Clojurians Log v2
asami 2023-01-31
When you call asami.core/transact (or asami.core/transact-async) then the argument can be:
• a seq of triples to be inserted
• a map
If it’s a map, then the accepted keys are:
• tx-data: a seq of :db/assert and :db/remove statements
• tx-triples: like the non-map version, a seq of triples to be inserted
• executor: a user supplied executor. Defaults to the Clojure Agent executor
• input-limit: limits the maximum number of triples that can be inserted in this transaction (I recommend never using this. It was a UI hack for Cisco)
• update-fn: a Graph update function.
You want to pass update-fn
This is a function that accepts a graph, and returns a graph. The Graph protocol is found in asami.graph/Graph
You can do a resolve-triple for something like [the-entity :count ?c] and then:
(graph-transact graph [[the-entity :count (inc c)]] [[the-entity :count c]])
So:
(asami.core/transact my-connection
{:update-fn
(fn [graph]
(let [c (ffirst (asami.graph/resolve-triple graph :my-entity :count '?c))]
(asami.graph/graph-transact graph [[:my-entity :count (inc c)]] [[:my-entity :count c]])))})Not pretty, I’m sorry
But possible
Thank you! "possible" is great!
When I get to SPARQL, then this is something I’ll have a query language to do for you. I just need more weekends. Well… weekends when the kids don’t need me
Hey 🙂 out of interest, have you seen that there's the basics of an "RDF" module in XT which reuses the RDF4J parser and compiles a small subset of SPARQL into XT's public query API? See https://github.com/xtdb/xtdb/blob/master/labs/rdf/src/xtdb/sparql.clj It's mainly been used for benchmarking with https://dsg.uwaterloo.ca/watdiv/ (etc.). Just thought I'd mention it just in case it helps for some inspiration or at least ideas about what not to do with your hypothetical weekends!
Thank you 🙂
Time for a little bit of personal philosophy: There is lots of work out there that I could be leveraging to help build out Asami faster. I really ought to. But I choose not to for a few reasons: • I try to keep the dependencies very minimal. This has proven to be very useful. • Each thing I implement myself is a learning experience. I appreciate this. • Most external systems are not Clojure/ClojureScript compatible. I’m also hoping to incorporate ClojureCLR support soon. It would be even cooler if I can make it work with ClojureDart or (eventually) Jank. I admit that there’s a certain amount of hubris/ego happening here. But it’s my hobby, so I’m allowed do it the way I’d like.
Also, I tend to be a little opinionated. It’s nice to have an environment where I get to express that
That last one is the most flexible though. I will often look back at something I wrote and realize that my position has evolved, and I wouldn’t do it like that anymore 🙂
I am completely on board with all of that! There's real beauty in pure Clojure.
This is why Raphael exists. Otherwise I’d have used riot from Jena
Years ago (like 12), I tried an RDF/XML parser, but the Clojure XML parser didn’t handle namespaces! I did try patching it, but I probably used non-idiomatic Clojure back then, and it was rejected. So I stopped. But it’s since been fixed, so I should try again.
Godspeed 🙂
Honestly, I like TTL a lot more. Now that it’s formally accepted as an RDF format, I feel like I’m justified in sticking to it 🙂
There are some government datasets in RDF/XML, so it’s still around
But I’m mostly inspired to consider RDF/XML because my mentor uses it
I author my OWL in TTL. She does it in RDF/XML. And honestly, she’s been working on it longer than I am (she worked on the DAML+OIL committee!)