kamituel

docker-machine

To run Docker on a Mac host you need docker-machine.

If, at some point, you’ll find out that the “marchine” is not running - and you’ll notice that when you’ll see messages like this:

Cannot connect to the Docker daemon. Is the docker daemon running on this host?

You can then start it:

$ docker-machine start $ docker-machine env

Container shell

To get a container shell prompt, you can launch a ‘bash’ command from within it:

$ docker exec -ti container-name bash

Okay, but how do I get container name?

$ docker ps

Copying files to/from container filesystem

$ docker cp src dest

For instance:

The majority of the time, querySelector() and querySelectorAll() are used to get hold of a DOM element. They both accept a CSS selectors.

It’s often forgotten, though, that major browsers do support the good, old XPath. And it can, sometimes, be extremely useful and powerful.

Example: you want to get a parent of a given node. So, having the h2, you want to get hold of div:

<div>   <h2 name=“Syntax”>Lorem Ipsum</h2> </div>

You cannot do that in CSS, unless you have a browser that supports CSS4 - and none of the major browsers support that as of this writing.

In CSS4, you will be some day able to write:

$div > *[name=‘Syntax’]

This would select a parent div element of an element that has a given name attribute.

In XPath, you can do that now:

//*[@name='Syntax']/..

Voila!

How do you use that selector in Javascript?

Steps to quickly iterate when writing integration tests, by avoiding JVM load overhead between each test run:

1. Add to project.clj

:dependencies [[lein-light-nrepl "0.1.0"]] :repl-options {:nrepl-middleware [lighttable.nrepl.handler/lighttable-ops]}

2. Run REPL

$ lein repl

It will print something similar to:

nREPL server started on port 56139 on host 127.0.0.1 - nrepl://127.0.0.1:56139 REPL-y 0.3.5, nREPL 0.2.6 Clojure 1.6.0

Note the port number - 56139 in this case.

3. In REPL, load clojure.test namespaces, and the namespace you want to test:

(require '[clojure.test :refer [run-tests test-vars]]) (require ‘test.something :reload-all)

The :reload-all option is handy, because it forces REPL to fully reload the namespace and all its dependencies.

4. Open LightTable and connect to this remote REPL. First, open “Connections” view, then click “Add Connection”, and then “Clojure (remote nREPL)”. Paste your port number in the dialog that will pop up.

5. Modify test file (in this example, the test/something.clj), then reload it in remote REPL by pressing ctrl + shift + enter (or cmd + shit + enter on Mac).

6. Run all tests for this namespace, or selected tests only:

(run-tests 'test.something) (test-vars [#'test.something/one-test-only])

Often I encounter an interesting library, which looks to be useful, but I have no immediate application for at the moment. Couple of weeks, months (years maybe) later, facing some problem, I recall that I've seen a tool which could help me solve it. But how was it called? Starring repos on Github is one way, of course, but it's pretty hard later to find the repo you need only by it's name, when this name is the one thing you don't recall ;)

This is an alphabetical ordered list, along with keywords, for an easier lookup later. I intend to update it regularly. No very well known libraries here, as I remember those anyway :)

ace - Ajax.org #Cloud9 #editor module.

annyang - A #javascript library for adding #voice commands to your site, using #speech recognition. #frontend

html2canvas - #screenshots with #javascript

json - a #CLI tool for working with JSON.

Minimist - parse command line arguments for #node.

Mori - #clojurescript's persistent data structures and supporting API from the comfort of vanilla @javascript. By David #Nolan.

node-exif - #node library to extract #exif #metadata from #images.

node-findit - walk a directory tree in #node.

Node Inspector - #node debugger basend on #blink #developertools.

oauth2orize - #oauth 2.0 authorization server toolkit for #node.

parse5 - WHATWG #html5 compliant parsing/serialization toolset for #node and #io.js.

rewire - easy dependency injection for #node #unit testing.

shortcuts - #keyboard #shortcuts in #javascript. #frontend

Skulpt - A #javascript implementation of the #python programming language.

Sweet.js - Macros for #javascript, can be used to improve syntax of the language.

waypoints - makes it easy to execute a function whenever you #scroll to an element. #javascript #frontend

I have been in a process of setting up an Atom Shell app, where I'd use Clojure Script for both browser (which is the backend in Atom Shell's nomenclature) and renderer sides.  Setting this up wasn't as straightforward as I'd hope to, but it can be done.

Using Clojure Script both for browser and renderer side would (hopefully) allowe me to reuse some of the code.

I used Leiningen with a lein-cljsbuild. I've set up two build profiles, one for the browser, another for renderer sides. The biggest challenge here was to make browser.js work with ecent Clojure Script (0.0-2913) and core.async. Depending on the optimization level and target set, I would repeatedly get errors such as:

goog.global.setTimeout undefined TypeError: cannot set property 'unicode' of undefined TypeError: undefined is not a function Cannot read property 'cljs$core$IFn$_invoke$arity$1' of undefined

Some of those errors were caused by core.async's go macro not being properly handled by Clojure Script / Google Closure combo.

Finally, I was able to find excellent blog post from Lauri Personen which showed the right path. In a nutshell, what needed to be done is to set :optimizations to :none, then :target to :nodejs and :output-dir to "out". With that build setup, instead of loading the compiled file directly, I needed to provide a small shim:

// Helper, generated by Clojure Script / Google Closure compiler, // which helps Node.js to load compiled files. require('../out/goog/bootstrap/nodejs') // Entry file which loads all JS files from the "out" directory. require('../js/browser.js') // File which contains the main method. require('../out/kamituel/gallery/browser/main') // Call the main method. console.log('shell.js dependencies loaded. Running _main().'); kamituel.gallery.browser.main._main()

Save this file as "js/shell.js" and point Atom Shell to it as a main file:

{ "name": "Gallery", "version": "0.0.1", "main": "js/shell.js" }

For a reference, the project.clj file:

;; cljs/browser - contains Atom Shell browser side namespaces (backend). ;; cljs/renderer - contains Atom Shell renderer side namespaces (frontend). ;; cljs/common - contains namespaces shared accross browser and renderer. ;; ;; Output files: ;; js/browser.js ;; js/renderer.js ;; out/ - browser's output files ;; ;; js/shell.js contains a shim which helps with loading compiled Clojure Script ;; browser code. (defproject gallery "0.0.1" :plugins [[lein-cljsbuild "1.0.6-SNAPSHOT"]] :dependencies [[org.clojure/clojure "1.6.0"] [org.clojure/clojurescript "0.0-2913"] [reagent "0.5.0-alpha3"] [org.clojure/core.async "0.1.346.0-17112a-alpha"]] :cljsbuild { :builds [ {:id "browser" :source-paths ["cljs/common" "cljs/browser"] :compiler { :output-to "js/browser.js" :optimizations :none :output-dir "out" :target :nodejs :pretty-print true}} {:id "renderer" :source-paths ["cljs/common" "cljs/renderer"] :compiler { :output-to "js/renderer.js" :optimizations :simple :pretty-print true}}]} :aliases {"watch" ["cljsbuild" "auto"]})

Occasionally, you might find yourself in a need of adding some library to the classpath of the running REPL instance. Furthermore, sometimes it's okay for this change to be temporary - library will not be used later.

This happened to me recently when I wanted to use criterium library to benchmark piece of my code.

First step (optional): Download the JAR you need.

Since library will be used one time only, it might make sense to download it, but without adding it as a dependency. You can use mvn command line tool do do the job:

$ mvn -DgroupId=criterium \ -DartifactId=criterium \ -Dversion=0.4.3 \ -DremoteRepositories="http://clojars.org/repo" \ dependency:get

It'll land in $HOME/.m2/repository obviously.

Second step: Add JAR to the classpath.

(-> (.getContextClassLoader (Thread/currentThread)) (.addURL (java.net.URL. "file:///Users/kls/.m2/repository/criterium/criterium/0.4.3/criterium-0.4.3.jar")))

Third step: Require namespace

(require 'criterium.core)

Fourth step: Use the newly loaded library

Postgresql tips

1. Expanded display

For a wide tables (with tens of columns) that wouldn't normally fit in one console line, you can enable extended display, which will show them, row by row, vertically.

db=# \x on

2. Enable logging of SQL statements

Scenario: you have an Android keystore (generated using keytool command, for instance), and you need to use it in an Javascript app. In my case, I had a signature generated using command line tools like openssl, and I wanted to verify it in the browser.

Solution: extract public and private keys. Use private key to sign some value in the command line, and public key to verify the signature - in the browser.

The following steps would give an overview of how it can work:

Step 0: Generate Android keystore

All the examples belowe assume the password "keystore" is used whenever prompted.

keytool -v \ -genkey \ -keystore kamituel.keystore \ -alias kamituel \ -keyalg RSA \ -keysize 2048 \ -validity 10000

Step 1: Convert JKS to PKCS12 structure

keytool \ -importkeystore \ -srckeystore kamituel.keystore \ -destkeystore kamituel.p12 \ -deststoretype PKCS12 \ -srcalias kamituel \ -deststorepass keystore \ -destkeypass keystore

Step 2: Export private key from the PKCS12 file

openssl pkcs12 \ -in kamituel.p12 \ -nodes \ -nocerts \ -out kamituel-private-key.pem

Step 3: Export certificate from the PKCS12 file

openssl pkcs12 \ -in kamituel.p12 \ -nokeys \ -out kamituel.pem

Step 4: Export public key from the PKCS12 file

openssl x509 \ -pubkey \ -noout \ -in kamituel.pem > kamituel-public.pem

Step 5: Sign the desired value

echo -n "some value" > value

openssl dgst \ -sha1 \ -sign kamituel-private-key.pem < value | xxd -p -c 256

Note: There is another command which could be used to sign a value (openssl rsautl -sign). It generates an invalid output (it's a raw signature, without a digest algorithm info in it), though, and will not work. See this excellent Stack Overflow answer for a great explanation.

Step 6: Verify it in the browser

vthis excear value = 'some value' // signed value , signature = '......' // ouput of command in step 5 , pubkey = '.......' // kamituel-public.pem , sig = new KJUR.crypto.Signature({"alg": "SHA1withRSA", "prov": "cryptojs/jsrsa"}) , key = KEYUTIL.getKey(pubkey) ; sig.initVerifyByPublicKey(key); sig.updateString(value); // Should be true. var isValid = sig.verify(signature);

To convert EPS file to an PNG bitmap, use the following command:

Pagination in Postgres:

SELECT *, COUNT(*) OVER() AS count FROM table WHERE ... LIMIT page_size OFFSET page_no * page_size;

Update remote

$ git fetch origin

Checkout remote branch

$ git checkout -b local-name origin/remote-name

Diff staged, but not commited changes

$ git diff [--staged | --cached]

Add files to the last commit (or edit message)

$ git commit --amend

Change message for any commit

$ git rebase -i PARENT_OF_FLAWED_COMMIT

Now change "pick" to "reword" for any commit you want to change message for. Save & quit. Now edit the message.

Branch visualization

$ git log --graph \ --oneline \ --decorate=[short|full] \ --all \ --simplify-by-decoration

Or:

$ gitx # Mac tool

Abort merge

$ git reset --hard HEAD

Skip staging area

$ git commit -a -m ""

Auto add all tracked files

$ git add -u

Undo "git add"

$ git rm --cached file_path

Undo "git commit"

$ git reset --soft HEAD^

Remove local branch

$ git branch -d branch_name

Remove remote branch

$ git push origin --delete branch_name

Rename current branch

$ git branch -m new_name

To rename remote branch, remove it and push again.

Apply last stash

$ git stash pop stash@{0}

Diff stash

$ git stash show -p stash@{0}

Checkout file at specified commit

$ git checkout commit file_path

Find commits containing file

$ git log --all -- file_path

List branches containing commit

$ git branch --contains REF

Push branch under different name

$ git push origin local_name:remote_name

Rebase feature branch based on master onto production

$ git rebase --onto production master

(details)

See which commits aren't yet pushed to the remote

$ git log origin/branch..branch

When “git branch -va” shows branches that are not present on the remote anymore

We all know Android applications are distributed as APK files, and that those files are a simple ZIP archives, but with changed extension. Peeking into them isn't obvious, though, because Java files inside are compiled and because XML files - most importantly manifest - are compiled as well.

So, we have an application on the phone. How do we look inside it?

First, we should copy it over from the phone to our computer. Let's use Google Keep as an example. To do that, we need to know the package name of this application:

$ adb shell pm list packages package:com.quickoffice.android package:com.google.android.keep ...

Once you find the package of interest in the list printed by the command above, you can get the path to the APK file:

$ adb shell pm path com.google.android.keep package:/data/app/com.google.android.keep-1.apk

Now let's get this file:

$ adb pull /data/app/com.google.android.keep-1.apk 6777 KB/s (6344188 bytes in 0.914s)

Great, we have now Google Keep APK on our computer. 

Let's now peek inside. To view the manifest file, use this command:

$ aapt -l a com.google.android.keep-1.apk

This command will output, among others, the manifest file. It's in a format where lines starting with E: are XML elements, and those starting with A: are attributes.

Okay, so now we can decompile the code. In order to do that, we need two tools: dex2jar and JD-GUI.

$ d2j-dex2jar -f -o output.jar com.google.android.keep-1.apk

We all know all those useful Bash variables: $0, $1, $2, ..., $@, $*, $# or $?.

But there is another - $_ - which is the last parameter of the most recently invoked command.

More formally:

At shell startup, set to the absolute pathname used to invoke the shell or shell script being executed as passed in the environment or argument list. Subsequently, expands to the last argument to the previous command, after expansion.

What so nice about it? Well, how many times have you invoked command similar to:

$ mkdir my_next_great_project $ cd my_next_great_project

Now you can simply do:

$ mkdir my_next_great_project $ cd $_

Or even:

Navigation

ESC - Move focus to editor

Ctrl + N - open class by name

Ctrl + Shift + N - open any file by name

Ctrl + Shift + A - Invoke action

Shift, Shift - Search everywhere

Ctrl + E - Recent files

Ctrl + Shift + E - Recently edited files

Ctrl + F12 - Functions and symbols from current file (outline). Once it pops up, start typing name to filter.

F2 - Jump to next error in current file

Editor

Ctrl + W - Incremental selection

Ctrl + Y - Delete line

Shift + Alt + Up - Move line up

Shift + Alt + Down - Move line down

Shift + F6 - Refactor/rename

Ctrl + / - Comment current line or selection

Ctrl + Shift + / - Uncomment current line or selection

Ctrl + Q - Show documentation for function

Ctrl + B - Go to delcaration

Ctrl + Shift + I - Quick definition lookup

Ctrl + Alt + T - Surround with template

Ctrl + J - Add template

REPL and structural editing

Those shortcuts you'll have to assign yourself. Default assignment for structural editing is described at the bottom of Cursive's page. Here's my setup:

⌘ / Ctrl + Alt + Shift + K - Load current file in REPL

Alt + A, N - Slurp backwards

Alt + A M - Slurp forward

Alt + A, , - Barf backward

Alt + A, . - Barf forward

Ctrl + ; - Splice

Ctrl + Shift + ; - Raise

Alt + , - Kill

I found myself in need of (once again) getting familiar with database layout I haven't designed myself (it's a fairly complex, Postgres database this time).

SchemaSpy, a small, open source tool, solved this issue well (and quickly).

General steps are:

Get SchemaSpy from project SourceForge page.

Make sure you have Java and GraphViz installed.

Find and download JDBC driver for the database of your choice. For PostgreSQL visit this page.

Run SchemaSpy.

java -jar schemaSpy_5.0.0.jar \ -t pgsql \ -db my_database \ -s schema_name \ -u username \ -p passphrase \ -o output_dir \ -host localhost \ -dp postgresql-9.3-1101.jdbc41.jar

It might be a bit overwhelming, at first, to get to know what methods are available when writing AngularJS end-to-end tests using Protractor/Selenium. That's partly because Selenium isn't Javascript native library, so you might find documentation to some other flavour of it, and partly because Protractor adds some Angular-specific methods to Selenium.

First and foremost documentation is being hosten on Protactors Github page. 

There, you'll find list of methods (with examples) for at least those classes:

webdriver.Webdriver - In your tests, it's bound to 'browser' object.

ProtractorBy - contains Angular specific "locator" methods. You should remember that this list is not complete, because you can of course use standard Selenium methods as well, like "by.id" or "by.linkText". You'll find list of all standard methods on Selenium website.

If you use string as an array indice, it will coerce to a number and work, i.e.:

var array = ['a', 'b', 'c']; array['1'] === 'b'

What's less common is to use object as an array indice: