e.g. an endpoint like /current-user , the handler could be called twice per network request

Maybe it's your browser that duplicates the requests? CORS os favicon.ico? I think it's worth checking the network tab in your browser and check what and who emits those extra requests.

the browser definitely doesn't do it

In the network tab, I see just one request

(def cusess-handlers
  ["/current-user-sessions" {:get {:handler (fn [r]
                                              (swap! dbggg conj r)
                                              (response {:ok true})
                                              ;;(get-current-user-sessions-handler r)
                                              )}
                             :swagger {:tags ["sessions"]}}])

In this instance, I can see there's one (swap! dbggg conj r) per request... i.e. the @dbgggg atom increments by one request every request on the endpoint

But then if I flip to

{:handler (fn [r]
                                              (swap! dbggg conj r)
                                              ;;(response {:ok true})
                                              (get-current-user-sessions-handler r)
                                              )}

Maybe there is something in your router or middleware that performs the request twice.

makes sense... i think there could be some exception retry handling or something

> In the network tab, I see just one request Browsers don't show some requests in there. The most robust way to determine it is to read your server's logs or use Wireshark.

Also, in case you're using Reitit there's a debug route transformer - it will print out request and response information before and after every middleware, including a colorful diff.

Thanks. I zero'd in on there not being a second network request. I found the unhandled exception and I don't know why, it's calling the handler to be called a second time.

What middleware do you have? Easy to imagine for it to be something a tad weird like

(try
  (handle (wrap request))
  (catch Exception _
    (handle request)))

Have you checked what goes on in the session middleware? Does it receive a cookie and match it to a session?

Oh, sounds like the 403 has been resolved. Nvm

Is this in emacs? I'm pretty sure the cljr- functions for thread and unwind are obsolete and simply alias the clojure- functions

yeah, emacs

There's #C099W16KZ btw. Why do you ask? Have you noticed a difference?

No, just configured all my clojure keybindings and saw an overlap

always use clojure mode functions over cljr, as they don't require compiling. I'm a bit biased as my cljr almost never works

You can't extend an existing (compiled) Java class to implement a Java interface. You might be able to: • Change that class in the source code so it extends that interface via regular Java means • Create a new class that inherits the existing one and implements the interface • Create a new class that delegates to the existing one and implements the interface

Unfortunately, I am working with an external library Java class, so I cannot modify its source code to extend the interface directly. Creating a new class that inherits or delegates to the existing one is also not an option for me. Do you know if there’s a protocol I could use instead to work around this problem?

Are any wrappers completely out of the question? Or only Java ones?

If you can't wrap instances of the existing class in anything at all, then I believe you're out of luck because the seq function can only be applied to the following things: • Implementers of one of these interfaces Seqable, Iterable, CharSequence, Map • Subclasses of the ASeq class • Instances of the LazySeq class • null • Arrays

If you can create a wrapper in Clojure, then you can simply use reify, just like the iteration function does.

Wrappers are not out of the question. In fact, I have considered creating a protocol that “mimics” the Java class and using reify to implement the protocol and Seqable interface. However, I’m curious if there’s a simpler or more straightforward way to achieve this.

e.g. (Integer/MAX_VALUE)

on an interface

not a class

oh! my bad!

no worries 🙂

user=> clojure.asm.Opcodes/ACC_PUBLIC
1
user=>

it is the same as with a class, and you don't need the parens for either

I keep getting could not initialize class …

that is usually a problem with a static init

yes it’s a static init

then you need to figure out why the static init is broken and fix that other wise the jvm won't load the class file and you can't read fields

there may be some way to read static fields from a class without running its static init, but for data types that cannot be serialized directly in byte code, those values need to be constructed and then set as the fields value, which is done in the static init

thanks for the hint

was using some random lib testing something, and that dep was messed up

kind of late but you can also add breakpoints to interface static initializers with intelliJ and then fire the load from the repl, maybe that helps debugging those kind of issues

yeah it was pretty cryptic, thanks for that tip too