I'm seeing if I can turn my project.clj into a deps.edn. Can't figure out using a github repo. E.g., :deps {com.github.cognitect/transcriptor {:git/url "" :sha "05e9eb093b21d9a3458e0d946ad9dab76ae1e1c8"}...} gives "Manifest type detected"

I do succeed, however, if I use com.cognitect/transcriptor {:mvn/version "0.1.5"}

(that last being, a verification that the library is includeable)

you need the url to the git repo

not the github webpage for the repo

so maybe https://github.com/cognitect-labs/transcriptor.git

OH

yes of course

reasonable idea! but, no

:thinking_face:

status code is a 301

also the guide suggests a form like {:git/url "" :sha "04d2744549214b5cba04002b6875bdf59f9d88b6"}}} which..... seems like what I used

aren't the two forms [email protected]:foo/bar.git and the https form?

so I'd try [email protected]:cognitect-labs/transcriptor.git

no, if yo go to repo and click the clone button the url it gives you is the https url with .git

the root of the project does not contain a deps.edn

I tried successfully with the following deps.edn:

{:deps {clt {:git/url "" :sha "05e9eb093b21d9a3458e0d946ad9dab76ae1e1c8" :deps/root "examples"}}}

yeah, that is the problem, it isn't a deps.edn project

Hi guys! Question: I have a spec defined like this:

(s/def ::person (s/keys :req [::foo ::bar ::baz]))

@vincentdm Not any easy way, no. This is a problem that Rich talked about at Conj and a future version of clojure.spec will address this.

Rings a bell! Okay, then I won't waste time in this avenue. Thanks!

The problem is that s/keys complects shape and requiredness.

Hi all, new here. Not sure what the correct etiquette for asking new questions is but...Is there a built in function similar to clojure.core/time but allows for prepending a string to the output of elapsed time output. I find when using lots of time calls it can get confusing where they are from. I know it's simple enough to write this myself using do but just curious if there is already something built in?

@chriskellendonk probably easiest is to use with-out-str and then modify the string you capture

Oh nice, thanks I didn't know that existed. I'll check it out looks pretty close to what I need

user=> (str "foo: " (subs (with-out-str (time (+ 1 1))) 1))
"foo: Elapsed time: 0.010721 msecs\"\n"

it's sloppy still, but it's a start

Yeah I will probably just write my own helper to wrap this

And how do you get the original value in that case?

ah good point, this does not pass the output through

Where did 2 go? 🙂

user=> (time (+ 1 1))
"Elapsed time: 0.038226 msecs"
2
user=> (with-out-str (time (+ 1 1)))
"\"Elapsed time: 0.009613 msecs\"\n"

yeah, you could use a promise or whatever to capture the return value

but that seems ugly

I was using

(time (do (print "Timed =>") (+ 1 1)))

which works, but was hoping this may be built in. Sounds like it's not though. Thanks for the help 👍

Here's what we use:

(defmacro with-time
  "Evaluates expr and returns a map of the :value and the :time (in ms)."
  [key & expr]
  (let [value (if expr key :value)
        expr (if expr (first expr) key)]
    `(let [start# (System/nanoTime)
           ret# ~expr]
       {:time (/ (double (- (System/nanoTime) start#)) 1000000.0)
        ~value ret#})))

(ins)user=> (defn timed [f] (let [res (promise) timing (with-out-str (deliver res (time (f))))] {:result @res :time timing}))
#'user/timed
(cmd)user=> (timed #(+ 1 1))
{:result 2, :time "\"Elapsed time: 0.029535 msecs\"\n"}

Return a hash map with the time and the value.

yeah, that macros probably the smarter bet

The docstring was not written with the public in mind 🙂

I still want the value to be passed through as normal with time. I just want to add my own custom string so it reads "My string => Elapsed time: 0.029535 msec"

that's straightforward to do without a macro, in that case

(with-time expr) produces {:time .. :value ..} whereas (with-time :x expr) produces {:time .. :x ..}

I don't know the macro syntax but something like

(defmacro time-with-str
  "Evaluates expr and prints the time it took.  Returns the value of
 expr."
  {:added "1.0"}
  [my-str expr]
  `(let [start# (. System (nanoTime))
         ret# ~expr]
     (prn (str my-str  " Elapsed time: " (/ (double (- (. System (nanoTime)) start#)) 1000000.0) " msecs"))
     ret#))

This still returns the normal output but lets me use a custom string to mark this specific call

(defmacro time-str [prefix & body] `(let [start# (System/nanoTime) ret# (do ~@body) time# (- (System/nanoTime) start#)] (println (str ~prefix time#)) ret#))

about the same thing

using prn in this context seems odd, as it just forces awkward quoting of the string part of the output

(. System (nanoTime)) might make sense in the context where clojure.core/time is defined, or might have been the right code when it was implemented, but it's not how we'd normally do interop

That’s good to know. And thanks for the implementation

oh now i see that clojure.core/time was using prn - I guess that' s to make it clearer in the repl that what you are seeing is separate from the return value?

Which predicate would return true on a vector, a list, a set or a sorted-set, and false on a map?

(sequential? ...) maybe

eh bah, not on set

I guess your best bet would be to assert that it is seq?, but not map?, I don’t think that all non-mapping collections have a common ancestor. Edit: or is coll? and not associative? is maybe better.

(prove me wrong tho!)

Are the semantics of #inst defined at the level of EDN or Clojure syntax?

^ Please ignore. #inst and #uuid are part of EDN.

Are there plans to add more tags in EDN? (Stuff like IP addresses and URLs) I did check out this discussion on google groups about #inst and tagged literals in general:

no plans to add those to core atm

you can of course add any tagged literals you like

How I do a validation on a map with prismatic schema that returns the data if the data is correct or empty string if the data is incorrect?

I ask on schema's channel?

or here?

hi all

can anybody help me with something? I'm trying to read a clojure file and analyze the forms, but it looks like autoresolved keywords can't be read 😞 I'm not sure what to do...

how are you reading it?

the reader has a pluggable system for resolution so it is possible to plug in your own handler there to give it the “namespace context” if needed

i figurred it out 🙂 thx

had to bind the reader-resolver to some NoOp impl of a LispReader$Resolver

i guess that's what you mean alex

yep

I keep getting this at various times:

SLF4J: Failed to load class "org.slf4j.impl.staticloggerbinder".
SLF4J: Defaulting to no-operation (NOP) logger implementation
SLF4J: See  for further details.

This isn't anything I'm depending on directly... presumably something else I'm depending on is

not sure if I should just ignore it or address it in my project.clj somehow, or what

it's red, that's never good

SLF4J is a logging API with pluggable impls

many Java libraries use it

I'm using a JGit wrapper, that's my first suspicion

what it means is that one of those libs wrote a log statement, and no log impl was found, so it’s using the default (which does nothing)

yeah, jgit uses it

and since I'm not logging anything, /shrug

you can add an additional dependency on one of the explicit slf4j impls and then use that to configure the logging

if you want the warning to go away

I get similar errors when I boot up CIDER on an empty lein project

I see that https://github.com/clojure/tools.logging exists.... if I include it, what else would I need to do?

well, that will only deepen your problems :) tools.logging is also a pluggable logging api.

you can plug slf4j in underneath it though :)

You can include the dep org.slf4j/slf4j-nop to explicitly tell it to not do anything

Thanks, Alex. I'll try that tonight when I'm at home

ha argh

oh, I see

that's.... not something I would have thought of 🙂 thanks Alex

(I am tickled by having the option of choosing a beta version of no-op)

choose your favorite vintage :)

this is one of those great tragedies of software right, there where once too many logging frameworks. Now there are too many logging framework frameworks

everything about Java logging should be set on fire

but how would I log that I did it

this is a problem that seems to plague the java ecosystem most

how did other platforms deal with it

Python has a built-in logging library that seem to be well-known enough and good enough that most people simply use it. I wouldn't be surprised if there are a few various implementations from years ago that are effectively "dead" now because of that.

also logging is something that is cross-cutting - needed at many levels, and its users expect to apply a single solution across libraries

so if something usable isn't in the default setup built in, you end up with lots of hacky attempts at modularity and compatibility and "dependency injection" that make the whole thing harder to understand or use properly

right, I must admit that I always use timbre in clojure. And most libs of my comp also use timbre. So basically, if something doesn’t its part of the problem

The other nice thing about Timbre -- and some of its satellite libraries -- is that you can "hijack" pretty much every Java logging library and force them all through Timbre, giving you centralized control.

yeah, was looking into that. Can you do it the other way around tho, or is it unwise to depend on Timbre if you publish a lib

Well... Timbre pulls in a bunch of other dependencies from the same author... and that has often bothered me enough to want to switch to something else...

...but whenever I've tried, the pain of logging usually makes me go back to Timbre 🙂

I think it's reasonable for a library to depend on tools.logging and then users can do what they want with it -- including using Timbre to control everything.

Right, that makes sense

… added slf4j-timbre to my project, now I get spammed with debug info of one of its libs

I’m back on the clojure wagon for a bit. Just got my vs-code set up with calva, and worked through the 4th chapter of Brave and True. It uses Lein - and I still don’t fully understand what the bounds of that are. When I start lein repl, it’s giving me a `Warning: implicit middleware found: cider-nrepl.plugin/middleware Please declare all middleware in :middleware as implicit loading is deprecated.` - is that something I should resolve, or ignore for now?

sounds like something the CIDER devs need to fix - or may have fixed in a newer version

@wmichaelshirk do you have an old project.clj or something with cider-nrepl in it? If you are just starting lein repl there should not be any CIDER deps in it at all.

sorry not project.clj, i mean profiles.clj

the version of CIDER that Calva would recommend is likely older than the one that makes the latest lein happy

Which lib is better for clojure selenium? https://github.com/komcrad/webdriver or https://github.com/igrishaev/etaoin or other

Are transducers commonly used in clojure?

@josh_tackett We've used clj-webdriver for years but it is no longer maintained. I've been looking at switching to Etaoin but haven't had the time to investigate it fully yet. I had not heard of that other one.

@joshlemer Define "commonly". I think they are being used more and more these days but I'm not sure I'd say they were "common" yet.

It can take a while before new features are really internalized by developers at large and become "the norm". We're only recently seeing "common" usage of the Socket REPL that was introduced in Clojure 1.8, for example, and people are still asking lots of questions about clojure.spec introduced in 1.9.

I think transducers are great -- but they can take a bit of getting your head around -- and a lot of material out there about Clojure was written before they were added so folks aren't as widely exposed to them yet as I think they will/should be.

@seancorfield I'm in the exact same position 🙂 Here are some notes I made about switching. Hope they are helpful: clj-webdriver to Etaoin - to -> go - current-url -> get-url - wait -> wait-visible - input -> fill-human - element & click -> query & click-el

We've started using HtmlUnit for some newer applications -- and nearly everything we have clj-webdriver tests for is legacy and will be retired (this year) -- so we must stick with HtmlUnit going forward rather than convert the legacy tests and/or take on a new Selenium library.

hello, I think I'm confused on how to bring in libraries. I want to use exponential math so am trying to bring in the math.numeric-tower library like so:

(ns armstrong-numbers
  (:require [clojure.math.numeric-tower :as math]))

I get the same error when I change the dash to an underscore.

Error:

Caused by java.io.FileNotFoundException
   Could not locate clojure/math/numeric_tower__init.class,
   clojure/math/numeric_tower.clj or clojure/math/numeric_tower.cljc
   on classpath. Please check that namespaces with dashes use
   underscores in the Clojure file name.

what are you doing to add that library to your project?

it doesn't come with clojure

hmmm. I was just trying to follow along with this: http://clojure-doc.org/articles/cookbooks/math.html

Do I need to put it in my project.clj dependencies?

that would be the standard way to do it if you are using lein, yes

what is the idiomatic way to do exponeniation (ex 3 ^ 3) in clojure?

user=> (Math/pow 3 3)
27.0

that's what comes with the vm at least

ahhh, ok, I think I had found that first. Math/pow is a java thing right?

I think numeric tower tries harder to use precise types

yeah, it's part of the java.lang.Math api https://docs.oracle.com/javase/8/docs/api/java/lang/Math.html

So, as I understand so far, clojure.spec and functions pre- and post-conditions server similar purposes (typechecking at runtime). Is clojure.spec supposed to replace the latter? Or is there something I'm not understanding about their usage?

if you review, that has the standard things like pow, abs, floor, ceil, cos, sin, sqrt etc. etc.

so what do I put in dependencies for the numeric tower? [clojure/math/numeric_tower "0.0.5"]?

@pberganza10 just definign a spec isn't enough to make it get checked, you can use spec checks in a :pre or :post, and you can explicitly turn on checking for a specced function

@chase-lambert I think the article you linked had a link to the numeric tower project, that will tell you the right way

it's probably more like [org.clojure/math.numeric-tower "..."]

yup, too obvious. sorry about that

there was a link right to the github at the top

yeah, and the github shows the dep vector to use