Clojurians Log v2

I got a question a really cool question... how do I synchronize logins / authentication between two separate ring servers? They are hosted at the same IP but I suppose I could use something like webhooks(tm?) to send a CSRF-free POST to the other box when there's a successful login and have the other "box" set that person as logged in... ? Hmmm. What do you think friends?

Hey guys, How could this be improved? Signal-cone receives two parameters: 1. An array of numbers (all greater than 0) 2. An array of 3 arrays - each with 2 numbers (all greater than 0) We can only use a range in which the lowest & the highest number does not fall out of the range of either of the other ranges. So in this call it would be [9, 15] Once we have our range we need return which numbers fall within that range. So here it would be [9, 11, 15] (defn return-inner-ranges [ranges] (into [] (map (fn [x] (into [] (range (get-in x [0]) (inc (get-in x [1])))))) ranges)) (defn get-shortest-range [ranges] (let [chosen-range (first (sort (return-inner-ranges ranges)))] (vector (first chosen-range) (last chosen-range)))) (defn get-within-range [numbers range] (filter (fn [x] (and (>= x (get-in range [0])) (<= x (get-in range [1])))) numbers)) (defn signal-cone [numbers ranges] (get-within-range numbers (get-shortest-range ranges))) (signal-cone [3,4,7,8,9,11,15,18,23] [[10 20] [8 16] [9 15]]) Thanks in advance!!

@santosx99xx Could you give me another input and output example?

If I put in [23, 55] and [[5 6] [77 2] [25 59]] what comes out?

from what I understand, no more than 3 ranges are given. It would probably be like this [[5 6] [25 59] [2 77]]. [25 59] should get chosen. And it would return all numbers that fall inbetween those two

that’s prob what I would do too

I guess only [59] would be returned.

@santosx99xx please let me know if i'm way off. I think I get it. I want to check numbers in [stairway] to see if they fit in the range of each item in [gold silver bronze]. (check-if-medal-lives-in-stairway-range [stairway] [[gold] [silver] [bronze]]) So if [stairway] is 77,78,79,80 and [gold silver bronze] are [[8 78] [78 79] [1 5]] the result will be all the numbers of [stairway] that fit comfortable in [gold] or [silver] or [bronze] ... so here the output would be [77, 78, 79] from the inclusionary endcaps on [gold] and [silver]. I feel that with [gold silver bronze] it would be better to have a function that receives one [stairway] and one [medal] and returns the set that fits, endpoints included, from stairway into medal. And then I would call that function 3 consecutive times, once on gold, once on silver, and once on bronze, and throw them all into the same #{set} so only the uniques remain. How to check which numbers exist within a specific range? It's a cool question. I think you can call (range) on the endpoints in something like [gold] .. (range 0 5) to get [0 1 2 3 4] Then after you have the whole range splayed out, you can see which numbers exist in both [stairway] and [medal] and keep the overlapping ones. You can use (clojure.set/intersection ) and if the inputs have overlapping entries (entries in common) the result will be those.

let me know if i didn't understand the challenge, i'm under the impression it takes 2 inputs and produces one output set, right?

Thanks, ill try that

the first input being just numbers, the second being 3 ranges. Id have to choose the range that falls inbetween the other two. so given ranges [[8 78] [78 79] [1 5]] - [8 78] would get chosen. say you were given a vector of [0 1 2 3... 100] it would return [8 9 10 11... 78]

(signalCone [[8 78] [78 79] [1 5]] [0 1 2 3 4... 100]) -> [8 9 10 11... 78]

How to know what symbols can be used for vars and what not? Like , is ignored but . is a valid var name. What is the rule to know what is allowed?

You can see what's allowed at: https://clojure.org/reference/reader#_symbols

How to convert PersistenceList to vector in clojure?

Hello Antti 👍

thanks, nice to be here

(vec list)

Hello there, Is it possible to reassign atom with same value ? For example:

(def image (atom "Inital"))
(if (or (= (get-in data [:ParkCategoryId]) 1) (= (get-in data [:ParkCategoryId]) 5) (= (get-in park [:data]) 6) )
  (reset! image image)
  )

(reset! image image) -> (reset! image @image)?

(swap! image identity) will keep the same value. you need to reset it to a value. image is an atom. (reset! image @image) is a way to do this. but what purpose do you want to achieve by setting something to its same value?

are you triggering a watch?

I am working on a function that is responsible for image icon In js I have as following

let image = 'inital';
   if (data.Available && data.AvailableFiltered)
        image = "AvailableNotCriteria";


    if (data.ParkCategoryId == 1 || data.ParkCategoryId == 5 || data.ParkCategoryId == 6)
        image = image;
    else if (data.ParkCategoryId == 2)
        image = "B_" + image;

This is in Js but this is what I am trying to do

Hi all. Are there some rules about fully qualified classes when using metadata? I get a

Syntax error (ClassNotFoundException) compiling deftype* at (com/eldrix/clods/fhir_r4.clj:49:1).
java.lang.Organization

(deftype OrganizationResourceProvider [^ODS ods]
  IResourceProvider
  (getResourceType [_this] Organization)
  OrganizationGetResourceById
  (^{:tag org.hl7.fhir.r4.model.Organization
     Read true}                                            
    getResourceById [_this ^{:tag    org.hl7.fhir.r4.model.IdType
                            IdParam true} id]

(deftype OrganizationResourceProvider [^ODS ods]
  IResourceProvider
  (getResourceType [_this] Organization)
  OrganizationGetResourceById
  (^{:tag Organization
     Read true}                                            
    getResourceById [_this ^{:tag    org.hl7.fhir.r4.model.IdType
                            IdParam true} id]

(:import (ca.uhn.fhir.rest.server RestfulServer IResourceProvider)
         (ca.uhn.fhir.context FhirContext)
         (org.hl7.fhir.r4.model Organization IdType Address Identifier)
         (ca.uhn.fhir.rest.annotation Read IdParam))

(if (or (= (get-in data [:ParkCategoryId]) 1) (= (get-in data [:ParkCategoryId]) 5) (= (get-in data [:ParkCategoryId]) 6) )
  (reset! image @image)

yes, but again, it does nothing

just don't do that

@mark354 try (with-meta …) rather than ^{} for those cases where you are experiencing issues

The ^{} shorthand is a reader macro so it occurs at read time, which means your Java classes have not been imported yet: https://clojure.org/reference/metadata#_metadata_reader_macros that is my guess anyway

I think you’re right; the issue seems to affect only the metadata within definterface and deftype that are declaring the structure of the eventual java classes - the metadata within the forms making up runtime expressions can be short-form. I might just leave it as fully qualified class names but thanks for the with-meta pointer.

Just got to ignore the advice of the IDE to shorten an already imported reference!

i have a function i'm importing from a shared lib, and that function is throwing. the stack trace refers to line numbers and namespaces from the required lib, but not the namespace that requires it - not the namespace that actually invokes the required function. is there anything i can do about this to improve the stack trace?

the stacktrace while show everything that is on the stack when the exception is thrown

if it doesn't show your calling code it either isn't on the stack (because the called code is being run in another thread) or some tool you are using is eliding it when printing/presenting the stacktrace to you

so a first thing to do is to just enter *e in the repl after you get the exception (assuming the exception is thrown the context of evaluating an expression at the repl)

that should (but not always) get you the default clojure printing of the last exception throw in the repl, which should be complete

> code it either isn't on the stack (because the called code is being run in another thread) i suppose with pedestal, it's on another thread

i don't have access to *e because the error is being automatically caught and logged for me

by pedestal

if it is on another thread the information just isn't there

the stacktrace in an exception is literally a trace of the call stack as methods are called on the jvm, and each thread has its own call stack

realizing the lazyseq returned by my function dramatically changes the stack trace for the better

perhaps that means my original problem was the second explanation - that my tool/framework is eliding the trace

ah, well, lazy-seqs can behave similar to threads

well, not like threads

a lazy-seq suspends a computation to be completed at some other time

and the stack may be different between when it is suspended and when it is completed

I have a question about lazy vs. non-lazy sequences and converting between the two. Here’s an interpreter session that’s illustrates my confusion:

(defn prnr [a] (prn a) a)
=> #'grdn.core/prnr

(->> [1 2 3 4]
     (map prnr)
     (take 1))
1
2
3
4
=> (1)
(->> (seq '(1 2 3 4))
     (map prnr)
     (take 1))
1
=> (1)
(->> (seq [1 2 3 4])
     (map prnr)
     (take 1))
1
2
3
4
=> (1)

@thisfred you are calling a side-effecting function prnr inside your map over a sequence

so if I pass a vector, map is not lazily evaluated, if I pass a seq literal it is, but if I convert a vector to a seq it once again isn’t.

thank you, yes

that’s the point

and so you are seeing some of the implementation details of clojure sequences

namely, that they are realized in chunks of 32

this is done for efficiency

if you instead do something like this, you'll see results that make sense, I suspect

(doseq [a (take 1 [1 2 3])] (prnr a))

the thing is that when you call map you are getting a new sequence, and it will be realized in chunks of 32, for efficiency

thank you so much, I’ll read up on doseq

that means that even though you end up discarding most of it (with take) your function that you pass to map still gets called for more elements of the seq

you're welcome!

that makes sense I guess, though I wish the documentation wouldn’t say map was lazy, in that case 😉

You're not the first person to be confused by this

however it does speed things up for clojure's lazy sequences

Perhaps the docs could be improved

(I know, side effects should probably be relegated outside of threading macros)

Anyway, really helpful!

compare with

(->> (seq (vec (range 99)))
     (map prnr)
     (take 1))

Ooh, yeah, thanks, that demonstrates it beautifully

so its still lazy, but they never give you (in the contract) the degree of laziness so to speak. and don't assume you know what it is, or rely on that for side effects or to prevent heavy computation

It is lazy, but it is not strictly lazy :-)

another common gotcha that you might misinterpret at smaller repl-test-sizes is that maps might appear to be ordered (keys {1 1 2 2 3 3}) is (1 2 3) but they are not ordered. just a quirk of small sized maps using an optimization

gotcha, thanks! I wasn’t expecting them to be, because of prior Python experience (although they since made the ordering stable there.)

The actual use case we have is we want to return the first non-nil result of an expensive calculation mapped over a vector. My colleague just pointed out that this works instead of using map + first or (take 1):

(->> vector
     (some expensive-calculation))

(just in case someone stumbles upon this in the future with that exact need 😉

That works, but the fact that the input is a vector means that it has already been fully realized. If every element is (or might be) expensive to calculate, and you don't need any after the first non-nil to be calculated, then that will cost more than a more lazy solution.

Yeah, in this case the input is fully realized (passed in from the frontend)

Oh, sorry, I misunderstood your statement. Thinking more on the actual issue.

if you want to control expensive evaluation using map and first is a poor way to get there. (loop [[x & xs] the-seq (when x (if-let [transformed (f x)] transformed (recur xs)))

(some f vector) is neat though