How does one pre-load a clojure file in the leiningen repl? - clojure

I have some clojure functions that I would like pre-loaded when I start the clojure REPL. The functions aren't much use unless you are using them within the context of a REPL.
If it helps, I generally use leiningen to start a clojure REPL for me.
How can I tell clojure (or leiningen, if it's not available through flat clojure) to pre-load a clojure file containing these definitions for me?

There are several ways to do this described in the leiningen sample project
one of my favorite methods is so put the code you want in the default repl namespace into
/path/to/project/dev/user.clj:
(ns user)
(def foo 42)
and add a line like this into the project.clj file:
(defproject hello "0.1.0-SNAPSHOT"
:description "FIXME: write description"
:url "http://example.com/FIXME"
:license {:name "Eclipse Public License"
:url "http://www.eclipse.org/legal/epl-v10.html"}
:dependencies [[org.clojure/clojure "1.5.1"]]
:source-paths ["dev"])
This makes it clear that this is for dev while still getting it loaded into the default namespace.
When you run nrepl-jack-in form emacs or "lein repl" form the shell, you should be greeted with a user> namespace with your code loaded:
; nREPL 0.1.6
user> foo
42

Related

Hugsql can not read my sql file

I am truely lost here. I have a very simple application. All it does is to insert a user into the user table in my Database. I using Postgres. The code is
(ns signupper.db (:require [hugsql.core :as hugsql]))
(hugsql/def-db-fns "sql/q.sql")
Inside the direvtory where db.clj is I made a directory called sql and inside of it there is a file called q.sql.
When I ran my REPL and type (require '[signupper.db :as db]) I get the following error message:
CompilerException clojure.lang.ExceptionInfo: Can not read file: sql/q.sql {}, compiling:(signupper/db.clj:4:1)
Any one has any idea?
Thanks.
Your sql directory needs to be on the path. Please check your project.clj file, under resource-paths, and verify your sql directory is accessible via one of the paths there stated.
If not, you might either move your sql directory or include the path into the resource-paths entry.
If you are using Leiningen you should add your sql folder under the :resource-paths key to your project.clj like this:
(defproject test-project "0.1.0-SNAPSHOT"
:description "FIXME: write description"
:url "http://example.com/FIXME"
:license {:name "Eclipse Public License"
:url "http://www.eclipse.org/legal/epl-v10.html"}
:dependencies [[org.clojure/clojure "1.8.0"]]
:main ^:skip-aot test-project.core
:target-path "target/%s"
:resource-paths ["sql" "resources"] ; <-- here
:profiles {:uberjar {:aot :all}})
Hugsql expects the path relative to the directory on the classpath. (This is the same as Clojure namespaces.)
So, your code should look like this:
(ns signupper.db (:require [hugsql.core :as hugsql]))
(hugsql/def-db-fns "signupper/db/sql/q.sql")
See the example in the docs: https://www.hugsql.org/#start-sql
My experience :
If your path is like this :
"/home/user/Desktop/sql/q.sql"
project.clj must be like this -->
:resource-paths ["/home/user/Desktop/sql" "other resources"]
and core.clj -->
(hugsql/def-db-fns "q.sql")
But if you write in core.clj
(hugsql/def-db-fns "/home/user/Desktop/sql/q.sql")
Computer sees like this :
"/home/user/Desktop/sql/home/user/Desktop/sql/q.sql"
Just had the same issue.
Apparently, the function def-db-fns starts the path from the folder src and not the root of the project.
I think one way you can solve this is putting your file inside the src directory, and call the function with
(hugsql/def-db-fns "q.sql")

Clojure compile time

I'm testing out Clojure for the first time, and wanted to write a simple websocket client application. I've found the library https://github.com/stylefruits/gniazdo and got the code to work (using lein run). However, compiling the code into a jar (either with lein jar or lein uberjar is either stuck or takes ages (aborted after ~1h))
steps:
lein new app testing
modified src/testing/core.clj and project.clj (see below)
lein jar (or lein uberjar)
For simplicity I've have this very simple code, which already takes ages to compile into a jar:
(ns testing.core
(:gen-class))
(require '[gniazdo.core :as ws])
(def socket
(ws/connect
"wss://some.url.com/"))
(defn -main
"I don't do a whole lot ... yet."
[& args]
(ws/close socket))
project.clj:
(defproject testing "0.1.0-SNAPSHOT"
:description "FIXME: write description"
:url "http://example.com/FIXME"
:license {:name "Eclipse Public License"
:url "http://www.eclipse.org/legal/epl-v10.html"}
:dependencies [[org.clojure/clojure "1.8.0"]
[stylefruits/gniazdo "1.0.1"]]
:main ^:skip-aot testing.core
:aot [testing.core]
:target-path "target/%s"
:profiles {:uberjar {:aot :all}})
output of running lein jar:
$lein jar
Compiling testing.core
2017-12-11 14:15:14.813:INFO::main: Logging initialized #1352ms
and nothing aferwards. Is this normal behaviour (and it just takes ages to compile) or am I missing something here? Clojure looks very interesting, but if it takes even for such a small program hours to compile, deployment could be a problem in my circumstances.
When that namespace is compiled Ahead of Time (:aot [testing.core] in your project.clj), this code gets evaluated during compilation:
(def socket
(ws/connect "wss://some.url.com/"))
This is likely what's causing the hang. The compiler never moves on from this because it has made a blocking call.
You could remove the :aot directive if you don't need it (and you probably don't). I think this can be a somewhat confusing default when creating a new Leiningen project.
You could do something like wrap that socket/conn in a delay:
(def socket (delay (ws/connect "wss://some.url.com/")))
And then deref/# it wherever you need the value. This just delays evaluation of ws/connect until you ask.

importing java lib in clojure, how does it work?

I'm trying to build my first clojure leiningen project but I have an issue using a specific java class in my code.
While coding, I was looking for a specific functionnality and found out about DatatypeConverter (http://docs.oracle.com/javase/7/docs/api/javax/xml/bind/DatatypeConverter.html).
Then I had to figure how to import the library. I don't know anything about Maven but I ended up somewhat (educated?) guessing I should looking for the library there https://search.maven.org/.
So there is what I ended up writing for my project.clj file:
(defproject game-backend "0.1.0-SNAPSHOT"
:description "FIXME: write description"
:url "http://example.com/FIXME"
:license {:name "Eclipse Public License"
:url "http://www.eclipse.org/legal/epl-v10.html"}
:main game-backend.core
:dependencies [
[org.clojure/clojure "1.8.0"]
[javax.xml.bind/jaxb-api "2.2.12"]
])
and here is my ns macro call in my core.clj file:
(ns game-backend.core
(:require [clojure.java.io])
(:import
(java.security DigestInputStream)
(java.io FileInputStream)
(javax.xml.bind DataTypeConverter)
)
)
and when I tun lein run I get the following error (a package was downloaded at some point in time): Exception in thread "main" java.lang.ClassNotFoundException: javax.xml.bind.DataTypeConverter, compiling:(game_backend/core.clj:1:1)
I(m have no idea how many steps I've done wrong (all of them?). Can you please enlight me on how it should be done?
Try a lowercase 't' DatatypeConverter
(ns game-backend.core
(:require [clojure.java.io])
(:import
(java.security DigestInputStream)
(java.io FileInputStream)
(javax.xml.bind DatatypeConverter)
)
)
Take a look inside your maven repository (.m2 directory). You will be able to find the jar file there. Then look at the .class files in that jar.
DatatypeConverter.class
That's one way to find that you should be using a lowercase 't'.
Also you can add multiple classes of a package:
(:import (java.io File Bits BufferedInputStream))

Running a leiningen Clojure project in LightTable

I'm trying to use Clojure to run my Leiningen project. Even though LightTable says it's connected in the connections pane, it won't execute unless I call the main function manually.
project.clj:
(defproject lein-test "0.1.0-SNAPSHOT"
:description "FIXME: write description"
:url "http://example.com/FIXME"
:license {:name "Eclipse Public License"
:url "http://www.eclipse.org/legal/epl-v10.html"}
:dependencies [[org.clojure/clojure "1.5.1"]]
:main lein-test.core)
core.clj:
(ns lein-test.core)
(defn -main [& args]
(println "asdf"))
When I eval the entire file with ctrl+shift+enter, I see nil by the ns and nothing in the console. If I add
(-main)
then the console shows 'asdf'. lein run in the command prompt yields the expected behavior. What am I missing?
Control+Shift+Enter evaluates the namespace.
Evaluating a namespace should not run any of its functions, unless you call them at the top level. In a given codebase there should ideally be only one function that is called at the top level (conventionally, the -main function), and one should set this up not by calling it in the namespace code, but as you have, via configuration.
Everything is working as expected here. You can put a call to (-main) in a commented block or a temporary section of the file for convenience while developing, or call it directly from the repl interface.

Project-level Leiningen Plugin

In Leiningen versions 1.x.x I was able to define a lein foo task valid only in a single project by putting the following in that project's project.clj:
(defproject tester "0.1.0-SNAPSHOT"
:description "FIXME: write description"
:url "http://example.com/FIXME"
:license {:name "Eclipse Public License"
:url "http://www.eclipse.org/legal/epl-v10.html"}
:dependencies [[org.clojure/clojure "1.4.0"]])
;; Create a task, "foo"
(ns leiningen.foo
(:require (leiningen [uberjar :as uberjar])))
(defn foo [project & args]
(println "Do something here first, then make the uberjar.")
(uberjar/uberjar project))
You can get a little more information about this here:
http://nakkaya.com/2010/02/25/writing-leiningen-plugins-101/
In 2.x.x, I'm not able to do this anymore (i.e., I get 'foo' is not a task. It seems way, way overkill for me to have to start a separate project for this task. Is it still possible to define a task within project.clj for leiningen 2.x.x?
The short answer is "no", but it is still fairly easy to define a project level task: Add :eval-in-leiningen true to your defproject definition and move the task definition to src/leiningen/foo.clj.
You can do this by using .lein-classpath to point to a directory outside of src containing the tasks. For example, if you have the plugin in src/leiningen/foo.clj, you can do, at the project root:
$ mkdir tasks
$ mv src/leiningen tasks/
$ echo tasks > .lein-classpath
The reason you might want to avoid :eval-in-leiningen true is that it has some funny behaviors when you're trying to do AOT compilation for a main class. Specifically, you get:
Compilation failed: java.io.IOException: No such file or directory, compiling:(testproj/core.clj:1)
When trying to compile/run a even a simple test example. More information at:
https://github.com/technomancy/leiningen/issues/769