I had first-hand experience with the way java developers avoid XML config. In my early days I wrote some javabean mapper framework but it wasn't liked by my peers becaused it was driven by XML. Back then EJB, Struts, and many web/enterprise frameworks used XML for config, so it was a no brainer to continue the trend. However the DX of XML in Java wasn't that great I have to admit. This is what made Spring so successful with its annotation-driven approach to config: you can annotate directly in the code, no need to manually create or edit XML files.

Clojure does not have this issue at all: XML easily translate into clojure data structures, and vice-versa, and from there you no longer feel any pain. I find it ironic that for a language in which data is code and code is data, you need to revert to java to integrate with annotations which is metadata added to code.

I guess the fundamental question is: do we want clojure to develop a parallel story for enterprise development, or do we want to improve clojure for better integration with enterprise frameworks. Back to slipway: what do you think can be done on the clojure side to make the slipway integration story easier to reproduce for other enterprise frameworks? I doubt the JDK will change for us so best is for us to adapt as much as possible (despite such conclusion [0]).

[0] https://blog.jakubholy.net/translating-enterprise-spring-app...