Clojurians Log v2
Clojure programming
Channels
Apps
malli
2025-05-18T06:14:25.335539Z
Hi folks, I've got a question about trying to use lazy-cat with malli. Here's a complete example:
(require
'[malli.core :as m]
'[malli.instrument :as mi])
(defn numbers [n]
(lazy-cat (repeat n n) (numbers n)))
(m/=> numbers [:=> [:cat :int] [:sequential :int]])
;(mi/instrument!)
(take 3 (numbers 10))mi/instrument! and re-run the example, I get a stack overflow. It seems like instrumentation causes the infinite seq to be fully realized. Is this a known issue or do I need to write the spec differently? 2025-05-19T13:03:17.546779Z
I guess it makes sense since the output would need to be fully realized to ensure everything is valid. The take happens downstream of the spec validation.
opqdonut 2025-05-19T13:05:38.450829Z
yep
2025-05-19T23:12:44.587509Z
Use :every for potentially infinite collections. https://github.com/metosin/malli?tab=readme-ov-file#seqable-schemas
2025-05-19T23:13:23.234809Z
It checks a bounded number of elements (100 by default IIRC).
2025-05-19T23:14:02.602999Z
Nice!!!
2025-05-19T23:14:06.098049Z
Thanks!
😁 1
2025-05-19T23:19:50.335049Z
It doesn't require sequential? tho fwiw. You might need to use [:and [:every :int] sequential?] if you care about that (relevant for generators).
👍 1
opqdonut 2025-05-20T04:51:05.276749Z
thanks, I didn't know about :every 😅
opqdonut 2025-05-19T05:36:45.423029Z
I don't think malli has anything for lazy validation of lazy sequences. I think you'll just need to skip instrumenting that function.
👍 1