Anyone good at math? It’s funny. I’ve got myself (again) into stupid argument about how Lisp is not less readable because of prefix notation and I’ve been thrown a “challenge” in my face: > simple formula > Combined trip distribution:

f(t) = A (t^B) exp(Ct) where t = t (i,j)

Then t(i,j)=A(i) f(i,j,t) / Sum(A(i) f(i,j,t))

I’ve tried to google and it’s not helping. I have found bunch of articles about trip distribution methods. Uniform factor method, Average factor method, Fratar method, Furness method, Gravity model, Tanner model, Intervening opportunities model, Completing opportunities model.. holy cow…

dumb, class in college related issue

is there a json parsing library for clojure that can handle multiple top level forms

delineated by arbitrary whitespace

cheshire/parsed-seq

@ag I am not familiar with that formula, although I see there is a Wikipedia page for it. Before someone can help you write code for it, I think you (or someone) needs to identify what the parts mean, e.g. f(i,j,t), is that effectively a 3-d matrix/array indexed by integers i,j,t? Or are some of those things real numbers, not integers? The Sum, what is it summing over? j an integer ranging from 1 to N? If so, is N an input? The A(t^B) in the first line, is that the same A as the A(i) in the second line, or a different thing?

Did the people giving the challenge give it in the form that you show there? Or did they just say "look up the Combined trip distribution formula and implement it"?

@andy.fingerhut is more positive than I would be presented with this. 🙂 it feels like as presented, it already hasn't been reduced into a form that is useful for computation, and it doesn't matter if it's in prefix or infix notation.

I mean, sometimes I like a challenge, but it helps if I know what is being asked first 🙂

ya. qualified my statement to not be a trolly post. 😄

A question related to reify: As I understand reify allows to create an anonymous record by extending an existing record. Is there a way to to the same but non anonymously i.e. to create a record that extends an existing record?

>As I understand reify allows to create an anonymous record by extending an existing record. this is not correct

reify doesn't create an anonymous record nor does it extend one

and there is no mechanism in clojure to extend a class to another class, so no to your question (well proxy I guess, but it's not what you're asking for)

as bronsa says, reify has nothing to do with records

sorry I meant, reify allows to create an anonymous type

> While deftype and defrecord define named types, reify defines both an anonymous type and creates an instance of that type. from https://webcache.googleusercontent.com/search?q=cache:y8by8CHJKvgJ:https://clojure.org/reference/datatypes+&cd=3&hl=en&ct=clnk&gl=il

yes, it creates an anonymous type, which is not a record

but it doesn't extend any concrete classes, nor records

if you want a non-anonymous reify, that's more or less deftype

modulo the IObj metadata support

Is there a way to use deftype and provide a “base” type and extend it (like with reify)?

reify can't extend a base type

reify can only implement interfaces/protocols

deftype can't either, only proxy and genclass can extend base classes

no concrete derivation, by design

which is also discussed in that same page you linked above

https://clojure.org/reference/datatypes#_datatypes_and_protocols_are_opinionated

"Concrete derivation is bad" to quote directly

@viebel Maybe you should describe the bigger problem you’re trying to solve by extending a base type?

I remember that once I used something (maybe in clojurescript) that allows to extend a record

you may be thinking of specify, which is a clojurescript only construct

yes. Exactly

it doesn't exist in clojure

Is there a good reason for that diff between clj and cljs?

specify doesnt extend a record either, it provides additional implementations to a particular instance of a type IIRC, and in clojure we can do something similar to specify now with extend-via-meta in protocols

I don't know why clojurescript implemented specify

but it's not part of clojure's design, and it's not particularly easy to implement on the JVM

to leverage the host capabilities, I believe

although this is maybe a better question for #clojurescript or #cljs-dev

Hola, I'm struggling a bit to get zippers to do my bidding: https://gist.github.com/devn/61b3690f4d6e059303874d547c21f803 What did I mess up here? I've tried a bunch of variations and can't seem to figure out why the (set? l) branch in the loop doesn't get hit.

use tree-seq

@hiredman can I use that to construct a get-in/update-in-esque path like this?

no

but a recursive for will do it

l is a loc, which is the data structure a zipper uses to keep track of its traversal, so it is never a set

using into and empty to reconstruct seqs will reverse them

your child|branch? function only considers seqs to be branches if they have children

I seem to recall suggesting a recursive for to you for this kind of thing previously

((fn f [path parent-id item]
   (let [id (java.util.UUID/randomUUID)]
     (cons (assoc (dissoc item :children)
                  :id id
                  :path path
                  :parent parent-id)
           (for [child (:children item)
                 i (f (conj path (:name item))
                      id
                      child)]
             i))))
 []
 nil
 {:name "root",
  :description "root",
  :children
  [{:name "Foo",
    :description "foo",
    :children
    [{:name "Bar",
      :description "bar",}
     {:name "Baz",
      :description "Baz",
      :children [{:name "Qux",
                  :description "qux"}]}]}]})

you may also want to checkout a lens library (I haven't used any so I cannot recommend any), at least in theory with those the "path" you build will be directly usable to update sub values

@hiredman appreciate that, and yes, I seem to remember you recommending that approach previously as well 😄

@hiredman ok, just had a chance to run it, the only missing bit there is that some :children can be sets, which means the path would need to be constructed as ("root" {:name "foo" :description "foo" :children ...} ...) in some cases

and for vectors, it needs to be the index of the item in the vector for get-in/update-in reasons

(for anyone following along at home, I updated my gist and wound up getting my zipper-based solution working https://gist.github.com/devn/61b3690f4d6e059303874d547c21f803) One question I have in light of this is: Is there any particular design reason why update-in doesn't handle sets? For instance, the ability to do:

(update-in {:a {:b #{{:c :d}}}} [:a :b {:c :d}] assoc :e :f)

update-in is for "associative structures"

and inherently leans on assoc

Hey guys, i'm making a guide to people who already do web development and want to have a head start in clojure

Here is the draft: https://community.inkdrop.app/note/ede8cdf47aafb065ac7c0964d4fca7d5/note:krng7hycU

I want to know if i should mention data input validation or this is out of the subject of my guide

i would talk about input validation using Spec

a quick glance on the draft is enough to know what i am presenting on this guide

Hello all, what happen with lein repl with clojure 1.10.1? The repl is starting in 1 second with cider, is that correct?

@fabrao What do you think has changed from previous versions of Clojure with lein repl and/or CIDER?

I think the way of loading the stuffs in memory

Can you be a bit more specific? And also which version of Clojure are you comparing this with? I can't think of anything that changed in Clojure. I believe CIDER's default middleware changed to be lazily loaded which may well improve startup time.

my guess is you have a user.clj file and you upgraded to a java version where it has the performance regression for code execute when running static inits

from 1.10.0 to 1.10.1

did you change JDK version?

oh, sorry, I think 1.10.1 is just the work around for that performance issue, so likely not it

I did update in JDK version yes

ah, nope, no work around for that yet

I´m asking this because before it started in 10-20 seconds

you should try again with the older jdk version if you can

I did update JDK and clojure

JDK 11+ has a perf problem when loading things in user.clj

1.10.1 is starting faster for you -- so that's the workaround described in https://github.com/clojure/clojure/blob/master/changes.md

yes, I read this, but 10-20 times faster?

If it was 10-20 seconds before, on 1.10.0, then you probably hit that user.clj loading issue yes.

Did you also update CIDER? (see my comment above about it switching to lazy loading middleware).

yes, I did update CIDER too

well, this is a HUGE performance increase, I´m very happy with result, I even thougth to buy Mac to speed up the stuffs

so basically it could be a lot of things

It's always better to update things one at a time and then you'll know exactly what caused any observable change.

Updating multiple parts of your tool chain at once is ... an exercise for the, er, brave 🙂

yes, I did not update some stuffs fearing breaking some, and I asked you just about this because it was breaking after update some libs

Why isn't there a ssecond? I've never been more disappointed in my life.

Why do you think there isn't?

Wait, ssecond or second?

@skuttleman What would ssecond do? (second (second x)) ?

REALLY!? Two seconds only no ssecond!? https://clojuredocs.org/clojure.core/fnext

I knew about the various first/`next` combo functions but never actually noticed these two docstrings were identical:

user=> (doc second)
-------------------------
clojure.core/second
([x])
  Same as (first (next x))
nil
user=> (doc fnext)
-------------------------
clojure.core/fnext
([x])
  Same as (first (next x))
nil
user=> 

So I guess ssecond would be fnfnext 🙂 🙂 and Clojure <> Lisp so we don't go down the caddadr path...

I suspect the various first/`next` combos all got in before the API was really being thought about? It would be interesting to hear from Stu or Rich on that...

ffirst and nnext are both actually used in clojure.core, so likely Rich considered them general purpose enough to include them in the API.

second is used in clojure.core, but not fnext

There is a great Sasha Baron Cohen interview about 15 years old where he asks some prominent financial expert why there isn't an $87.21 dollar bill

The expert asks him how he would pay if the bill was $87.22

He of course responds, "I'd pay with the $87.22 bill"

That's why there isn't an ssecond