hdkdicnsnjes

Real

Isn’t this just dunning kruger? Try and build something substantial for your gene science thing rather than some notebook script or CRUD app and see what complexity you run into.

Write a macro DSL in Clojure for it then throw it up as a library.

To get to a higher level of abstraction than the general purpose languages you have to get more specific, maybe someone just hasn’t done it yet for your use case.

Does SQL or regex as a language rot? I’d say the abstraction above using SQL or regex to implement it rots, because it’s not formalised as a language.

Does the implementation leak through things like JSON or a EBNF grammar? Performance can be a concern but the implementation doesn’t leak through the language unlike a library api. If you have access to the compiler you can adjust the implementation to get whatever performance you’d like without affecting the abstraction.

Edit: JSON is a perfect example of resistance to rot and a good abstraction. You will always be able to read it in any given language because it is a language and not some library. Non DSL code should be considered as compiler output and that is what 99.9% of software devs spend their time writing. They mock people who would write assembly then go and create some fucked type hierarchy in Haskell for some business problem and call it a good abstraction. The complexity just multiplies until we end up with the current state of the industry.

Ah shit, right, it’s probably for the best lisp isn’t mainstream.

Imagine the software industry if lisp was mainstream.

The nuts and bolts will be embeddable DSL’s, they are the only code which does not rot. Put them in a hierarchy with one compiling down into another or several and you have your building.