Clojurians Log v2

I think that "clojure" has a runtime, "clojurescript" dosent Kind of we can implement "clojurescript" on JVM or "clojure" in JS But with this definition in mind, the reader conditional that currently use :clj to "macro time" in :cljs turns wired.

Depends what you mean by runtime

Cljs does have a big pile of support code, the main difference is the support code for cljs is a mix of clojure and clojurescript, the support code for clojure is a lot of java

By runtime I mean something like a stateful singleton that hold things like keyword cache and threads.

not sure what threads has to do with it

and clojurescript does its best to cache keywords

Anybody has idea on how we can run PACT Test as provider for amqp tests in Clojure, I’m using https://github.com/DiUS/pact-jvm

I use leiningen and I use pomegranate to add libraries needed when I experiment in the repl to avoid restarting the repl. Any other alternatives? quick google search shows alembic and add-lib

I'm pretty sure that's it for now.

No idea, but have you also looked at the Clojure solutions available? If you need to generate fake requests, there is e.g. ring-mock

In Rich Hickey’s talk Simple Made Easy he talks about set functions as a source of simplicity opposed to imperative loops/folds as a source of complexity

Does anyone have an idea of what he means exactly when he talks about set functions

I’m assuming it’s not clojure.set

Hmmm… appears to be

Best I can make out of it is functions that take sets (in the mathematical sense, which for clojure could be lists, vectors, maps, sets, etc) and apply a function on each element in the set and then returns the output. So map, reduce, filter, etc would all apply

I believe that's a set, used as a function, like so: (#{1 2 3} 1) => 1 (#{1 2 3} 42) => nil

The functions you're quoting here operate on the Seq abstraction, which is a very central Clojure thing. Maybe every seq is a set in the mathematical sense, but I doubt that was the point there.

I believe he's referring to mathematical sets and implementations of them in clojure. Functions on sets like intersection/ union / conj / disj are simple because they are idempotent. Sets are simple because ultimately they're only about membership. I do not think he meant Seqs and all the functions that surround Seqs. Those properties of sets allow for declarative expression and manipulation of membership, whereas other data structures may require careful policing and ordering to achieve the same which leads to more complex and imperative code.

it's about what collection functions that complect what's to be done and how that's done. he pointed loop and fold as examples. although both operates on whole collection (what), both operate from first element, one-by-one, to last element (how). like when using map or filter, we don't care the how, the order of things or the order of how things work--whether the work done from beginning-to-end, end-to-beginning, or even worked in parallel. I guess Rich named those 'set functions' because sets don't have ordering in its members.

Right… I have a hard time thinking it relates to clojure.set as you can do so much more in a imperative loop/fold than what clojure.set offers

map, filter, etc are equivalent power to imperative loop/fold, but are simpler

Hi

Is there a variant of concat that allows you to do that (concat '(3 4) (5) 1 6 [3 [6] ]) returning '(3 4 5 1 6 3 [6])

??

Without doing that (mapcat #(if (coll? %) % (list %) ) '(1 [3 5] (3)))

@vfernandez11 While I don't recommend using it, flatten comes close:

user=> (flatten '((3 4) (5) 1 6 [3 [6]]))
(3 4 5 1 6 3 6)

In the Clojure core library, I think that is the most concise understandable way

You are rigth

But I'm solving 4clojure problems and I'm implementing flatten

With the restriction of using flatten

what do people use for code formatting lately, that works with different editors.

Understood, thanks.

cljfmt, clj-kondo (both have good support for major IDEs)

clj-kondo doesn't do formatting

I saw zprint and joker do it too, but not sure how widely used they are

@gtrak This is a fork of cljfmt which looks pretty great: https://github.com/greglook/cljstyle by @greg316

thx, I'll check that out.

been a bit of a lull while I’ve been busy with other projects, but I’m planning to get to a 1.0 release later this year with better config and more options 😄

while on the subject, can puget also pprint code or is it more tuned towards EDN @greg316

we're spoiled with prettier, black, and ocamlformat, but trying to get clojure in now.

puget is very much about EDN/data rather than code, though it does do one or two code highlighting things

Anyone have some solid examples of this library in action by chance? https://github.com/weavejester/ring-oauth2 Perusing Github didn’t bring up much

I'd like to do something like this using spec, and I'm having a hard time.

(spec/def ::int-to-int-fn
  (spec/fspec
    :args (cat :n int?)
    :ret int?))

(defn add-5 [n] (+ 5 n))

(defn subtract-3 [n] (- n 3))

(spec/fdef add-5 ::into-to-int-fn) ;; this draws an error

(spec/fdef subtract-3 ::int-to-int-fn) ;; this draws an error

(spec/fdef add-5
    :args (spec/cat :n int?)
    :ret int?)

You can also do the opposite and write a maybe-cat transducer. Just copy cat's definition and add a coll? check which would branch to reduce or rf, I'll find an example later and send it if you're interested

While running this from the command line clj -Sdeps '{:deps {org.apache.pdfbox {:mvn/version "2.0.21"}}} I get the following errors and a stacktrace

Error building classpath. 3 problems were encountered while building the effective model for org.apache.pdfbox:pdfbox:2.0.21
[ERROR] Failed to determine Java version for profile jdk9 @ org.apache.pdfbox:pdfbox-parent:2.0.21
[ERROR] Failed to determine Java version for profile jdkGte9 @ org.apache.pdfbox:pdfbox-parent:2.0.21
[ERROR] Failed to determine Java version for profile jdk11 @ org.apache.pdfbox:pdfbox-parent:2.0.21org.apache.maven.model.building.ModelBuildingException: 3 problems were encountered while building the effective model for org.apache.pdfbox:pdfbox:2.0.21
[ERROR] Failed to determine Java version for profile jdk9 @ org.apache.pdfbox:pdfbox-parent:2.0.21
[ERROR] Failed to determine Java version for profile jdkGte9 @ org.apache.pdfbox:pdfbox-parent:2.0.21
[ERROR] Failed to determine Java version for profile jdk11 @ org.apache.pdfbox:pdfbox-parent:2.0.21    at org.apache.maven.model.building.DefaultModelProblemCollector.newModelBuildingException(DefaultModelProblemCollector.java:197)

@heykieran That doesn't look like a valid coordinate to me -- you have just the group ID with no artifact ID.

This works:

clj -Sdeps '{:deps {org.apache.pdfbox/pdfbox {:mvn/version "2.0.21"}}}'

You don't get an error on the first attempted call to spec/fdef on add-5? I did, in my attempt to reproduce what you see.

Perhaps if you do (spec/def add-5 ::int-to-int-fn) it achieves what you want (note spec/def instead of spec/fdef)

Thanks @seancorfield - I was trying to simplify a larger problem where I was getting an error dep’ing a local jar. I ran the cli in the same directory, which merged the deps, and what I was seeing is the original error. Sorry, clj never got to tell me I was an idiot for using bad coords. It had baled already. Back to square one for me.

newer clj will fail earlier with a little better pointer

No I got an error for both of them. Typo on my part.

I can spec/def add-5 ::into-to-int-fn without an error, but what I'm looking for to achieve the equivalent of giving :args, :ret, etc. with a single token declaring a common i/o signature

For example if I spec/def add-5, then ask for docs on add-5, the 'spec' section of the output is nil.

I do not know if there is a way to achieve what you want with spec. (I'm not saying there isn't -- just that I'm not enough of a spec expert to give a definitive answer either way)

It’s a long shot, but any obvious things I should look for (and might cause an error like this) while dep’ing a jar that I know can be referenced in Java, its classes imported, and used without issue?

what error are you seeing? that stuff above?

Yeah it's starting to look like it's out of scope for spec. It'd be cool though. Thanks for your help!

Yup. The only dep in the deps file is a local jar. Even just trying to start the repl throws the error. Not much info there, sorry.

@heykieran Is it anything we can see publicly, to help you debug it?

@seancorfield this is ironic. Once I got it working, I was going to write a blog post about how to merge data to Word files and push to docusign. The jar does the merging. However, I haven’t got the ok on that yet. And I don’t want to color outside the lines. I’ll see if I can reduce it to as simple a case as possible. Appreciate the offer.

You may get a more expert answer by asking in #clojure-spec channel

what you're seeing above is almost certainly due to jdk profile stuff in maven - you might need to explicitly pull in one of the classifier jars (maven has a whole system for figuring this out which is not really be used through clj)

Question about polymorphism: should I limit myself to just built-in tools like protocols where they fit, or is it okay to have something in front of protocols that might perform automatic conversion to target type? For example, if there is a map with a special key, I might supply a particular behavior for these maps, while other maps are treated by default protocol impl

as a general practice, I nearly always front a protocol method with a function

and that function can do some automatic conversions?

that's up to you, but it's often a good place for custom processing to grow

I just worry that might be a source of complexity where easy to use makes it hard to reason

can you set JAVA_HOME and see if that helps?

if the API is "it has to be something implementing protocol, and we implement Object and nil for you", that's quite easy and straightforward, but for some common uses there might be wrappings required...

pdfbox parent pom has a property that uses ${env.JAVA_HOME}

I think that's what's failing (and what's not failing for me because I have it set)

well if you're the one doing the wrapping for an extender at use then you're in control. depends how the wrappings come into play

Unfortunately, JAVA_HOME is set. I think it’s the way the jar (a kind of Uber jar, with everything included) is being created. I may pull it apart and then reassemble. Even if I take out the pdf dependencies in the jar, I get an error about commons-io. Two funny things, the base jar without its dependencies imports fine (but doesn’t work) and the uberjar works fine in Java. Appreciate your help.

Users have to do the wrapping, but I can infer that wrapping from what they supply to me from the data they supply, and it will work in 99% of cases. And for remaining 1% where they had the key but "didn't mean it", they will be able to supply the data in a way that I'll be able to treat as if they "didn't mean it"

oh, that could be it - uber will sometimes strip or replace things that collide and you can run into some weirdnesses that way

Hmm, I guess that might be worth it...

Yes I'm interested

Something like

(defn maybe-cat
  "Like [[clojure.core/cat]] but falls through on non-sequential elements."
  [rf]
  (let [rrf (preserving-reduced rf)]
    (fn
      ([] (rf))
      ([result] (rf result))
      ([result input]
       (if (sequential? input)
         (reduce rrf result input)
         (rrf result input))))))

then just use it in a transduction

(into [] maybe-cat coll), or any other API

@gtrak @greg316 https://bogoyavlensky.com/blog/clojure-formatting-cljstyle/ <- someone blogged today!

oh wow

hah, nice!

hello, I want to create a Slack bot where functions react to certain messages on chat depending on regexp matches to call the right function. i.e: a "cake" message gets dispatched to the (defn cake ...) method. I would like to specify the regexp that matches a function next to it in its own source file, but what is the best way to aggregate them all so that the dispatcher fn knows all the available regexps/defn couples? should I use some kind of reflection code or is there a better way?

Having just turned off the bot which was a part of our product's Slack integration: we have centralized all command handling in the HTTP handler function, since you're just going to get a raw message text and you'll have to parse the commands yourself. Disclaimer: that was an old school bot, it had to listen to all incoming messages. I never worked with slash commands, which are newer style and might do some command parsing

there's definitely more than one way to do it. the approach that first comes to mind: • define your cake method and add some meta data with the regexp

(defn ^{:slackbot.ns/regexp #"cake.*"} cake ...)

can I lookup in a wildcard namespace? like myapp.commands.*? maybe I'd just lookup all the subnamespaces of myapp.commands.*?

it's my first clojure project so I'll have lots to learn but I figured metadata was a likely answer

you can, but it's easier if the namespaces are already loaded when you try to find the handlers

another idea I had was to make create a macro that, when called, added a regexp to defn entry to a collection of maps and then defined the defn as well?

the other option that I've used is have a macro like defslackhandler and that handler will "register" the handler in a global list. "registering" just means swapping into an atom in the slack namespace

what would be the more idiomatic way?

probably, https://github.com/clojure/spec-alpha2/blob/master/src/main/clojure/clojure/alpha/spec.clj#L440

seems like that's how spec registers specs

alright, thank you very much I'll go the macro route

the def macro translates to a register call which adds an entry to an atom

gonna be a fun exercise in demystifying macros 🙂

You can also inspect ns with ns-interns and ns-publics

https://github.com/vlaaad/reveal/blob/master/src/vlaaad/reveal.clj#L167

doesn't ns-interns require that the namespace is already loaded, not just available on the classpath?

Reveal uses that to build cli command list from defs

Yes, it has to be loaded

so will I need to "preload" all my calls to defslackhandler?

Do you think you need dynamic loading and multiple namespaces? Maybe single ns with command list would do?

Or maybe even a single def with a map...

I am reimplementing a bot with thousands of lines of ruby and 50 commands

if it's all in one file it's gonna be a mess...

I think that's fine for a single map from regexps to handlers if handlers are extracted to other namespaces

I like to keep code that works tightly together in the same file, which I why I want the regexp and the defn next to each other

you just need to make sure an load all the namespaces with handlers from the namespace that contains your main

so that I don't have to switch emacs buffers around to make modifications

Hello. How can i add my_project/src/clj/server.clj to classpath? Inside my project.clj i have :source-paths ["src/clj" "src/cljs"] but it doesn't work and i dont know why:disappointed:

can you give a few more details as to what you are trying and the error message you're receiving?

I try to start backend REPL with `lein run` and get: Can't find 'grump2.server' as .class or .clj for lein run: please check the spelling. Syntax error (FileNotFoundException) compiling at (/tmp/form-init3649709 871957213789.clj:1:73). Could not locate grump2/server__init.class, grump2/server.clj or grump2/server.cljc on classpath. And it doesn't matter where is `server.clj` file: `grump2/server.clj`, `grump2/src/server.clj` or `grump2/src/clj/server.clj`

do you have (:gen-class) in your ns declaration?

no

is it in project.clj should be?

oh seem i understood

i need add this to server.clj?

I would try that, and it's expecting the file in:

<myproject>/src/clj/grump2/server.clj

I think. if I'm reading the error messages correctly

i've created src/clj/grump2/ dir and added server.clj file and db.clj because i have import there. And get same error but now it can't find db.clj

do you have the command you're running and the error message you're getting?

excuse me

i'have fixed the error

thank you very much:grin:

now i have new bugs but it's working now and it's great

Can i write to you sometimes for help?

I'd recommend posting in #beginners. if I'm around and available, I'll try to help!

Great! Thank you 🙂