Fork me on GitHub

@dave.dixon just curious Re part of issue 371: if you had something like s/querydef would you still want ns kw names for defquery?


What is s/querydef?


Doesn’t exist. 🙂 Thinking it could be like s/fdef for queries


That's a good idea. But I'd still want keyword names. It's not a big deal, just a nice option to give more flexibility to DSL authors. I have it working in my local branch in CLJS, pretty minor change.


What do the vars end up being named? e.g. (defquery ::some-alias/my-spec ..) => #’rule-ns/ … ?


There are no vars. I'm not defining the queries as ns-level defs. They're grouped inside of a named object, and that symbol winds up holding the collection of productions that is provided to defsession.


Like most of your suggestions and changes I like that. Using defs/the ns as a kind of data structure is something I like sometimes and not others


Is that part of your proposal? If not how would you envision this working for users of Clara? Would it affect session reloading?


I'm putting together my own DSL that has those characteristics, not proposing to alter the existing rules DSL packaged with clara. I handle reloading in the implementation of my DSL. Since rules are just grouped in vars, it's pretty straightforward, don't have to deal with the subtleties of CLJS namespace reloading etc.