@hiredman thanks 🙂

I have the following setup: 1. there is a LAN 2. on this LAN, there is: server: running http-kit (w/ websocket), port 8000 boot reload node https forwarding forwards 8443 to 8000 using https://github.com/nodejitsu/node-http-proxy laptop: can reach server:8000 and server:8443 chromebook, ipad: can reach server:8000 , but NOT server 8443 3. what tests should I be running to debug this?

Is the laptop on LAN and the tablets on WLAN?

laptop, chromebook, ipad are all on WLAN solved issue: as it turns out: 1. I have a self signed SSL certificate. 2. laptop has the *.cer installed, chromebook/ipad do not 3. chromebook/ipad: instead of asking me if I want to trust the self signed certificate, just decides to say: network reset error

Say I have a side-effecting thing which I need to serialize access to.. I manage serializing access to it through an atom holding a state ― :idle or :in-use. Does that make sense to you too? Would you rather use an old-school java lock idiom or something else more clojuresque?

the problem with this strategy is you need to make producers actively poll the atom when the consumer is busy

you can't really solve this problem without some kind of queue

I would look at agents, or core.async if you need backpressure

right, it's really a lazy hack avoiding a queue solution

I assume agents will only help as part of a queue solution, unless I'm missing something

the use case you describe is pretty much idiomatic for agents

Yeah, I believe agents were included in the language in order to tackle this issue, if you're interested in resources, the o'reilly clojure programming book has a chapter which goes over how the data is accessed concurrently in each scenario, and when to use each concurrency type (atom, ref, agent, var)

thanks guys :thumbsup:

I guess a java locking idiom would actually do the same as the atom approach, I mean hog the CPU while waiting, or would it be more like the agent approach in its implementation and impact?

I guess it depends on if whether or not you want to use the Java way of dealing with the issue, the tools suggested are just safer to use. I'm sure you could use a lock between two threads

agents take a function similarly to a thread, i've never attempted to use them, I tend to stick to core.async with channels

Quite frankly, using clojure's locking, was 3 lines of code to touch and serves well as a stop-gap solution until I have time to fiddle with core.async. This makes sense, as clojure's basic idioms aren't geared toward side effects, side effects are a second-class citizen.

https://clojuredocs.org/clojure.core/locking

Why isn't delay an instance of IBlockingDeref? The blocking aspect of calling deref on an unrealized delay is not mentioned in deref's documentation and the variant of deref with timeout (3-arity function) cannot be used with delay. I don't see any conceptual difference between e.g. promise and delay in this case, so I think that 3-arity deref should work.

Congrats with your post on http://clojure.org @val_waeselynck! Amazing job: https://clojure.org/guides/repl/introduction

This will be a great resource. 🙂

Has anyone used spyscope? https://github.com/dgrnbrg/spyscope

@borkdude @val_waeselynk that's great - I'd suggest one more trick for the "troubleshooting" entry - if you accidentally end up in a namespace created by in-ns (typo or forgot to require), you can get back to a semi-sane state by calling (clojure.core/refer-clojure) - after that you can do what you need to get into the ns you needed (require it, fix a typo, whatever)

prs welcome

@kingcode I use spyscope all the time, it's great

@schmee I am trying it out with the latest version, and keep getting the following error #spy/p (+ 1 2 3) ;;=> 6 ;;=> 6 #spy/d (+ 1 2 3) ;;=>ClassCastException java.io.StringWriter cannot be cast to clojure.lang.Associative clojure.lang.RT.assoc (RT.java:778)

seems the maintainer doesn't update the project anymore

but there is a workaround here: https://github.com/dgrnbrg/spyscope/issues/26#issuecomment-320614139

the fix is in master but the jar hasn't been released

@schmee Thanks! Will try your fix. I forgot to mention that it works fine with clojure 1.7.0 and spyscope 0.1.5, but would like to use clojure 1.9.0 if possible..Thanks again.

I was trying to understand how to make a reducible collection for a reader following this gist: https://gist.github.com/borkdude/ddc9433c396b1ff43a091cf2901b9dc7#file-text_xform-clj-L15-L26 however, I don't understand the line (recur (f state line)). Where does f come from? Where does state come from?

@spei f is the argument fn passed to the top call to reduce (line 17). ’state is the loop local which becomes (f state line), and is initialized to the value of ’init, itself another top level argument to reduce. So essentially, invoking ’lines-reducible with the requested BufferedReader will yield an object which is reducible: each element is taken from reading a line in the closed-over reader.

@kingcode But the function is called as #(lines-reducible (io/reader %)), no where is f and state ever given

@spei correct. f is provided by the client code (you?) and state is built up at each step, starting with ’init

@spei just like you would with (reduce + 20 {1 2 3]) replace [1 2 3] with your buffered reader, + with f and init with 20.

@spei state is the final result returned by reduce

maybe i'm misunderstanding how reify works. I just don't see any code that calls the reduce

@spei that's the transduce

([xform f init coll]
     (let [f (xform f)
           ret (if (instance? clojure.lang.IReduceInit coll)
                 (.reduce ^clojure.lang.IReduceInit coll f init)
                 (clojure.core.protocols/coll-reduce coll f init))]
       (f ret))))

@spei example: (def my-reducible (lines-reduce (BufferedReader. (FileReader. (File. <somepath>))))) (reduce my-reducible my-own-func “”)

@spei don’t worry about ’reify. It simple means “I’m giving you something that can reduce itself”. Try the example I typed in a repl with a real file, and you’ll see it works. Use conj and [] for f and init

@kingcode i see, that makes sense

@spei (or somthing similar, I probably made typos…)

is there a way to turn the line-seq into a IReduceInit? is that even a valid question? does the lines-reduce function return an IReduceInit?

@spei “I don’t see any code that calls the reduce” That is not the point of reify. Reify prepares something for you to make your call(s) on it, using your own function (f) and initialized value (init)..Hope this helps..

Have you tried it with an example? I think it will help you it you try it…

@kingcode I think it can be confusing when you look at the example in the gist - there's nothing that call reduce; because there's just transduce which calls .reduce internally

@jumar Aw OK…sorry, will try it myself

@spei Here is a 2-line code example you can try with the gist code (def br (BufferedReader. (StringReader. “Line 1\nLine 2\nLine 3\nLine4”))) (.reduce (lines-reducible br) conj []) ;;=>[“Line 1" “Line 2” “Line 3" “Line4”]

that kinda makes sense

@spei Here is the same example using a transducer instead: (def br (BufferedReader. (StringReader. “Line 1\nLine 2\nLine 3\nLine4”))) (transduce (mapcat #(lines-reducible %)) conj [] [br])

@spei you could add more sources (make br1 ,br2 and shove them as last input arg to (transduce …[br br1 br2])…

@spei Good luck and thanks for sharing…

@kingcode @jumar thanks for the help, it makes quite a bit more sense now.

Is there any documentation around regarding writing custom test.check generators? I’m trying to write a generator to sample a collection with a non-level distribution.

I'm guessing you've seen the stuff in the GitHub repo?

Yeah scanning through all of it atm.

Is there a Clojure equivalent to Scala’s Slick DB?

Nothing I know of that’s released and really tries to take Clojure’s native datastructures and make them transparently durable besides hitchiker-tree. I’m trying to build something like that but it’s very much experimental.

ztellman’s new bifurcan libraries are working towards offering durable models of data which can be round-tripped efficiently but I don’t know how much of that works yet.

clojure’s own datastructures are all Java serializable, so you could go put something like this together yourself using say datascript or one of the other in-memory datalogs as a frontend.

Datomic is probably the closest thing to a database for Clojure out of the box, but it isn’t part of Clojure it’s a separate product.

@arrdem re: generators, did you see my conj talk?

@gfredericks I don’t think so, but after some digging I concluded that test.check just isn’t appropriate for the generator semantics I want since I just want to kick out random examples with a known distribution and no shrinking so really I just want to hand-roll my own sampling machinery for this one special case.

Will definitely check it out tho!

@arrdem Ugh, that stinks. I have a colleague who’s insisting on NOT using HoneySQL/YeSQL style, which I think works fine for scoped micro-services.

is there a builtin function which does truthy -> true falsey -> false ? input: arbitrary object, output: boolean

i.e. (if x true false)

@qqq clojure.core/boolean

user> (boolean nil)
false
user> (boolean false)
false
user> (boolean 1)
true
user> (boolean (Object.))
true
user> 

@jmckitrick 😕 that’s definitely the common path AFAIK

@arrdem Maybe Korma??

I haven’t worked with any of the three personally but it feels like I’ve heard more negative experience reports about Korma than the other two.

Hmm