New to travis and looking for nice pipelines to borrow from. Would like to have it cover publishing to clojars and dockerhub. Preferably even multiple jdk versions and images. Any suggestion is appreciated.

@seancorfield is there a simple way to make next.jdbc return column names (keys in the map) without namespacing them with the table name?

@vale check out https://github.com/seancorfield/next-jdbc/blob/master/doc/result-set-builders.md#rowbuilder-and-resultsetbuilder

Good Morning everybody. I am stuck like for two days and maybe somebody can help me? If this message is too long or I should ask elsewhere tell me 🙂 Thanks. Function below takes list of maps like

'({:country_code "CZ" :direction_id "1" :direction_name "1"
  :zone_id "x" :zone_name "x" :hub_id "x" :hub_name "x" 
:dealer_number "1025" :dealer_name "x"}
{:country_code "CZ" :direction_id "2" :direction_name "2"
:zone_id "x" :zone_name "x" :hub_id "x" :hub_name "x" 
:dealer_number "1025" :dealer_name "y"}
...)

(defn group-and-transform
  [datamart group-key]
  (for [m (group-by group-key datamart)]
    (let [info (extract-info 
                (apply merge-with (comp distinct flatten (into list)) (val m)) 
                group-key)
          kpis (calculate-kpis (val m))]
      (assoc
      (merge info kpis)
       group-key (key m)))))

(defn extract-info [data group-key]
  (let [country (if (list? (:country_code data)) 
                  (first (:country_code data))
                  (:country_code data))
        info (try
               (merge
                (case country
                  "CZ" {:country_code "" :direction_id "" :direction_name ""
                        :zone_id "" :zone_name "" :hub_id "" :hub_name "" 
                        :dealer_number "" :dealer_name "" :type group-key}
                  :default {})
                  (select-keys data
                               [:country_code :country_name :direction_id :direction_name
                                :zone_id :zone_name :hub_id :hub_name :dealer_number :dealer_name]))
               (catch Exception e (println (str "Error!" e))))]
    info))

(group-and-transform  (take 3 dm) :zone_id)
Error! java.lang.IllegalArgumentException: No matching clause: clojure.lang.LazySeq@896
({:kpi773 -100, :kpi755 -61.60700M, :kpi760 -61.60700M, :kpi768 1, :kpi807 0, :kpi1326 1, :kpi756 1, :kpi771 0, :zone_id "51", :kpi770 0, :kpi758 129.00M} {:kpi773 0, :hub_id "", :kpi755 0, :kpi760 0, :zone_name "Z≤na 52", :direction_name "", :kpi768 1, :type :zone_id, :kpi807 0, :kpi1326 1, :country_code "CZ", :country_name "Czech republic", :hub_name "", :kpi756 1, :dealer_number "20318039", :dealer_name "K N car spol. s r.o.", :kpi771 0, :zone_id "52", :direction_id "", :kpi770 0, :kpi758 57.00M})

In your function group-and-transform you have the expression (into list). list has no local value defined, so it refers to the function list in Clojure

If you were to write a custom CLI to better combine your git & ticketing system's API, would you reach for Clojure or ClojureScript? I'm having trouble deciding between the two.

Try evaluating small subexpressions and see what results you get for those smaller things, then build up from there.

For example, consider the difference betwen these:

user=> (into list [5 3])
Execution error (ClassCastException) at user/eval4356 (REPL:111).
class clojure.lang.PersistentList$Primordial cannot be cast to class clojure.lang.IPersistentCollection (clojure.lang.PersistentList$Primordial and clojure.lang.IPersistentCollection are in unnamed module of loader 'app')
user=> (into (list) [5 3])
(3 5)

flatten is often a bad idea, because it flattens arbitrary depths of data structures. It isn't clear to me what your intended behavior is that you want to get from this expression? (apply merge-with (comp distinct flatten (into list)) (val m))

Hello Andy, thank you for your interest! About the second function -> I need to merge some values accross multiple maps into one value. When I did only into list it ended like (1 1(1(1(1(1))) so I use flatten to get (1 1 1 1 1 1) and then distinct to get (1). I feel that this is not the best way how to do it but I did not find better yet. Edit: When I use (into (list))) and call (group-and-transform (take 2 dm) :zone_id) it leads to

Error printing return value (ClassCastException) at clojure.core/comp$fn (core.clj:2570).
clojure.lang.PersistentList$EmptyList

Error printing return value (StackOverflowError) at clojure.core/distinct$step$fn$fn (core.clj:5048).
null
class clojure.lang.ExceptionInfo

country (if (list? (:country_code data)) 
                  (first (:country_code data))
                  (:country_code data)) 

So my question ends like this: How to tranform this: '(((("CZ" "CZ") "CZ") "CZ"))) (but ideally also this): "CZ" into ("CZ") not producing lazy sequence. Thank you

Instead of answering how to transform this '(((("CZ" "CZ") "CZ") "CZ")))into what you want, I would ask you: what expression are you using that returns that result? Because I suspect that changing that expression to return ("CZ" "CZ" "CZ" "CZ") instead is very likely possible, and will avoid any need to add a step to flatten it.

So using merge-with with the function list is causing the first two maps to produce a list of two elements, then if you merge in a 3rd map, you get the list of two elements, plus the 3rd thing, in a list. and so on.

So if you had a different expression instead, that took a sequence of maps as input, you want it to return one map that contains keys that are in any of those individual input maps, and the values associated with the keys to be lists of the values from each input map?

Anyone familiar with clojure.test.check.clojure-test/defspec opaquely failing builds (e.g. a CircleCI one)? I can successfully run a defspec-based test in Emacs, or over a vanilla repl. In CI, the build just fails with Tests failed. and no other useful output

Does anyone here know of a macro or something else for retrying a failing serializable PostgreSQL transaction a specified max number of times?

I’m sure someone has already thought about Clojure over llvm (like Kotlin native). As I understand llvm could be very dynamic (as Swift for Tensorflow shows), so good REPL experience is possible. What do you think? Are there any progress in this area? Or what are the main obstacles?

https://clojure.org/about/rationale#_languages_and_platforms

llvm basically has none of the things called out as being provided by platforms there. there is nothing stopping you from implementing something like clojure on the llvm, but you will need to implement everything from scratch, and won't even be able to share your facilities with other languages that use llvm

the fact that llvm has vm in the name and so does the jvm causes people to think they are comparable, they are not, at all

basically a native code version of clojure throws away a lot of great things about clojure (good interop because everything on the jvm has the same memory model, layout, and calling conventions, a number of great gcs to choose from, etc)

llvm doesn't even provide a standard multithreading abstraction

Thank you. I mean yes, jvm and llvm are quite different, but the way I look at this is more similar to ClojureScript view. Something that’s useful in a context, where Clojure (or ClojureScript) is not applicable. Where you need fast startup time, low footprint, easy C, C++ interop or platforms like iOS. I know many of the things above are possible with graalvm or js, but the idea to have native binaries produced by mature established infrastructure like llvm seems very attractive to me

there are a number of projects that translate (some subset of) clojure code in to c++

https://ferret-lang.org/ is maybe the most maintained

Thank you, I’ll look at this

if given some kind of project requiring c++, I would likely write clojure to generate c++ instead of writing a clojure to c++ compiler

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

is my understand of reader conditionals correct when I'm thinking that these are processed at read-time by the language and the language picks only the parts that it's interested in and then processes the the forms?

e.g. you cannot write a macro that emits reader conditional forms which are then again processed by the reader?

the reader hands things out already filtered -- eval doesn't see them

that's what I mean yeah

that's right, you cannot write such a macro. conditionals are at read-time

phew, ok. thanks 🙂

you can ask the reader to preserve them

there is the reader-conditional function to create data representations of reader conditional forms. when is that useful? I have used it once when I generated code to disk which was then executed by some other process

that's what you get if you ask the reader to preserve

ah kewl, ok. thanks a lot