Using `cl-defstruct` structure with `cl-macrolet-symbol` for very convenient read/write access to structure slots by name.

convenient-struct-handles.el

(defmacro with-convenient-slots (slots obj &rest body)
  "Convenience wrapper for read/write access to struct data.
Bind slot names SLOTS (a list) within the `my-struct'
record OBJ, and execute BODY."
  `(cl-symbol-macrolet
       ,(cl-loop for slot in slots
                 collect `(,slot (,(intern (concat "my-struct-"
						   (symbol-name slot)))
				  ,obj)))
     ,@body))

;; Example usage
(cl-defstruct my-struct "A struct" foo bar)
(setq myobj (make-my-struct :foo 1 :bar 2))
(with-convenient-slots (foo bar) myobj
  (setq foo (+ 123 bar)
	bar (+ 456 foo)))

convenient-struct-orig.el

;; Original version: more general, but less efficient

(defmacro with-convenient-slots (slots obj &rest body)
  "Convenience wrapper for read/write access to struct data.
Bind slot names SLOTS (a list) within the `some-data-type' 
record OBJ, and execute BODY."
  (declare (indent 2))
  `(cl-symbol-macrolet
       ,(cl-loop for slot in slots
                 collect `(,slot (cl-struct-slot-value
				  'my-struct
				  ',slot ,obj)))
     ,@body))

The code cleanup for reasonable size functions with well named slots makes it well worth it I think.

It depends on a variety of factors. Another cost is to code readability: when other programmers are reading your code, the more non-standard macros used in the code, the heavier the cognitive load is to understand the code.

What approaches do you use to keep structs under control?

I don't suppose that I use any, really, other than to try to keep the package name prefix and slot names short. The package I use them in the most thoroughly is https://github.com/alphapapa/ement.el. There are a few places where having a place-setting macro would be helpful, but it doesn't seem to be a big deal. I also use highlight-function-calls-mode, which shows function names (including accessors) in a different face, which helps visually identify them and confirm that they are typed correctly, and I would miss that if I used such a macro instead.

RE pcase-let, I didn't see your post but it's an interesting idea: pcase-symbol-macrolet?

Hm, maybe so, or maybe pcase-letf, to coincide with cl-letf. But I'm doubtful that such a macro would be accepted into Emacs core, anyway. It would probably take some wrangling!