sequentiality is not a hard requirement, it just needs to advance in the same direction (decrease in this case)

i know someone who had a similar problem in c++ (unique ids) and they just decided to use the current time multiplied by the hardware thread index as the unique id

ew

we do have uuids if we need to go that far

leave it to c++ programmers to make something so ugly xd

for a moment I considered to use (clojure.lang.RT/nextID), but then again, I need to find cljs alternative too (also make it negative), and the same argument applies: too much obscure code for a negative int

what are you working on btw, if you don't mind indulging my curiosity

I'd use datascript's fn, but I try to keep dependencies at zero. which brings me to next question: how do you depend on a few library's functions, w/o depending on the library in project.clj? I went with config hash-map, which gets passed all over the place under the hood. Are there other alternatives?

(:require [lib :refer [foo bar]]) i think

oh wait i misunderstood you

yes :)

Are you checking for the availability of the lib?

I want to package code as a lib foo, which uses fns from lib bar, but do not include it as a dependency, rather ask to supply particular functions at client setup time

@misha make a function that takes the implementation as an arg, and supply a simple reasonable default

I've stored the config in an atom. Rather than passing it around

I tried datascript with react native, and it is very slow on iphone 5. I moved it to a separate thread, but RN adds like 50ms to a messages between threads, + datascript is not instantaneous itself, so I decided to build a cache layer to be used in a main thread, for fast reads and optimistic updates, which will talk to a db-thread

@noisesmith there are no alternative implementations: I need 4 particular datascript functions, but want to avoid bumping up dependencies.

normal clojure dependency resolution makes it simple for the end user to select their version

db-thread then will have all the time in the world to: sync with datomic, transact stuff, etc.

this isn’t something you need to solve

like

:dependencies
 [["foo-1" :exclude ["bar-2"]]
  ["bar-3"]]

or just :dependencies [[bar “3”] [foo “1"]]

the default rule is to use the first provided, and closest to the top level

imperative!

All that Rich'es talk was for nothing then

that’s why you don’t get somebody else’s version of clojure.core - because everyone puts clojure as their first dep as lein puts it there

so it overrides the various clojure.core versions referenced by libraries

this is not critical, I know, it's just what I took stab on to keep things interesting

it’s also possible to use a :provided scope for a dependency, which declares that you never ship that lib and the user will always provide it

but really just letting people override versions in the normal way suffices for this

I did not find a good/any description of :provided when I was looking some time ago, and that sounds like the best option so far (as long as I will put it in a readme, hehe)

that would just be annoying - people would ask why they have to add a dep by hand for something you use

or no? would people always be using the lib already if they use yours?

yes, they would, since it is a caching layer specifically for datascript. zero value w/o one. on the other hand, I am always annoyed when I need to go and bump version all over the util-libs I have, instead of just do it in a client application

Hello everyone, I’m new to clojure, could anyone tell me how to avoid use exceptions in clojure? For example, I have an function to check user input params and block the coming steps when the param is invalid, how to refactor the following code?

(defn check-params
  "Check my params"
  [params]
  (if (:required-key params)
    params
    (throw (Exception. "Missing required keys"))))

@sailxjx I prefer returning an ex-info instance. That way you can have a function that check if the returned value was an error and fetch extra information is necessary (logging) for example.

@carocad return an ex-info or throw an ex-info?

Return it. Check out the instaparse project. They use that approach as well. You can the check the type of the returned value as see if it is an exception. If so then you know the processing failed

@carocad what is the benefit compared to a try/catch ?

@leonoel this blog post explains it better than I could in a single message: http://yellerapp.com/posts/2015-06-22-exceptions-vs-failure-values-a-battle-to-the-death.html

Hi, how can ring get the form data from a request, I currently can get query-params , multipart, keyword-params, but I don't know how to get the js formdata.

(def svc-app
  (wrap-routes main-routes
               #(-> %
                    wrap-reload
                    keyword-params/wrap-keyword-params
                    cookies/wrap-cookies
                    params/wrap-params
                    mp/wrap-multipart-params)))

I found if I send the POST request using x-www-form-urlencoded, I will get the form-params, but if I send the POST request using js/formData the form-params will like this:

:form-params {"------WebKitFormBoundaryu7BKnalMAs1Vg5qP\r\nContent-Disposition: form-data; name" "\"token\"\r\n\r\n91827fc3227c5ba631ba53435327374273c45fa0\r\n------WebKitFormBoundaryu7BKnalMAs1Vg5qP\r\nContent-Disposition: form-data; name=\"statement\"\r\n\r\n1212\r\n------WebKitFormBoundaryu7BKnalMAs1Vg5qP--\r\n"}

at a clojure repl, is there . away to type in the name of a java class and get the documnetatio nf said class ?

(I'm thinking via reflection or something)

@qqq I don't think it's possible via reflection. Some environments support javadoc documentation for given symbol (e.g. cider-doc) - for standard java classes you can still use javadoc function (which will only open online documentation)

jumar: I see, so basically all tools call javadoc

I wouldn't say that - Cursive certainly has internal mechanism how to show javadoc for given class. not sure about cider, but in cider it also works for non-JDK classes for which you add sources to the classpath

right, but Cursive is also using javadoc to get the actual docs right?

what I meant to say: "the standard way 5toget docs for a class is via javadoc; and you can figure out how to display it later"

alright, I've had enough of clojure.java.jdbc I want something higher level

what higher level lib should I look into?

(j/with-db-connection [conn db-spec] (j/execute! conn (j/drop-table-ddl :letters))) gets me d.font=> PSQLException ERROR: statement cannot follow a schema change in a transaction org.postgresql.core.v3.QueryExecutorImpl.receiveErrorResponse (QueryExecutorImpl.java:2455) how do I drop a table in jdbc ?

try set transaction is false in execute!

(execute! db sql-params) (execute! db sql-params opts) The :transaction? option specifies whether to run the operation in a transaction or not (default true).

@lincpa: that fixed it; thanks!

hey guys, what tool do you use for database migration ?

Hi, how can I get all routes under some /path in compojure? for example, I want to match all /path, /path/a, /path/a/b/c, /path/v/e/r/y/long and so on to run in one handler? Or, maybe a wildcard matches such as /path/*?

i never seem to have a need for macros, am i doing clojure wrong

@michaellindon The author of book Clojure Programming says: You can go a long way before you start to use macros in clojure.

@cmal one of the main selling points of Clojure is the ease with which one can create macros... seems like I'm missing out on one of the main advantages of the language. Either my applications rarely need them or my brain is not wired to think in macros

First rule of macro club: don’t use macros.

Second rule (for experts only): unless you really need to

macros tend to be a modification of a reasoning model (in my experience). So it would just mean your reasoning model is in line with the base patterns of the language.

plus things get hairy quick with macros, so you want to tread carefully anyway.

thats good to know. It felt like I was missing out on some superpowers by not using macros

In some way you are, but normal code doesn’t need superpowers

Do protocols have an argument count limit?

CompilerException java.lang.RuntimeException: Unable to resolve symbol: ct->tensor in this context, compiling:(/home/chrisn/dev/cortex/src/cortex/verify/tensor.clj:138:17) 
user> (defprotocol test-proto
        (test-fn [arg0 arg1 arg2 arg3
                  arg4 arg5 arg6 arg7
                  arg8 arg9 arg10 arg11
                  arg12 arg13 arg14 arg15
                  arg16 arg17]))
test-proto
user> (test-fn 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17)
CompilerException java.lang.IllegalArgumentException: No single method: test_fn of interface: user.test_proto found for function: test-fn of protocol: test-proto, compiling:(*cider-repl cortex*:51:7) 
user> (defprotocol test-proto
        (test-fn [arg0 arg1 arg2 arg3]))
test-proto
user> (test-fn 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17)
CompilerException java.lang.IllegalArgumentException: No single method: test_fn of interface: user.test_proto found for function: test-fn of protocol: test-proto, compiling:(*cider-repl cortex*:56:7) 
user> (test-fn 1 2 3 4)
IllegalArgumentException No implementation of method: :test-fn of protocol: #'user/test-proto found for class: java.lang.Long  clojure.core/-cache-protocol-fn (core_deftype.clj:568)
user> 

I would expect the last error every time (no implementation).

No, my error.

0 based indexing...

I have a somewhat noob'ish question if anyone feels like entertaining it! I'm working on a transpiler-of-sorts, and it uses clojure.walk/postwalk to assign types to expressions, starting at the leaves of the tree (that comes from edn) and going up. as far as I can see... when I modify leaf nodes and then try to 'access' those leaf nodes from their parent in the tree, they aren't modified

so... is postwalk sort of creating an immutable copy and the tree you're accessing doesn't change as you traverse it?

what do you mean by “modify leaf”?

here's the source for reference: https://gist.github.com/anonymous/a7c1e7f4f9a7cf3bce3209c8a40fac09#L121

by modify leaf, i mean... when it's at the bottom of the tree, it uses with-meta to add metadata, and then i'm trying to basically progressively move that metadata up the tree

(clojure.walk/postwalk #(if (number? %) (inc %) %) {:foo 1 :bar {:nested 41}})
=> {:foo 2, :bar {:nested 42}}

ba!

i've been confused because when i get to the bit where it looks for child nodes to have metadata ([here](https://gist.github.com/anonymous/a7c1e7f4f9a7cf3bce3209c8a40fac09#L134)) it isn't finding that metadata set

john!

but if i log the places that postwalk traverses, i know that the metadata has been set... somewhere

let me see if i can make a distilled example..

whoah. okay, was in a totally different bug land than i expected 🙂 til that list? is not a very robust way to check for general listiness, seq? covers all the bases in this case

@ba lol sorry 🙂

@tmcw postwalk creates a new collection, the new collection it created has your new metadata

we don’t really update things in place, other than vars, atoms, refs, agents

and a hash map is none of those

with-meta returns the hash-map but with different metadata, not changing the metadata on the item you were looking at

gotcha, i'm more wondering... the function that's called for each node by postwalk - is it called on the new tree (and thus has the modifications of previous traversal steps) or are the methods always called on the old, original tree

it should be on the new one, but I wouldn’t be surprised if there was a subtle metadata bug - those happen a lot

👍 that's what the current operation of the code suggests, which is great

(not the thing about metadata bugs, i hope i don't run into one of those :)

it might save some trouble to define a two-way transformation into a node structure that explicitly shows that data (at the cost of hiding the “real value” under a sub key of course)

similar to how clojure.tools.analyzer.jvm represents the AST https://github.com/clojure/tools.analyzer

huh... why wouldn't analyzer use the metadata system?

because they found it more reliable to describe everything about an ast as data - in terms of printing, less gotchas where you unexpectedly lose metadata, etc.

but someone who works on that project might have a better answer than I

got it, cool! i'll consider that if i end up needing to store / debug information more in this tree ast

metadata isn't very robust and not all expressions can accept metadata

also the AST that tools.analyzer produces is not always 1:1 with the underlying code

that's a bummer about not being able to trust metadata, was pretty hopeful about that part of the language

@bronsa what do you mean by "not very robust"?

I think the intended use case was suppose to be "about" data, and never to be used in a way that affects the identity/equality of the value. Not sure if that pertains to this use case or not.

it’s not robust in that many people use functions that don’t preserve metadata - clojure.core probably fixed this in most cases by now, but random other libs likely haven’t

this means to use metadata reliably you need to agressively re-apply metadata after using the data

ok. i'd call that a matter of transparency rather than robustness, fwiw. metadata behaves the way it's designed to behave, i hope, since i'm using it a lot.

@mobileink yeah, I meant that as "using metadata to represent ASTs isn't a very robust approach"

maybe i'll need to rewrite my libs for v. 2. 😢

gotcha.

metadata is fine per se

Hello all, how to write new String[]{"id","text"} from Java in clojure?

(into-array String [“id” “text”])

hummmm, thanks

are there any tricks for debugging test assertions that print exactly the same?

there’s extra parens in the fail case

lein difftest makes such things more obvious in my experience

i'm not sure the parens are... real? like if I add (first, then the 'actual' result becomes 'module', same as the expected

clojure.test does not insert fake parens

here's what that looks like in ultra:

no extra parens there, but a puzzling comparison of nils

nil means “no difference”

the difference is the two 0

maybe someone did (symbol “0”) which is not equal to 0 but prints the same (and is an abomination btw)

oh... :thinking_face:

but the symbol function totally lets you do that

peregrine.circle=> (symbol "0")
0
peregrine.circle=> (= 0 *1)
false

👍 back on the hunt, thanks

you could add a test that the 0 there is not a symbol - narrow down to the failure

Hey folks, I was wondering if I could get some tips on pre/post conditions. When looking at http://blog.fogus.me/2009/12/21/clojures-pre-and-post/ and in practice, when the precondition fails all you get is an AssertionError. Any way to get a better report of the error, especially the location where it happened?

It'd be nice if it told you the function or ns where it occurred.

stack traces will tell you both the function and ns

and when assertion errors are thrown you should get stack traces

well, I thought so, but maybe I am doing something wrong. Here's an example:

(defn constrained-fn [f x] {:pre [(pos? x)] :post [(= % (* 2 x))]} (f x)) => #'enhanced-hickory.core/constrained-fn ((try (constrained-fn #(* 3 %) 2) (catch Exception e (.printStackTrace e)))) java.lang.AssertionError: Assert failed: (= % (* 2 x))

ah, wait...

there’s extra parens around that try…

(try (constrained-fn #(* 3 %) 2) (catch Throwable e (.printStackTrace e))) => nil java.lang.AssertionError: Assert failed: (= % (* 2 x)) at enhanced_hickory.core$constrained_fn.invokeStatic(form-init6196173666840420140.clj:1) at enhanced_hickory.core$constrained_fn.invoke(form-init6196173666840420140.clj:1) at enhanced_hickory.core$eval15641.invokeStatic(form-init6196173666840420140.clj:2) at enhanced_hickory.core$eval15641.invoke(form-init6196173666840420140.clj:1) at clojure.lang.Compiler.eval(Compiler.java:6978) at clojure.lang.Compiler.eval(Compiler.java:6941) at clojure.core$eval.invokeStatic(core.clj:3187) at clojure.core$eval.invoke(core.clj:3183) at clojure.main$repl$read_eval_print__9983$fn__9986.invoke(main.clj:242) at clojure.main$repl$read_eval_print__9983.invoke(main.clj:242) at clojure.main$repl$fn__9992.invoke(main.clj:260) at clojure.main$repl.invokeStatic(main.clj:260) at clojure.main$repl.doInvoke(main.clj:176) at clojure.lang.RestFn.invoke(RestFn.java:1523) at clojure.tools.nrepl.middleware.interruptible_eval$evaluate$fn__996.invoke(interruptible_eval.clj:87) at clojure.lang.AFn.applyToHelper(AFn.java:152) at clojure.lang.AFn.applyTo(AFn.java:144) at clojure.core$apply.invokeStatic(core.clj:657) at clojure.core$with_bindings_STAR_.invokeStatic(core.clj:1963) at clojure.core$with_bindings_STAR_.doInvoke(core.clj:1963) at clojure.lang.RestFn.invoke(RestFn.java:425) at clojure.tools.nrepl.middleware.interruptible_eval$evaluate.invokeStatic(interruptible_eval.clj:85) at clojure.tools.nrepl.middleware.interruptible_eval$evaluate.invoke(interruptible_eval.clj:55) at clojure.tools.nrepl.middleware.interruptible_eval$interruptible_eval$fn__1041$fn__1044.invoke(interruptible_eval.clj:222) at clojure.tools.nrepl.middleware.interruptible_eval$run_next$fn__1036.invoke(interruptible_eval.clj:190) at clojure.lang.AFn.run(AFn.java:22) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617) at java.lang.Thread.run(Thread.java:745)

and an assertion error is not an exception

sorry for the span

yeah, needed to catch the Throwable

not Exception

so to get usable info from pre/post I need to wrap it in a try/catch somewhere? That seems to be the answer.

enhanced_hickory.core$constrained_fn - that’s your namespace and function, right at the top

> It'd be nice if it told you the function or ns where it occurred.

Assert failed: (= % (* 2 x))

at enhanced_hickory.core$constrained_fn.invokeStatic(

Yeah, what I was hoping for was a better message without wrapping the call.

(constrained-fn #(* 3 %) 2) java.lang.AssertionError: Assert failed: (= % (* 2 x))

also, with assert (as opposed to :pre and :post) you can provide a string describing the issue

^that

AssertionErrors are meant for catching things during development, I would skip the try/catch and use the file / line info from the stack trace to see which code I need to fix

Do you not have some version of this?

this uncaught assertion error has a stack trace

I'm in the cursive repl.

w/o the try/catch all I see is this:

(constrained-fn #(* 3 %) 2) java.lang.AssertionError: Assert failed: (= % (* 2 x))

ah. i'm surprised

and it tells you the function which failed the assertion?

no, it just dumps the assert

@markbastian if that’s all you see, clojure.repl/pst will show the whole trace

Yeah, that's my whole issue. I was expecting better reporting (the whoe stack trace)

cool

usually clojure.repl is in scope so you can just run (pst)

when in doubt, (use ‘clojure.repl) is a good idea when developing after switching namespaces

can you post a screenshot?

in lein repl i get at least the function that fails the constraint even though i don't get a stack trace

ah, that's cool

meaning pst

ah

So, pst def solves the problem for now. Thanks!

yeah- assertion errors and pre / post are mainly for dev time, it’s totally legit to turn off assertions in production

if you get an uncaught error in an app, you should get a stack trace unless someone is over-eager with a try/catch

i wonder if there's a setting you could set or maybe they could update that. you could try #cursive

yeah, this seems like a cursive issue more than anything.

Proposal: "The Clojure community should dump slack in favor of something to be named later." Please respond by selecting one of the following, where 🐓 means "dunno", or "undecided" or "no opinion". 👍 👎 🐓

motivation: this comes up sometimes. let's get something approximating real data.

well looks by a vote of 0-1 we'll stay on slack

hahaha! wait, you didn't vote! seriously, low response rate (let's give it 24 hrs) means it is not perceived as an issue. which would be good to know since clearly some do think its an issue.

late friday was probably not the best timing. 😉

ok just ate my doggiefood. what about you? heh.

my guess is most of us just haven't even thought about it, and probably don't much care. but you never know what real data might show.

I got hung up on "to be named later"

@mobileink : I just cast the deciding vote, takikng it from 2-2-1 to 2-3-1; get hacking, dude

without valid arguments for something better, hard to commit

the other problem is that with the 10k limit, and the inability to pin posts, few ppl will see mobileink's voting system before it vanishes

if the poll disappears, I would vote to move off of slack. but i would be unable to vote in this case. Slack's limitations prevent me from complaining about slack's limitations

@john: just wanted to gauge slack response. proposing a specific alternative would bias the tesult.

@qqq: methodologically, is it ok if we take all the people who did not see it before it vanished as "yes"?

@mobileink : personally, I would be okay if you took all the people who voted 'no' and interpreted them as 'yes' 🙂

I think if we had a better system, for wiki links . ;pinning stuff in discussion, people would also provide answers

dude, this is clojure, not php.

right now, it's an "eh, wasted effort"; whereas if we knew that it had long term value, we would be more likely to write wiki like responses

it'd also be useful to be able to answer questions by poiting at wiki docs instead of retyping ansers to repeat questions

yeah, hence the poll. you and me (and others) would like to see more, but if the vast majority of users are satisfied with what we have, it would be a waste of time to try to sell them on sth else.

i think the way to bridge this is by writing a slackbot that (1) archives everything and (2) starts auto answering questions

plenty of lorem ipsum generators around.

https://clojurians-log.clojureverse.org/clojure/2017-05-19.html

he's a log of today

better than nothing, i admit. but undeniably hackish. slack search does not talk to it akaik. and, frankly, i am far too lazy to go grubbing around the web to find old slack msgs.

There is a discord channel that got started today https://www.reddit.com/r/Clojure/comments/6c4z91/a_discord_for_clojurians/

so far we have a response rate of 9 out of approximately 10,000. Hear me, O People! I am compelled to say that I am a little dissapointed by your lack of enthusiasm. Vote - use it or lose it, and be condemned to program in Java for the rest of your sorry life! heheh.

ugh, did that have to call it "clojurians" again 😑

you got a better idea? sadly, i don't. "clojurites"? "clojacks"? "clojicians"?

our local clojure meetup was “clojerks” for a while

I first attended just so I could say I was an official clojerk

it’s “clojure pdx” now, but seattle’s was still called seajure last I heard

but how to pronounce the "j"? hard or soft? perhaps that could be our shibboleth.

@mobileink I think it was extensively debated before and "clojurians" won, so I probably shouldn't complain

oh, i like seajure (seizure?)

Are there any projects like skummet (which seems unmaintained) that create emit AOT-compiled clojure that can be tree-shaken?

I’m trying to get my binary size down. There are a ton of unused classes in my jar.