With a large set of examples for are, how do I know which one failed? I get output like:

FAIL in (required-args-tests) (refl_test.clj:66)
expected: #{}
  actual: :payne.refl/missing-args
    diff: - #{}
          + :payne.refl/missing-args

FAIL in (required-args-tests) (refl_test.clj:66)
expected: #{}
  actual: :payne.refl/missing-args
    diff: - #{}
          + :payne.refl/missing-args

FAIL in (required-args-tests) (refl_test.clj:66)
expected: #{}
  actual: :payne.refl/missing-args
    diff: - #{}
          + :payne.refl/missing-args

FAIL in (required-args-tests) (refl_test.clj:66)
expected: #{}
  actual: :payne.refl/missing-args
    diff: - #{}
          + :payne.refl/missing-args

the test is written like:

(t/deftest required-args-tests
  (t/are [fn-var expected] (= expected (#'r/required-args fn-var))
    #'amazonica.aws.ec2/describe-vpcs
    #{}

    #'amazonica.aws.ec2/describe-egress-only-internet-gateways
    #{}

...

Obviously I could change the equals to include the fn-var input, but that seems silly

this is a long-standing problem with are (but really with macros in general) - I think the infamous http://dev.clojure.org/jira/browse/CLJ-865 is related

@mobileink btw, Cognitect does on-site Clojure intro training. let me know if you need any more info.

why am i now seeing the antiforgery token error when i run my luminus project locally?

@thug.nasty are you passing in the anti forgery token?

Luminus enables CSRF protection by default, so you have to provide the token in form submissions

@yogthos yes, the token is being passed around

passed in*

what's the error you're seeing?

well, nothing now. I also just restarted the server

oh wait. does the system clock have anything to do with the generation of the token?

if you had a stale token on the page that would do it

for example if you reloaded the server and didn't reload the page

ahh that’s much more like it, thanks!

Hi, coding clojure since quite a while again. Was there a trick to make changes in code automatically be loaded when the app has been started with lein run?

Using compojure-api and couldn't figure out how to mangle headers with it without creating a separate namespace that gets the response and hacking the headers there

@leena there is #ring-swagger for compojure-api related stuff, try there?

Messing around and I have a map of name-spaced keywords generated by core.spec generator, and I want to return that from an api. Return JSON is all the namespaced keywords. Is there a way to easily turn all the namespaced keywords into just their regular keyword format?

clojure.walk/stringify-keys works

Can I run another task whenever lein deps is run?

So the above does not add them.

Compojure related question. I have defined a route GET “/xxx/:id”. When sending a request to /xxx/123 I would like to have a middleware that prints “/xxx/:id”. Is there any easy way to do so? (:uri request) is not enough.

@psalaberria002 why is it not enough?

I want to get it with ":id”, the same way it’s defined in the route

For the hot realoading, I found this and it works: http://www.http-kit.org/migration.html So made a main namespace that gets routes from the handler namespace (using compojure-api)

what do you want to print?

“/xxx/:id"

I want a unique identifier for each different route

you want to print the string :id? why?

so /xxx/1 and /xxx/2 will both return xxx/:id

for that kind of thing I recommend bidi instead of compojure

with bidi you associate routes with handlers, which could be keywords

it's more "decomplected" than compojure

thanks for the tip. I was thinking there may be a function in compojure that does exaclty what I want. will have a look into bidi, but I am afraid it will take some time to change the api from compojure to bidi.

hehe yeah I understand

Hello folks, I am working on a Clojure project for my study. My school always forced students to design the system with UML before implementing the functionality. That was easy to do with Java as a language, but I am not sure how this is done with a functional language as Clojure. I have read the articles on the net about this and I am wondering if you folks can give me an insight on how you are tackling designing a Clojure application.

UML is for OO languages... since Clojure is not OO, UML does not fit...

I don’t think that’s entirely true

it depends on how detailed you want your UML to be

there are still probably entities in your system that are ~same as the noun-like objects you’d have in OO (but they’re maps or records)

to can model the sub-systems without the need to reference “classes"

beyond that, I think it’s more useful to think in terms of data flow

I would represent the data as components and functions as how they interact with each other

Does anyone know if Buddy or any other lib can do the same thing as verify_id_token as in python's oauth2client? https://github.com/google/oauth2client/blob/master/oauth2client/client.py#L1530

I have used the java lib [com.nimbusds/nimbus-jose-jwt “4.18”] to do that

I've seen a lot of things for benchmarking CPU & time taken. But I don't seem to be able to find anything about memory consumption, have I missed it, or is there a JVM limitation in this regard and JMX is what I must delegate to?

What kind of things are you trying to measure?

UML has more than class and package diagrams, what about communication diagrams, sequence diagrams, state models, etc.

Communication diagrams fit really well with component/service design regardless of programming paradigm

agreed

sequence diagrams are great!

and sequence diagrams are not bad for function invocation (although higher order is challenging in some cases)

I flip flop between communication and sequence diagrams. Sequence for message order and communication to explore components/services

I think among Rich’s many skills are his adeptness at capturing a complex architecture in a single diagram that includes the essentials and omits the rest

these diagrams are typically a mixture of data and process

and not UML :)

I think that's possible but I don't have the intuitive insight to get there in one step...I need the intermediate diagrams to derive the final one

some good examples are in http://cdn.cognitect.com/presentations/2014/insidechannels.pdf

that’s Rich’s talk from EuroClojure 2014

I have been reminiscing about old fashioned Data Flow Diagrams (DFDs) recently....takes me back 25 years!

DFDs are great for functional (but again, challenging for higher-order stuff)

Is there a quick/easy way to have the java build spit out the last piece of source it was on when it died? I’m looking for a needle in a haystack of markdown being parsed as a macro, and it’s painstaking

java.lang.ClassFormatError: Unknown constant tag 112 in class file is a little opaque.

Wait, never mind. I think the parser is dying on sheer length

What do people use for generating documentation? I’m thinking of using swagger for apis but not so sure about for regular docs.

try Marginalia

https://github.com/weavejester/codox

I use codox and spit them to /docs. Then, if you’re on GitHub, setup the repo’s GitHub Pages to point to /docs

then your docs are published to https://<user|org>.

Thanks!

@lvh that's why I never use 'are'... haven't used it in years for this reason.

I Codox you can also embed Klipse snippets

*In

That’s interesting

Marginalia let's you embed Markdown everywhere and build html with side notes about whole source code

*lets

example: http://gdeer81.github.io/marginalia/

@miikka I guess a graph of memory consumption by a call.

@tsulej That’s super pretty.

@talexxx check out https://www.overops.com, I use it for keeping an eye on jars that I'm running - it's very handy for seeing what failed and why, especially as it displays the values of the parameters that were being used at the time. Must have saved me hours of scrolling through log files, it's free to use on a single jvm, then you pay if you want multiple ones

preferred way of sorting contents of an atom... ?

(swap! atm sort) ?

(let [a (atom [1 4 3 2])]
  (swap! a sort))
(1 2 3 4)

guess it depends what's in the atom

How about an atom with a nested vector... such as

(def nf-atom (atom {
                                              :nf [{:not-used stuff}]
                                              :things-i-wanna-sort [{:id 1 :message "I could be first..."}
                                                                                               {:id 2 :message "... or I could be first"}]}))

pardon the horrendous formatting

why is "atom" relevant here, is this question relaly "how do I sort this complicated sturcture" ?

Pretty much. I have an atom that holds a lot of application state, but I wasn't sure how to sort just a subset of it...

(def nf-atom (atom {
  :nf [{:not-used stuff}]
  :things-i-wanna-sort 
  [{:id 1 :message "I could be first..."}
     {:id 2 :message "... or I could be first"}]}))

you want to read up on swap!, update-in, and sort

sort will let you specify the sort function, update-in will let you choose which subset of data to sort (leaving the rest unchanged), and swap! will let you update the atom

(swap! nf-atom update :things-i-wanna-sort #(sort-by :id %))
{:nf [{:not-used stuff}], :things-i-wanna-sort ({:id 1, :message "I could be first..."} {:id 2, :message "... or I could be first"})}

do you know maybe why I have an error in this case

((comp int #(long 1234567890123)))

but not in this case

(int (long 1234567890123))

?

clojure.lang.RT.intCast checks the result of the cast and if it's different than passed throws exception. That's ok. But id doesn't in second case.

*than passed argument

I have a src/foo/bar.cljc -- there's someway to have the cljs version of foo/bar use a macro from the clj version of foo/bar

how do I do that without getting a circular dependency error?

@tsulej I get an error in both cases -- the number is too large to be cast to an int type

forgot to mention that unchecked-math was set to true

user=> (set! *unchecked-math* true)
true
user=> ((comp int #(long 1234567890123)))

IllegalArgumentException Value out of range for int: 1234567890123  clojure.lang.RT.intCast (RT.java:1205)
user=> (int (long 1234567890123))
1912276171

you should set the unboxed math warning

I forget the name

anything the behavior of int is different there because the higher order usage isn't inlined by the compiler

with unchecked-math, the int inlines as unchecked-int

you may also want to check out https://github.com/ztellman/primitive-math

aaaah, ok, thank you

((comp unchecked-int #(long 1234567890123))) works perfectly

and thanks for primitive-math this is the solution for my issues