Spritely's NLNet grant: Interface Discovery for Distributed Systems
I've been putting off making this blogpost for a while because I kept thinking, "I should wait to do it until I finish making some sort of website for Spritely and make a blogpost there!" Which, in a sense is a completely reasonable thought because right now Spritely's only "website" is a loose collection of repositories, but I'd like something that provides a greater narrative for what Spritely is trying to accomplish. But that also kind of feels like a distraction (or maybe I should just make a very minimal website) when there's something important to announce... so I'm just doing it here (where I've been making all the other Spritely posts so far anyway).
Spritely is an NLnet (in conjunction with the European Commision / Next Generation Internet initative) grant recipient! Specifically, we have received a grant for "Interface Discovery for Distributed Systems"! I'll be implementing the work alongside Serge Wroclawski.
There are two interesting sub-phrases there: "Interface Discovery" and "Distributed Systems". Regarding "distributed systems", we should really say "mutually suspicious open-world distributed systems". Those extra words change some of the requirements; we have to assume we'll be told about things we don't understand, and we have to assume that many objects we interact with may be opaque to us... they might lie about what kind of thing they are.
Choosing how to name interfaces then directly ties into something I wrote about here more recently, namely content addressed vocabulary.
I wrote more ideas and details about the interfaces ideas email to cap-talk so you can read more there if you like... but I think more details about the interfaces thoughts than that can wait until we publish a report about it (and publishing a report is baked into the grant).
The other interesting bit though is the "distributed" aspect; in order to handle distributed computation and object interaction, we need to correctly design our protocols. Thankfully there is a lot of good prior art to work from, usually some variant of "CapTP" (Capability Transport Protocol), as implemented in its original form by E, taking on a bit of a different form in the Waterken project, adapted in Cap'N Proto, as well as with the new work happening over at Agoric. Each of these variants of the core CapTP ideas have tried to tackle some different use cases, and Goblins has its own needs to be covered. Is there a possibility of convergence? Possibly... I am trying to understand the work of and communicate with the folks over at Agoric but I think it's a bit too early to be conclusive about anything. Regardless, it'll be a major milestone once Spritely Goblins is able to actually live up to its promise of distributed computation, and work on this is basically the next step to proceed on.
When I first announced Spritely about a year and a half ago I included a section that said "Who's going to pay for all this?" to which I then said, "I don't really have a funding plan, so I guess this is kind of a non-answer. However, I do have a Patreon account you could donate to." To be honest, I was fairly nervous about it... so I want to express my sincere and direct appreciation to NLnet alongside the European Commission / Next Generation Internet Initiative, along with Samsung Stack Zero, and all the folks donating on Patreon and Liberapay. With all the above, and especially the new grant from NLnet, I should have enough funding to continue working on Spritely through a large portion of 2021. I am determined to make good on the support I've received, and am looking forward to put out more interesting demonstrations of this technology over the next few months.