don't know how happy I am to see a lisp language project for decentralized social AND wasm. It's nice to have projects that you're interested in utilizing languages that make you happy. <nanomonkey>Thanks Zarutian_iPad, that's an excellent article. I was more specifically asking about how Spritely Goblins was implementing capabilities, as there seems to be some variation on how capabilities are accomplished based upon the scope. In the case of a distributed network, where you don't have any guarentees on the hardware that the vats are running on, it would seem that you need to incorporate some encryption <nanomonkey>For example, the "revoker" addressed in that article would be difficult to implemnt once you've jumped from one virtual machine to a multitude of nodes, each running their own client version. <nanomonkey>Specially if you're using a distributed hash table or gossip protocol to distribute entities. Coordinating garbage collection is one thing, but removing capabilites seems a step up. I'm really curious what tactics have been thought out. <nanomonkey>That name (una-something) sounds familiar, I believe I've come across that. Refresh my memory if you can. <nanomonkey>Anyways, I don't want to add to the Spritely teams load, I was just wondering if there were already documents covering how they are implementing some of the technical details. I will look through the repo for more info. <nanomonkey>Oh, no I haven't digested that article yet. Thanks Zarutian_iPad! <nanomonkey>That's a fascinating article...but now I have off topic questions about NAT traversals! Ah well, I should take my dog for a walk and get some sun. Thanks again. <Zarutian_iPad>sneek, tell nanomonkey well first version of the captp for goblins is over the Tor network to not have to deal with NAT punching <sneek>nanomonkey, Zarutian_iPad says: well first version of the captp for goblins is over the Tor network to not have to deal with NAT punching <Zarutian_iPad>sneek, tell oyster_smasher5000 spurning hvort að þetta virkar rétt eður ei. <sneek>oyster_smasher5000, Zarutian_iPad says: spurning hvort að þetta virkar rétt eður ei. 