Clojurians Log v2

native-image built is currently broken for some reason. I will look into this, @timok maybe you have some capacity, if not then it is fine. I see now why parts of it fail, I made the hitchhiker-tree optional as a dependency and it now is missing in the native pipeline it seems:

Compiling datahike.http.writer
Compiling datahike.impl.entity
Compiling datahike.index
Compiling datahike.index.hitchhiker-tree
Execution error (FileNotFoundException) at datahike.index.hitchhiker-tree.upsert/loading (upsert.cljc:1).
Could not locate hitchhiker/tree__init.class, hitchhiker/tree.clj or hitchhiker/tree.cljc on classpath.

Full report at:
/var/folders/g9/7rkt8rt1241bwwhd3_s8ndp40000gn/T/clojure-4946539122686929424.edn
Error while executing task: ni-cli

@whilo I'm free this weekend. I'll test all the changes and give feedback on Monday.

Awesome. I would also like to release https://github.com/replikativ/konserve-jdbc/ to get this backend out as well.

I can also take a look at this next https://github.com/replikativ/datahike/pull/755 and I will also need the auto-gc for some of the larger scale experiments I run.

I can check the CI... have to build it up on my fork first to not spam the main branch

Thanks for the clarification. Your work is fascinating. EAV/datalog with P2P sounds like a dream come true. And it does look like your simulation approach would work for the purposes of the aforementioned talk.

> distributed simulation based inference Do you mean inference based on the output of a distributed simulation? What would P2P/DHT bring to the table compared to the use of a distributed memory grid? Are you thinking of simulation to make systems more resilient, a bit like what this interesting talk suggests in order to make better architectures: https://youtu.be/D8qQUHrksrE

Just to clarify my previous comment, when viewing the talk I mentioned, it became clear to me that the kind of simulation it suggests is not just monkey testing the infra, it's simulating the environment outside the IS or even the business. It's closer to E2E testing but with scenario that are a bit more farfetched.

I mean probabilistic inference, which is a generalization of many other forms of inference (including logical ones). I wrote my PhD thesis on Bayesian inference and it covers this a little bit https://open.library.ubc.ca/soa/cIRcle/collections/ubctheses/24/items/1.0449997. No need to understand this technically though (the introduction tries to explain it intuitively). Coding models are in effect probabilistc inference models that are trained on massive data sets/tasks. The most challenging problems we face, climate change, reorganizing societies, etc. can be addressed through simulations and trying to infer ahead of time what is even worth trying (i.e. using the simulation for planning). Fuzz testing of code is a special case of doing sampling based inference on correctness predicates (tests).

I brought up p2p/dht because kabel could bootstrap connectivity inside out even under adversarial conditions, it is always something I thought about doing for replikativ, but didn't have the time to. Maybe I could give a coding assistant a spin and add routing tables and information, which then allows peers to explore and rewire a network.

It is not necessary for the simulation angle, but you generally want a resilient and flexible network substrate, I think.

I have only skimmed the talk, but yes, testing business models/ideas is in general the strategy and my goal.

Not just code.

It changes the perspective on what code is, probabilistic programming turns it into a model.

Very cool! So, it really would be nice to get rid of manual boilerplate when merging like

{:adds-in-branch
   (sort (d/q '[:find ?db-add ?e ?a ?v ?t
                :in $ $2 ?db-add
                :where
                [$ ?e ?a ?v ?t]
                [(not= :db/txInstant ?a)]
                (not [$2 ?e ?a ?v ?t])]
              branch-db base-db :db/add))

   :retracts-in-branch
   (sort (d/q '[:find ?db-add ?e ?a ?v ?t
                :in $ $2 ?db-add
                :where
                [$ ?e ?a ?v ?t]
                [(not= :db/txInstant ?a)]
                (not [$2 ?e ?a ?v ?t])]
              base-db branch-db :db/retract))

   :adds-in-main
   (sort (d/q '[:find ?db-add ?e ?a ?v ?t
                :in $ $2 ?db-add
                :where
                [$ ?e ?a ?v ?t]
                [(not= :db/txInstant ?a)]
                (not [$2 ?e ?a ?v ?t])]
              main-db base-db :db/add))
}