hugoduncan

$ lein2 modules test
------------------------------------------------------------------------
 Module build order:
   pallet-test-env-parent
   lein-test-env
   pallet-test-env
------------------------------------------------------------------------
 Building pallet-test-env-parent 0.1.2-SNAPSHOT (default)
------------------------------------------------------------------------
Using JLine for console I/O; install rlwrap for optimum experience.

hugoduncan

(ns pallet.crate.maven.support-test
  (:require
   [clojure.test :refer :all]
   [pallet.actions :refer [exec-checked-script exec-script* minimal-packages
                           package-manager]]
   [pallet.api :refer [converge group-spec plan-fn]]
   [pallet.build-actions :refer [build-actions build-session]]
   [pallet.core.api :refer [phase-errors]]
   [pallet.core.session :refer [with-session]]
   [pallet.crate :refer [is-64bit?]]

hugoduncan

(ns lein-test-env.plugin)

(def profiles
  {:aws {:pallet/test-env {}
         :dependencies '[[com.palletops/pallet-aws "0.2.0"]
                         [ch.qos.logback/logback-classic "1.1.1"]
                         [org.slf4j/jcl-over-slf4j "1.7.6"]]}})

(defn middleware
  "Middleware to add test-env profiles"

hugoduncan

(ns pallet.api-examples.plan
  "Examples of pallet's plan api"
  (:refer-clojure :exclude [sync])
  (:require
   [pallet.actions :refer [exec-script*]]
   [pallet.core.executor.ssh :refer [ssh-executor]]
   [pallet.plan :refer [execute-plan execute-plans plan-fn]]
   [pallet.session :as session :refer [set-user]]
   [pallet.user :refer [default-user]]
   [pallet.utils.async :refer [sync]]))

hugoduncan

Criterium - reliable micro-benchmarks

abstract

We’ve all timed a clojure function using clojure’s `time` macro, and then wondered why we don’t get stable results. We’ve seen Rich show off order of magnitude speed differences in new versions of clojure using the time macro. But what happens when you want reliable benchmarks, that can track small changes in timings? `time` falls short.

hugoduncan

Scaling the Infrastructure Continuum

The choices of infrastructure is ever increasing - dedicated servers, VPS, cloud images, virtualisation and containers.

Each of these choices has advantages that suit a particular use case. A single application, when viewed from the development and operation perspective has many use cases however. Local virtualisation makes good sense for development. Ephemeral cloud nodes fit well for testing. Dedicated hardware is good choice for production in constant load environments. An image based solution is great

hugoduncan

{-# LANGUAGE ScopedTypeVariables #-}

module Test () where

-- a type class to specify parsing and encoding of a type
-- (in reality I'm using hedn, which is similar to aeson)
class T a where
  f :: String -> a
  g :: a -> String

hugoduncan

(ns com.palletops.brick-server.fs-watch
  "Watch a filesystem path"
  (:require
   [clojure.java.io :refer [file]]
   [clojure.set :refer [map-invert]]
   [clojure.tools.logging :refer [debugf]]
   [clojure.string :refer [join]])
  (:import
   [java.nio.file
    ClosedWatchServiceException FileSystems Path StandardWatchEventKinds

hugoduncan

(defvar clojure-test-ediff-buffers nil)
(defun clojure-test-ediff-cleanup ()
  "A function for ediff-cleanup-hook, to cleanup the temporary ediff buffers"
  (mapc #'kill-buffer clojure-test-ediff-buffers))

(defconst re1
  "Expected \\(?:.\\|\n\\)*, got (not ([^ ]+ \\(\"\\(?:.\\|\n\\)+\"\\) \\(\"\\(?:.\\|\n\\)+\"\\)))"
  "Match predicate on strings test")
(defconst re2
  "Expected \\(?:.\\|\n\\)*, got (not ([^ ]+ \\(\\(?:.\\|\n\\)+\\) \\(\\(?:.\\|\n\\)+\\)))"

hugoduncan

(defmacro mfn*
  [[argv] body & args]
  (let [[argv] args]
    body))

(defmacro mfn
  "An macro form which looks like an anonymous function, and can be used to
  inline a function call "
  [args body & arg-vals]
  `(let [~args ~(vec arg-vals)]