hmm. Proxy is also a no go. It modifies the class, but not the class instance.

Could you use deftype? Implement Closeable. Just use a plain function (close-xxx)? If this is something used as part of configuration, could you use Integrant (or similar) and do whatever you need in the halt method?

You should probably wrap the object and delegate all method calls except close to the underlying object. Is SomeType an interface or a concrete class?

> Is SomeType an interface or a concrete class? A concrete class.

Then can you use proxy to instantiate an object of SomeType and make it delegate all the stuff except close to the wrapped object?

Isn’t proxy wraps a class instead of an instance?

With proxy you make an instance of SomeType and implement it in the way that it passes all the calls to the wrapped object except the close method call.

I wonder how you wrap an object with proxy.

https://github.com/jumarko/clojure-experiments/pull/28

Btw. I realized you don't need to implement any other methods except close - see https://github.com/jumarko/clojure-experiments/pull/29/files#diff-813208f4514480f7a8f4fc6259a5ee673db0c2c7700c06ca3369ab7d5af09c1cR88

So a minimal example would be something like:

(let [filename "empty.zip"
      original-zip (ZipFile. filename)
      proxy-zip (proxy [ZipFile] [filename]
                  ;; overriding close method
                  (close [] (println "Ignored!")))]
  [[(.getName proxy-zip) (.getName original-zip)]
   ;; check the REPL - you will see "Ignored!" printed one time
   [(.close proxy-zip) (.close original-zip)]])

@UGC0NEP4Y does this approach work for you or you hit some issues?

@U06BE1L6T Thanks for the code. Take the Zip class for example, if https://github.com/jumarko/clojure-experiments/blame/master/src/clojure_experiments/java.clj#L101 the filename is not given, so the only thing accessible original-zip, what can we do?

How can the filename (or similar information required by the constructor) be not given? If you didn't know that, you couldn't construct the original object.

> I have a java function from a 3rd party lib, e.g., (my-func) which returns an object of type SomeType. It’s hidden in the 3rd party lib inside the function my-func.

If you are not able to construct the object then I'm afraid this is not possible. In most cases, there will be a constructor which you can use and some kind of "getter" to get the information you need from the object. Is this is an open source library? It would be great if you could provide a code sample.

The problem was addressed to use another lib and the problem has gone. I was trying to understand the possibility.

Even clojure.lang.Numbers have boolean marked as the return value for lt and others

This is the general pattern for Clojure functions, when you expect a boolean it will specify "non-nil" instead of "true", because this allows the contract to not be broken if the actual return value is changed down the line. In the specific case of the math comparison operators, there's not likely to be any change there, but many functions that might conceptually return a boolean return an object because of the pattern of returning "the most useful thing". e.g. or and and in other languages might return only a boolean, but in Clojure they return more valuable values.

Thanks!

No problem!

Yes, for or, and et al I understand this completely, but for math ops it was not clear. 🙂

Rich Hickey has objections to pattern matching in the "Simple Made Easy" talk.

I think it's less about readability and more about coupling. I think that any time you have a case/cond/etc you create a branching code structure that is closed for extension. Pattern matching (in a lot of languages) is often a dispatch-on-type kind of operation, and often dispatching on types that you don't own. This sounds a lot like polymorphism and could perhaps be better served using protocols or multimethods. I've not done a lot of erlang, so I don't feel in a position to comment on that vs something like core.match ... but I think the point is more about questioning if you're using the right language construct for the job rather than a hard and fast rule 😉

https://gist.github.com/reborg/dc8b0c96c397a56668905e2767fd697f#why-no-pattern-matching

@U06BE1L6T, thanks, this should be pinned basically everywhere haha

@U0P0TMEFJ, makes sense, thank you

Still, if you want to at least see what pattern matching could bring you can use either https://git.sr.ht/~technomancy/fnl-match or https://github.com/clojure/core.match 🙂

And #meander

and https://github.com/xapix-io/matchete 🙂 there are many more or less active libraries providing pattern-matching abilities

While not a fan of pattern matching, it has its uses. Wrangling HTML, for example

As far as I can gather, tags can be moved: https://stackoverflow.com/a/20076601/202538

Yes, tags can be moved, but this is actually the reason the sha is included in the first place. If we only had the tag then we would have no way of knowing if the tag was moved. With the sha included, if the tag is moved then the build will fail, and you as the developer have the opportunity to then go browse your dependency to determine the most appropriate version to use.

Great, so the sha is mandatory, right?

Yes. Adding the tag however allows you to use a short sha instead of the full one.

Ah, nice! Thanks for the explanation 🙂

https://clojure.org/reference/deps_and_cli#_git here is a full reference for those options

The fact that it goes and checks the tag against the sha on every single build is actually annoying because it means you can never build offline: https://clojure.atlassian.net/browse/TDEPS-223

if that's a concern to you at all, I suggest using full the sha and no tag instead

tag = semantic meaning for humans sha = specific commit for comparison

tag + short sha gives you semantics plus verification against a moved tag

if tag and short sha don't point to the same peeled commit, you'll get an error

I believe the same even happens with tag + full sha

yes, correct

Static methods are called differently: (JShell/create).

I think so. It prints for me.

what's your dev tooling?

just inf-clojure with a custom print fn in the repl that basically tries to prevent massive output, turns out to be related to that but not sure how yet

can you post your clojure.main/repl invocation here?

in inf-clojure I’m seeing:

database-test=> (take 10
                      (eduction
                        (map println)
                        (range 100)))
0
1
2
3
4
...
31
32
(nil nil ... nil)

(clojure.main/repl
  :print (fn [x]
           (binding [*print-length* 100
                     *print-level*  20]
             (if (> (count (pr-str x)) 3e6)
               (log/warn "<string too big>")
               (pprint/pprint x)))))

yeah it's definitely that if statement that's causing it, a little confused as to why

yeah i get the same with your repl. You should add :read clojure.core.server/repl-read so you can quit it with :repl/quit

it works fine for me to just create a new one, but thanks for the tip that's probably better

it is because of the call to pr-str

I’m using

(clojure.main/repl
 :print (fn [x]
          (tap> [:print-called x])
          (try
            (if (> (count (pr-str x)) 10)
              (log/warn "<string too big>")
              (pprint/pprint x))
            (catch Exception e (tap> "caught an exception") (tap> e))))
  :read   server/repl-read
  :prompt (fn [] (printf "sub %s=> " (str *ns*))))

pr-str is realizing the eduction inside a binding of *out*

ah good catch @U0NCTKEV8. that has with-out-str in it.

user=> (pr-str
 (take 10
       (eduction
        (map println)
        (range 100000))))
"(0\n1\n2\n3\n4\n5\n6\n7\n8\n9\n10\n11\n12\n13\n14\n15\n16\n17\n18\n19\n20\n21\n22\n23\n24\n25\n26\n27\n28\n29\n30\n31\n32\nnil nil nil nil nil nil nil nil nil nil)"
user=>

interjection but how do I customize my REPL like that? Is there a walkthrough on it somewhere?

then because of the call to take, the eduction is wrapped in a lazy seq, which caches the results of the compuation, so the second time through with pprint you just get a seq of nils

if you are using nrepl i think you are out of luck. If you are using a regular repl then the forms i’ve posted here will work fine for you. the server alias is clojure.core.server Here’s my standard repl:

(clojure.main/repl
 :prompt (fn [] (printf "%s=> " (peek (str/split (str *ns*) #"\."))))
 :eval (fn [f] (binding [clojure.test/*test-out* *out*] (eval f)))
 :read server/repl-read
 :print fipp/pprint)

regular REPL as in started with clj or clojure and not through Cider or the like

> pr-str is realizing the eduction inside a binding of out aha, thanks!

> regular REPL as in started with clj or clojure and not through Cider or the like yeah then you can just call clojure.main/repl with the args described here: https://clojuredocs.org/clojure.main/repl

@U03QBKTVA0N yeah. CIDER connects through nrepl which can’t work with subrepls like this. If you try it you’ll see if asks for input from the minibuffer (i believe, been a while)

Also check out https://insideclojure.org/2020/02/11/custom-repl/

cool, well I'll explore my options. thanks! (excuse the off-topic noise) @U11BV7MTK @U064UGEUQ

searching for artifacts doesn't use repository settings from dependencies, so you'll need to configure any additional repositorie

yeah I found this https://clojure.atlassian.net/browse/TDEPS-46

> you'll need to configure any additional repositorie you mean that all users of the library will have to add it to their setup?

yes, same is true of any one using maven

damn, I guess I'll just deploy that jar on clojars under my group :man-shrugging:

thanks!

you can see the classes are from different "module loaders"

but why would that only happen sometimes?

in other words, it implements IFn, just not the particular one that's being asked about. depends on your classloader hierarchy / environment

dunno, what is your code doing?

the code is doing a filter over a map, some very simple code. We have some custom classloader stuff so I guess I need to go chase down why it's loading things in the wrong order or whatever. Thank you both very much!

Could this be having two Clojures loaded on two classloaders?

it is

Hi, welcome! may I suggest posting in the #beginners channel?

In the #beginners channel, people have opted in to help folks who are new to Clojure in depth. Questions in this channel tend to get answers that assume a fair bit of Clojure knowledge -- so you're going to get better help in #beginners

Thanks guys!!!

Welcome to Clojure, @U02PR896TMG!