I have it reproducible, every time doing:

rm -rf ~/.m2/repository/org/bytedeco/opencv* && clj

{:deps {org.deeplearning4j/deeplearning4j-nlp {:mvn/version "1.0.0-M2.1"}}}

Clojure CLI version 1.11.1.1200

workaround is, as before:

clj -Sthreads 1

Also check with -Sforce to make sure you have a fresh classpath computation, but please put on Ask Clojure and I will take a look when I have a chance, might not be today

Done: https://ask.clojure.org/index.php/12730/error-could-acquire-write-lock-artifact-org-bytedeco-opencv

what would the recommended type be?

there is a suggestion for application/edn

I will talk to Rich and look into it again

thanks

Hello. If you want to greet people and chat about random stuff, please do that in #off-topic. This channel is reserved for discussions and questions about #clojure :)

Or #C0218CKTVEK

the list at the top of the Compiler is probably mostly that list - https://github.com/clojure/clojure/blob/master/src/jvm/clojure/lang/Compiler.java#L40

oh of course, thanks

well, particularly https://github.com/clojure/clojure/blob/master/src/jvm/clojure/lang/Compiler.java#L105

https://clojuredocs.org/search?q=diff

Yeah, I just found it. Thank you!

Not reloading the NS with *env* fixes the issue, but I would still like to understand where it came from in the first place.

Because it is a different var

Refresh blows away the existing vars so you get a new var with the same name

But previously compiled code that may still be running is using the old var

But that defmethod is the "old code" that apparently sees the old var - since it's in the same NS, shouldn't it start seeing the new var right away? And defonce doesn't prevent the new var from appearing, right? `But that

Depends on this that and the other

Which namespace the code is doesn't come into it, what comes into it is when the code that is running looked up the var by name

Which generally only happens when code is loaded (there is a static init that looks up vars by name and then stores the var it finds in a static field of most fn classes generated by the compiler)

So it is a temporal difference (code loaded at different times coexisting) not a spacial one (different namespaces or whatever)

I see, thanks. Still feels itchy because I haven't traced it with a debugger to the exact root cause, of course. :) But that can wait, given that not reloading that particular ns works just fine.

The refresh is an epoch, a line in time that for things to work correctly code shouldn't cross, but for some reason you are having code survive across the line

I mean, I dunno, could be something else, but seems most likely

Right, makes sense.

By returning a value do you mean an exit code or printed return?

printed return

I mean you can use -M -e to evaluate and print the result of any expression

Is that sufficient?

I got (ClassNotFoundException) after running clj -M -e com.namespace/foo

So I just wanna make sure we are on the same page, In com.namespace i have (def foo "bar") and I am just trying to return "bar"

With -e you'd still need to require the ns. -e "(requiring-resolve 'com.namespace/foo)"

(and you may need (deref ..) wrapped around that since requiring-resolve returns a Var)

Like so:

(~/clojure)-(!2007)-> clojure -Tnew app :name flava/dave
Creating project from org.corfield.new/app in dave

Mon Mar 06 14:02:46
(~/clojure)-(!2008)-> cd dave/

Mon Mar 06 14:02:49
(~/clojure/dave)-(!2009)-> vi src/flava/dave.clj

Mon Mar 06 14:03:03
(~/clojure/dave)-(!2010)-> cat src/flava/dave.clj
(ns flava.dave
  (:gen-class))

(def foo "bar")

(defn greet
  "Callable entry point to the application."
  [data]
  (println (str "Hello, " (or (:name data) "World") "!")))

(defn -main
  "I don't do a whole lot ... yet."
  [& args]
  (greet {:name (first args)}))

Mon Mar 06 14:03:06
(~/clojure/dave)-(!2011)-> clojure -M -e "(requiring-resolve 'flava.dave/foo)"
#'flava.dave/foo

Mon Mar 06 14:03:26
(~/clojure/dave)-(!2012)-> clojure -M -e "(deref (requiring-resolve 'flava.dave/foo))"
"bar"

Mon Mar 06 14:03:37
(~/clojure/dave)-(!2013)->

Oh cool!

thanks!

Store relationships as separate data. In other words, normalize the data.

@U2FRKM4TW's suggestion is a good one. Depending on the use case, you can also use things like zippers!

There are also in-memory dbs that can help.

Normalization is smart. I’m considering zippers but they feel very low level for the kind of operations I’m working with.

What kind of operations are you working with?

lol I’m working with source code traversal, which is the bread and butter of zippers, namely rewrite-clj. But rewrite-clj is a little slow and I find zippers to be hard to work with/conceptualize, so I’m exploring other options for fun.

in functional programming you usually use write a recursive transformation and use the call stack as the parent pointer

that can be a bit awkward sometimes of course, but for tree rewriting it definitely works

zippers add quite a bit of overhead. you can also do what those other languages do: use mutable references

I ran into a similar problem when trying to port a markdown parser from JS to Clojure

In analysing sports games, we originally had something that used a zipper. Currently experimenting with something that promises to be much faster based on a kind of doubly linked list of events allowing for lazily looking outwards in either direction from current context. That's an interesting point about normalisation. Will need to think about that.