Hey folks, how are you dealing with consistently and automatic code formatting in your teams (assuming that people are using different editors/IDEs)? I'm wondering if is there out in the wild something similar to lint-staged (https://github.com/okonet/lint-staged) for Clojure code formatting.

@wcalderipe The "Clojure way" of formatting is pretty uniform across most editors/IDEs I think -- and there's the community style guide.

https://guide.clojure.style/

There are linters but they're more about code constructs and some of the non-layout stuff in the style guide.

CIDER and cursive usually have some small differences. I think colin does a lot of work to be amenable to us folks stuck on a 1970s editor 🙂

@dpsutton I'm a bit surprised that Cursive doesn't use "standard" paredit/parinfer etc indentation?

@wcalderipe There is this https://github.com/weavejester/cljfmt

@seancorfield There really is no standard, and formatting varies pretty widely in the wild.

This is the closest thing to anything that could be considered a standard, if only because it’s the only option which isn’t form-dependent: https://tonsky.me/blog/clojurefmt/

I'm a bit surprised to hear that -- I very rarely notice much variation across all the open source projects I work with. Perhaps there are small differences I just don't see at this point...?

I notice little differences in formatting between our code. Mostly in cases where the old version of clojure-mode I have gets it wrong.

Tonsky's post is hardly a "standard" given how much it disagrees with the style guide and how CIDER / paredit / parinfer generally work 🙂

I have an key binding that reformats the current buffer, and often hit it without thinking about it (old habits)

It’s not a standard, but it’s the only thing that could eventually become one IMO.

@hiredman maybe you're more OCD about code layout than me? 🙂 (and I thought I was pretty OCD 🙂 )

https://github.com/ztellman/potemkin/blob/master/src/potemkin/utils.clj#L1-L7

So I then see the little differences in the git diff

Ah, I usually ignore whitespace in git diffs 🙂

Oh man, my last team, we had tests that would fail if code wasn't formatted the "correct" way

I've been on teams like that -- in other languages.

So you would get build failures from the ci server, and I used to think that was cool

What a waste

Life's too short to argue about whitespace and indentation, eh...

Indeed

And things like 1-space vs 2-space list indents etc are the source of much debate: https://github.com/bbatsov/clojure-style-guide/issues/126. There are a couple of pull requests on the style guide repo trying to amend that.

I mean, it’s totally bikeshedding, my point is that it’s not really decided, and the style guide isn’t really gospel.

I had a PR on the style guide once, it got rejected, but to be fair I deleted all of it and replaced it with a sick Bruce Lee quote

I remember seeing that thread and just sort of rolling my eyes and getting on with my life ¯\(ツ)/¯

Thanks for your explanation and for the shared references@seancorfield @dpsutton @cfleming. I have see differences between editors and IDEs code formatting and it does surprises me that Clojure doesn't have an "one to go format" out there. Personally, I see code formatting as noise in PRs it does taking people away from what really matters.

If someone sends me a PR with spurious reformatting in it, I just reject it. Show some respect when you work on an OSS project: follow the existing style/layout when you submit proposed changes.

If I submit a PR to someone's project, I try hard to follow whatever style they're already using. My proposed changes aren't about the layout.

(I will say that if you have the "tabs vs spaces" wars going on in your language community, you're going to see a lot more wild formatting differences in PRs... Clojure's mostly "2 spaces" so that sort of thing seems much rarer... mostly 🙂 )

For all the power programmers wield, we still can’t let everyone use whatever visual layout and keep the central source in some normalized format. It’s always baffled me how we can’t separate presentation and storage in this specific instance, where we always go on about it everywhere else.

i try to follow the existing style -- if i can figure out what it is 🙂

it would be nice to get at least some macros as values/functions... for example, or to provide a default in a map: (update opts :encoding or "UTF-8")

you mean, added to clojure.core? Such additions are far and few between these days (saying no to an addition is temporary, saying yes to an addition is forever).

If that isn't what you meant, such a function is trivial to write, yes?

it's trivial, sure, so it's about addition to clojure language. I just thought that there are usecases for "soft macros", or maybe they should be called "inlineable functions", so they are both functions which you can have as a value and do optimized stuff like short-circuiting and not evaluating some args it all when called as is

probably not gonna happen ¯\(ツ)/¯

Hi - sorry to interject current conversation, please thread replies on this message? I'm having issues with atoms and pr-str, edn/read-string... trying to figure out how to work around this: Clojure:

(clojure.edn/read-string (pr-str (atom "hello")))
Execution error at user/eval61644 (form-init4368157810432879853.clj:1).
No reader function for tag object

(cljs.reader/read-string (pr-str (atom "hello")))
#error {:message "No reader function for tag object.", ....

(pr-str (atom "hello"))
=> "#object [cljs.core.Atom {:val \"hello\"}]"

All clojure function calls cause all args to be evaluated before the function is called. Short-circuiting is impossible via a function call, unless you use techniques like passing futures/promises/lazy-seqs as args.

yes the idea I've had is about extending the language itself, and probably not going to happen..

No worries. I've had plenty of ideas that are probably not going to happen, too 🙂

A question about class loader for importing java classes (in repl): Does it use the per-thread context class loader? The example on pomegranate home page just doesn't work for me:

(use '[cemerick.pomegranate :only (add-dependencies)])
(add-dependencies :coordinates '[[joda-time "2.10.4"]]
                    :repositories (merge cemerick.pomegranate.aether/maven-central
                                        {"clojars" ""}))

user> (import org.joda.time.DateTime)
Execution error (ClassNotFoundException) at java.net.URLClassLoader/findClass (URLClassLoader.java:382).
org.joda.time.DateTime

In "Programming Clojure" book, it is claimed that for Clojure the number 0 is True but when I tried it on the REPL (true? 0) gives me -> false Any explanation?

Even (= 0 true) gives me -> false.

Is this outdated material or something?

@arto.eg this exactly checks if the value is true not truthy

So truthy has another check function?

so in clojure, falsy values are false and nil. everything else is truthy

so checking not falsy is probably what you want?

(if 0 :truthy :falsey) => :truthy

probably surround your expr that you wanna run when truthy with a when-not

Great. Makes sense now! 👍

https://clojuredocs.org/clojure.core/when-not

(and x true) to test for truthyness

(boolean x) to test for truthiness!

Is it a bad idea to use take-while on the output of pmap?

Depends on the behavior you expect.

I'm expecting map but parallel, stopping after the first element that fails the predicate, in the sequence that map would do

I don't think pmap would give you that. I think it looks ahead, but I'm not certain.

I solved this problem with (sort) but I am not satisfied. Is there a better solution?

clojure.core/max comes to mind

It's restricted for this exercise as you can see in the red box.

aw

It's restricted to allow beginners to think harder, which is good. But by using sort I just fooled the exercise so I wanted a better answer 🙂

I think this is a fine answer 🙂

Maybe I can become more specific.. I didn't think that using If conditions is Idiomatic.

So I wanted to know if there is a better clojure way than sort.

if you wanted to see the "proper" way to do it, I would just (source max) at a repl and see what it does.

Never thought of that 🙂 Good point!

it's good to not do that at first when you're doing the 4clojure problems, but eventually you want to know "the right way" so that you can learn how you should do it

👍

I remember when i first started the 4clojure problems all of my solutions had loop/recur... but I learned not to do that by studying the code in core.

It can be pretty dense, but it'll get easier and easier to reason about what the code is doing as you get better

Thanks for sharing that, because It's exactly what I tend to think when I start solving a problem.... and I am not happy loops come to mind first.

yeah, it's hard to get out of the mindset

though, loops aren't the worst ever

but eventually you start to see things in terms of reduce, map, etc

(p.s. try to do it with reduce)

Sometimes I am not able to draw a line between destructuring and pattern matching but anyway I would probably prefer them from loops.

This was how I did it: (fn[f & r](reduce (fn[a b](if (> a b) a b)) f r))

(fn [& coll] (reduce (fn [x y] (if (> y x) y x)) coll)) might work also

Yes guys I agree, I think I need to refactor my mind around reduce much more often.

It's a functional gem.

Generally, any time you have N things and want a single result use reduce.

I should say "Single result after one pass over all N things".

I'm going to be the black sheep here and suggest that you don't study the internals of the core namespace - there are precious few exercises at 4clojure where you actually need that level of optimization, more often than not it's the knowledge of what's in the standard library that will be most helpful. And for that, http://clojure.org/api/cheatsheet is indispensable. 🙂

It depends what you're trying to get out of 4clojure. If you want to know exactly how you'd do it if you were making the same trade-offs as Clojure, then by all means go for it. If you're just letting it guide you to new parts of the core library, then don't bother.

I think the advantage of looking at the core library is just that it's a specific, obvious place to look. You won't necessarily need the optimizations that they make, but it's just something to look at at all.

when you're starting out it's not obvious where to look for code

I wish that 4Clojure would just show you some random solutions from people (when you haven't followed anyone yet) when you solve a problem rather than making you follow people first.

what should I do with warnings like this in JVM 11?

WARNING: An illegal reflective access operation has occurred
WARNING: Illegal reflective access by clojure.lang.InjectedInvoker/0x00000008003cb040 (file:/Users/borkdude/.m2/repository/org/clojure/clojure/1.10.1/clojure-1.10.1.jar) to method com.sun.org.apache.xerces.internal.jaxp.SAXParserImpl.parse(org.xml.sax.InputSource,org.xml.sax.HandlerBase)
WARNING: Please consider reporting this to the maintainers of clojure.lang.InjectedInvoker/0x00000008003cb040
WARNING: Use --illegal-access=warn to enable warnings of further illegal reflective access operations
WARNING: All illegal access operations will be denied in a future release

enjoy?

https://clojure.org/guides/faq#illegal_access

I think you can use --illegal-access=debug to get more info about what is happening

it will often point to clojure here but only because the clojure reflector is involved

(and fwiw, "future release" here does not include at least up through Java 13, so not sure how serious that admonishment is)

thanks, enjoying it 🙂

That message is annoying because it prints to stderr. On prod app startup, this results in several lines getting logged as error messages. We have to implement custom filters to prevent those messages from triggering alerts.

anyone know a way to make (prn (quote (quote foo))) print ’foo instead of (quote foo)?

the only thing I can think of is a postwalk with a list -> string substitution, but even then that wouldn't quite work with prn...

I know there’s hooks via print-method, maybe I can mess with the print-method for List

user=> (w/postwalk (fn [f] (if (and (list? f) (= (first f) 'quote)) (symbol (str "'" (second f))) f)) [:a (quote (quote foo))])
[:a 'foo]

(prn for that prints the same string you see in the result)

why do you want to do that?

was looking at this SO question and assuming zprint was getting its base string from there. I may be wrong about that now I look closer though https://stackoverflow.com/questions/58308404/configure-symbol-quote-expansion-in-clojure-zprint

there actually is custom support in the clojure printer for printing code, don't remember if it does the quote translation

user=> (require '[clojure.pprint :as pp])
nil
user=> (binding [pp/*code-table* (assoc @#'pp/*code-table* 'clojure.spec.alpha/def #'pp/pprint-hold-first)] (pp/with-pprint-dispatch pp/code-dispatch (pp/pprint '(quote foo))))
'foo

it does

you don't even need all that, there's some extra stuff for spec support in that (dredging out of my notes)

pprint actually always did the correct thing, this is about zprint 🙂

Is there any concise overview to clojure interfaces and protocols? I mean seq, seqable, reducible etc. My intent is to understand for example why implementing IReduceInit for io/reader allows to use it in the context of a coll (e.g. in sequence and eduction)

well, in general, Clojure is mostly implemented in terms of Java interfaces (and to a lesser extent, Clojure protocols). All of the core library is written to use those abstractions, not concrete implementations. If you know which things to fill in, you can make your stuff work like core stuff.

self reducibility is implemented in the (open) CollReduce protocol, which has a coll-reduce function

it's extended to IReduceInit/IReduce, which are Java interfaces that concrete Java classes can implement to hook into that protocol

if you're doing Java stuff, it's probably best to implement one of those interfaces. if you're doing Clojure stuff, it's probably best to extend CollReduce on your thing

Is there any news about updating the Clojure error messages? Any references to such news or it's not going to be implemented?

For folks working on 4clojure problems, there is now a dedicated channel #4clojure

hey everyone, quick question. Is it possible to extend a Jvm type to support clojure functions like conj/cons/assoc ? In the same way of extend-protocol ?

is it a type you make and control? or an arbitrary thing?

if the former, then yes, by implementing the right interfaces. if the latter, then no.

hey why i cant bind model of a table to a atom with seesaw? Im doing sth like this (def model-atom (atom [:columns .... :rows ....])) (bind model-atom (property :model table)

@carocad you can't since those aren't protocols, but you can make a protocol yourself if you want

or a multimethod

or... if you are feeling absolutely evil

(ns my.ns
  (:refer-clojure :rename {cons core-cons}))

(defmulti cons (fn [head tail]
                 (type tail)))

(defmethod cons :default
  [head tail]
  (core-cons head tail))

(alter-var-root #'clojure.core/cons (constantly cons))

then you can extend it however you want

which i cant seem to get to work at the repl, but in principle does the evil voodoo

Possibly because Clojure core itself is direct linked?

clojure core is aot compiled and direct linkeed

so internal calls to cons will not see this redef

yep that