Dicanio Rol

Dicanio Rol

1594319580

How to Save and Load of Keras Sequential and Functional Models - Part II

Part I of this series, we have learned different approaches to save architecture only or save weights only or entire Keras model. However, we had covered only small part of saving entire model. In this article we will focus only on saving entire model when we have a custom metric/loss or a custom layer.

Outline of this article is as follows

  • Overview of saving and loading entire model
  • Saving entire model before or after training
  • Saving entire model during training
  • Saving entire model with a custom metric/loss
  • Saving entire model with custom layer

1. Overview of saving and loading entire model

As mentioned in Part I of this series, entire Keras model consist(the following is from TensorFlow website)

An architecture, or configuration, which specifies what layers the model contain, and how they’re connected
set of weights values (the “state of the model”)
An optimizer****state_ (defined by compiling the model)_
A set of losses and metrics (defined by compiling the model)

Entire Keras model can be saved to a disk in two formats (i) TensorFlow SavedModel ( tf ) format, and (ii) H5 format.

Entire Keras model can be saved either during training or before/after training the model. We will see see more details and examples in the following sections.

How to save entire model?

Entire Keras model can be saved using Saved model API by model.save(‘MyModel’,save_format='tf') or model.save('MyModel_h5',save_format='h5') . The tf format is default which means if you don’t provide save_format argument, then the model is saved in TensorFlow SavedModel tf format.

Why do we need to save entire model?

  • Sharing entire model is simple and error prone. You can share it with your team or client so that they can reproduce exactly same result as you are
  • As saving entire model includes optimizer state, you can restart the training where you left off.
  • Entire model can be easily converted to TFLite format so that you can deploy the model on mobile devices
  • Entire model can be converted to TensorFlow.js Layers format, which can be loaded directly into TensorFlow.js for inference or for further training.

#keras #machine-learning #tensorflow #deep-learning

What is GEEK

Buddha Community

How to Save and Load of Keras Sequential and Functional Models - Part II

Dotnet Script: Run C# Scripts From The .NET CLI

dotnet script

Run C# scripts from the .NET CLI, define NuGet packages inline and edit/debug them in VS Code - all of that with full language services support from OmniSharp.

NuGet Packages

NameVersionFramework(s)
dotnet-script (global tool)Nugetnet6.0, net5.0, netcoreapp3.1
Dotnet.Script (CLI as Nuget)Nugetnet6.0, net5.0, netcoreapp3.1
Dotnet.Script.CoreNugetnetcoreapp3.1 , netstandard2.0
Dotnet.Script.DependencyModelNugetnetstandard2.0
Dotnet.Script.DependencyModel.NugetNugetnetstandard2.0

Installing

Prerequisites

The only thing we need to install is .NET Core 3.1 or .NET 5.0 SDK.

.NET Core Global Tool

.NET Core 2.1 introduced the concept of global tools meaning that you can install dotnet-script using nothing but the .NET CLI.

dotnet tool install -g dotnet-script

You can invoke the tool using the following command: dotnet-script
Tool 'dotnet-script' (version '0.22.0') was successfully installed.

The advantage of this approach is that you can use the same command for installation across all platforms. .NET Core SDK also supports viewing a list of installed tools and their uninstallation.

dotnet tool list -g

Package Id         Version      Commands
---------------------------------------------
dotnet-script      0.22.0       dotnet-script
dotnet tool uninstall dotnet-script -g

Tool 'dotnet-script' (version '0.22.0') was successfully uninstalled.

Windows

choco install dotnet.script

We also provide a PowerShell script for installation.

(new-object Net.WebClient).DownloadString("https://raw.githubusercontent.com/filipw/dotnet-script/master/install/install.ps1") | iex

Linux and Mac

curl -s https://raw.githubusercontent.com/filipw/dotnet-script/master/install/install.sh | bash

If permission is denied we can try with sudo

curl -s https://raw.githubusercontent.com/filipw/dotnet-script/master/install/install.sh | sudo bash

Docker

A Dockerfile for running dotnet-script in a Linux container is available. Build:

cd build
docker build -t dotnet-script -f Dockerfile ..

And run:

docker run -it dotnet-script --version

Github

You can manually download all the releases in zip format from the GitHub releases page.

Usage

Our typical helloworld.csx might look like this:

Console.WriteLine("Hello world!");

That is all it takes and we can execute the script. Args are accessible via the global Args array.

dotnet script helloworld.csx

Scaffolding

Simply create a folder somewhere on your system and issue the following command.

dotnet script init

This will create main.csx along with the launch configuration needed to debug the script in VS Code.

.
├── .vscode
│   └── launch.json
├── main.csx
└── omnisharp.json

We can also initialize a folder using a custom filename.

dotnet script init custom.csx

Instead of main.csx which is the default, we now have a file named custom.csx.

.
├── .vscode
│   └── launch.json
├── custom.csx
└── omnisharp.json

Note: Executing dotnet script init inside a folder that already contains one or more script files will not create the main.csx file.

Running scripts

Scripts can be executed directly from the shell as if they were executables.

foo.csx arg1 arg2 arg3

OSX/Linux

Just like all scripts, on OSX/Linux you need to have a #! and mark the file as executable via chmod +x foo.csx. If you use dotnet script init to create your csx it will automatically have the #! directive and be marked as executable.

The OSX/Linux shebang directive should be #!/usr/bin/env dotnet-script

#!/usr/bin/env dotnet-script
Console.WriteLine("Hello world");

You can execute your script using dotnet script or dotnet-script, which allows you to pass arguments to control your script execution more.

foo.csx arg1 arg2 arg3
dotnet script foo.csx -- arg1 arg2 arg3
dotnet-script foo.csx -- arg1 arg2 arg3

Passing arguments to scripts

All arguments after -- are passed to the script in the following way:

dotnet script foo.csx -- arg1 arg2 arg3

Then you can access the arguments in the script context using the global Args collection:

foreach (var arg in Args)
{
    Console.WriteLine(arg);
}

All arguments before -- are processed by dotnet script. For example, the following command-line

dotnet script -d foo.csx -- -d

will pass the -d before -- to dotnet script and enable the debug mode whereas the -d after -- is passed to script for its own interpretation of the argument.

NuGet Packages

dotnet script has built-in support for referencing NuGet packages directly from within the script.

#r "nuget: AutoMapper, 6.1.0"

package

Note: Omnisharp needs to be restarted after adding a new package reference

Package Sources

We can define package sources using a NuGet.Config file in the script root folder. In addition to being used during execution of the script, it will also be used by OmniSharp that provides language services for packages resolved from these package sources.

As an alternative to maintaining a local NuGet.Config file we can define these package sources globally either at the user level or at the computer level as described in Configuring NuGet Behaviour

It is also possible to specify packages sources when executing the script.

dotnet script foo.csx -s https://SomePackageSource

Multiple packages sources can be specified like this:

dotnet script foo.csx -s https://SomePackageSource -s https://AnotherPackageSource

Creating DLLs or Exes from a CSX file

Dotnet-Script can create a standalone executable or DLL for your script.

SwitchLong switchdescription
-o--outputDirectory where the published executable should be placed. Defaults to a 'publish' folder in the current directory.
-n--nameThe name for the generated DLL (executable not supported at this time). Defaults to the name of the script.
 --dllPublish to a .dll instead of an executable.
-c--configurationConfiguration to use for publishing the script [Release/Debug]. Default is "Debug"
-d--debugEnables debug output.
-r--runtimeThe runtime used when publishing the self contained executable. Defaults to your current runtime.

The executable you can run directly independent of dotnet install, while the DLL can be run using the dotnet CLI like this:

dotnet script exec {path_to_dll} -- arg1 arg2

Caching

We provide two types of caching, the dependency cache and the execution cache which is explained in detail below. In order for any of these caches to be enabled, it is required that all NuGet package references are specified using an exact version number. The reason for this constraint is that we need to make sure that we don't execute a script with a stale dependency graph.

Dependency Cache

In order to resolve the dependencies for a script, a dotnet restore is executed under the hood to produce a project.assets.json file from which we can figure out all the dependencies we need to add to the compilation. This is an out-of-process operation and represents a significant overhead to the script execution. So this cache works by looking at all the dependencies specified in the script(s) either in the form of NuGet package references or assembly file references. If these dependencies matches the dependencies from the last script execution, we skip the restore and read the dependencies from the already generated project.assets.json file. If any of the dependencies has changed, we must restore again to obtain the new dependency graph.

Execution cache

In order to execute a script it needs to be compiled first and since that is a CPU and time consuming operation, we make sure that we only compile when the source code has changed. This works by creating a SHA256 hash from all the script files involved in the execution. This hash is written to a temporary location along with the DLL that represents the result of the script compilation. When a script is executed the hash is computed and compared with the hash from the previous compilation. If they match there is no need to recompile and we run from the already compiled DLL. If the hashes don't match, the cache is invalidated and we recompile.

You can override this automatic caching by passing --no-cache flag, which will bypass both caches and cause dependency resolution and script compilation to happen every time we execute the script.

Cache Location

The temporary location used for caches is a sub-directory named dotnet-script under (in order of priority):

  1. The path specified for the value of the environment variable named DOTNET_SCRIPT_CACHE_LOCATION, if defined and value is not empty.
  2. Linux distributions only: $XDG_CACHE_HOME if defined otherwise $HOME/.cache
  3. macOS only: ~/Library/Caches
  4. The value returned by Path.GetTempPath for the platform.

 

Debugging

The days of debugging scripts using Console.WriteLine are over. One major feature of dotnet script is the ability to debug scripts directly in VS Code. Just set a breakpoint anywhere in your script file(s) and hit F5(start debugging)

debug

Script Packages

Script packages are a way of organizing reusable scripts into NuGet packages that can be consumed by other scripts. This means that we now can leverage scripting infrastructure without the need for any kind of bootstrapping.

Creating a script package

A script package is just a regular NuGet package that contains script files inside the content or contentFiles folder.

The following example shows how the scripts are laid out inside the NuGet package according to the standard convention .

└── contentFiles
    └── csx
        └── netstandard2.0
            └── main.csx

This example contains just the main.csx file in the root folder, but packages may have multiple script files either in the root folder or in subfolders below the root folder.

When loading a script package we will look for an entry point script to be loaded. This entry point script is identified by one of the following.

  • A script called main.csx in the root folder
  • A single script file in the root folder

If the entry point script cannot be determined, we will simply load all the scripts files in the package.

The advantage with using an entry point script is that we can control loading other scripts from the package.

Consuming a script package

To consume a script package all we need to do specify the NuGet package in the #loaddirective.

The following example loads the simple-targets package that contains script files to be included in our script.

#load "nuget:simple-targets-csx, 6.0.0"

using static SimpleTargets;
var targets = new TargetDictionary();

targets.Add("default", () => Console.WriteLine("Hello, world!"));

Run(Args, targets);

Note: Debugging also works for script packages so that we can easily step into the scripts that are brought in using the #load directive.

Remote Scripts

Scripts don't actually have to exist locally on the machine. We can also execute scripts that are made available on an http(s) endpoint.

This means that we can create a Gist on Github and execute it just by providing the URL to the Gist.

This Gist contains a script that prints out "Hello World"

We can execute the script like this

dotnet script https://gist.githubusercontent.com/seesharper/5d6859509ea8364a1fdf66bbf5b7923d/raw/0a32bac2c3ea807f9379a38e251d93e39c8131cb/HelloWorld.csx

That is a pretty long URL, so why don't make it a TinyURL like this:

dotnet script https://tinyurl.com/y8cda9zt

Script Location

A pretty common scenario is that we have logic that is relative to the script path. We don't want to require the user to be in a certain directory for these paths to resolve correctly so here is how to provide the script path and the script folder regardless of the current working directory.

public static string GetScriptPath([CallerFilePath] string path = null) => path;
public static string GetScriptFolder([CallerFilePath] string path = null) => Path.GetDirectoryName(path);

Tip: Put these methods as top level methods in a separate script file and #load that file wherever access to the script path and/or folder is needed.

REPL

This release contains a C# REPL (Read-Evaluate-Print-Loop). The REPL mode ("interactive mode") is started by executing dotnet-script without any arguments.

The interactive mode allows you to supply individual C# code blocks and have them executed as soon as you press Enter. The REPL is configured with the same default set of assembly references and using statements as regular CSX script execution.

Basic usage

Once dotnet-script starts you will see a prompt for input. You can start typing C# code there.

~$ dotnet script
> var x = 1;
> x+x
2

If you submit an unterminated expression into the REPL (no ; at the end), it will be evaluated and the result will be serialized using a formatter and printed in the output. This is a bit more interesting than just calling ToString() on the object, because it attempts to capture the actual structure of the object. For example:

~$ dotnet script
> var x = new List<string>();
> x.Add("foo");
> x
List<string>(1) { "foo" }
> x.Add("bar");
> x
List<string>(2) { "foo", "bar" }
>

Inline Nuget packages

REPL also supports inline Nuget packages - meaning the Nuget packages can be installed into the REPL from within the REPL. This is done via our #r and #load from Nuget support and uses identical syntax.

~$ dotnet script
> #r "nuget: Automapper, 6.1.1"
> using AutoMapper;
> typeof(MapperConfiguration)
[AutoMapper.MapperConfiguration]
> #load "nuget: simple-targets-csx, 6.0.0";
> using static SimpleTargets;
> typeof(TargetDictionary)
[Submission#0+SimpleTargets+TargetDictionary]

Multiline mode

Using Roslyn syntax parsing, we also support multiline REPL mode. This means that if you have an uncompleted code block and press Enter, we will automatically enter the multiline mode. The mode is indicated by the * character. This is particularly useful for declaring classes and other more complex constructs.

~$ dotnet script
> class Foo {
* public string Bar {get; set;}
* }
> var foo = new Foo();

REPL commands

Aside from the regular C# script code, you can invoke the following commands (directives) from within the REPL:

CommandDescription
#loadLoad a script into the REPL (same as #load usage in CSX)
#rLoad an assembly into the REPL (same as #r usage in CSX)
#resetReset the REPL back to initial state (without restarting it)
#clsClear the console screen without resetting the REPL state
#exitExits the REPL

Seeding REPL with a script

You can execute a CSX script and, at the end of it, drop yourself into the context of the REPL. This way, the REPL becomes "seeded" with your code - all the classes, methods or variables are available in the REPL context. This is achieved by running a script with an -i flag.

For example, given the following CSX script:

var msg = "Hello World";
Console.WriteLine(msg);

When you run this with the -i flag, Hello World is printed, REPL starts and msg variable is available in the REPL context.

~$ dotnet script foo.csx -i
Hello World
>

You can also seed the REPL from inside the REPL - at any point - by invoking a #load directive pointed at a specific file. For example:

~$ dotnet script
> #load "foo.csx"
Hello World
>

Piping

The following example shows how we can pipe data in and out of a script.

The UpperCase.csx script simply converts the standard input to upper case and writes it back out to standard output.

using (var streamReader = new StreamReader(Console.OpenStandardInput()))
{
    Write(streamReader.ReadToEnd().ToUpper());
}

We can now simply pipe the output from one command into our script like this.

echo "This is some text" | dotnet script UpperCase.csx
THIS IS SOME TEXT

Debugging

The first thing we need to do add the following to the launch.config file that allows VS Code to debug a running process.

{
    "name": ".NET Core Attach",
    "type": "coreclr",
    "request": "attach",
    "processId": "${command:pickProcess}"
}

To debug this script we need a way to attach the debugger in VS Code and the simplest thing we can do here is to wait for the debugger to attach by adding this method somewhere.

public static void WaitForDebugger()
{
    Console.WriteLine("Attach Debugger (VS Code)");
    while(!Debugger.IsAttached)
    {
    }
}

To debug the script when executing it from the command line we can do something like

WaitForDebugger();
using (var streamReader = new StreamReader(Console.OpenStandardInput()))
{
    Write(streamReader.ReadToEnd().ToUpper()); // <- SET BREAKPOINT HERE
}

Now when we run the script from the command line we will get

$ echo "This is some text" | dotnet script UpperCase.csx
Attach Debugger (VS Code)

This now gives us a chance to attach the debugger before stepping into the script and from VS Code, select the .NET Core Attach debugger and pick the process that represents the executing script.

Once that is done we should see our breakpoint being hit.

Configuration(Debug/Release)

By default, scripts will be compiled using the debug configuration. This is to ensure that we can debug a script in VS Code as well as attaching a debugger for long running scripts.

There are however situations where we might need to execute a script that is compiled with the release configuration. For instance, running benchmarks using BenchmarkDotNet is not possible unless the script is compiled with the release configuration.

We can specify this when executing the script.

dotnet script foo.csx -c release

 

Nullable reference types

Starting from version 0.50.0, dotnet-script supports .Net Core 3.0 and all the C# 8 features. The way we deal with nullable references types in dotnet-script is that we turn every warning related to nullable reference types into compiler errors. This means every warning between CS8600 and CS8655 are treated as an error when compiling the script.

Nullable references types are turned off by default and the way we enable it is using the #nullable enable compiler directive. This means that existing scripts will continue to work, but we can now opt-in on this new feature.

#!/usr/bin/env dotnet-script

#nullable enable

string name = null;

Trying to execute the script will result in the following error

main.csx(5,15): error CS8625: Cannot convert null literal to non-nullable reference type.

We will also see this when working with scripts in VS Code under the problems panel.

image

Download Details:
Author: filipw
Source Code: https://github.com/filipw/dotnet-script
License: MIT License

#dotnet  #aspdotnet  #csharp 

Sasha  Roberts

Sasha Roberts

1659500100

Reform: Form Objects Decoupled From Models In Ruby

Reform

Form objects decoupled from your models.

Reform gives you a form object with validations and nested setup of models. It is completely framework-agnostic and doesn't care about your database.

Although reform can be used in any Ruby framework, it comes with Rails support, works with simple_form and other form gems, allows nesting forms to implement has_one and has_many relationships, can compose a form from multiple objects and gives you coercion.

Full Documentation

Reform is part of the Trailblazer framework. Full documentation is available on the project site.

Reform 2.2

Temporary note: Reform 2.2 does not automatically load Rails files anymore (e.g. ActiveModel::Validations). You need the reform-rails gem, see Installation.

Defining Forms

Forms are defined in separate classes. Often, these classes partially map to a model.

class AlbumForm < Reform::Form
  property :title
  validates :title, presence: true
end

Fields are declared using ::property. Validations work exactly as you know it from Rails or other frameworks. Note that validations no longer go into the model.

The API

Forms have a ridiculously simple API with only a handful of public methods.

  1. #initialize always requires a model that the form represents.
  2. #validate(params) updates the form's fields with the input data (only the form, not the model) and then runs all validations. The return value is the boolean result of the validations.
  3. #errors returns validation messages in a classic ActiveModel style.
  4. #sync writes form data back to the model. This will only use setter methods on the model(s).
  5. #save (optional) will call #save on the model and nested models. Note that this implies a #sync call.
  6. #prepopulate! (optional) will run pre-population hooks to "fill out" your form before rendering.

In addition to the main API, forms expose accessors to the defined properties. This is used for rendering or manual operations.

Setup

In your controller or operation you create a form instance and pass in the models you want to work on.

class AlbumsController
  def new
    @form = AlbumForm.new(Album.new)
  end

This will also work as an editing form with an existing album.

def edit
  @form = AlbumForm.new(Album.find(1))
end

Reform will read property values from the model in setup. In our example, the AlbumForm will call album.title to populate the title field.

Rendering Forms

Your @form is now ready to be rendered, either do it yourself or use something like Rails' #form_for, simple_form or formtastic.

= form_for @form do |f|
  = f.input :title

Nested forms and collections can be easily rendered with fields_for, etc. Note that you no longer pass the model to the form builder, but the Reform instance.

Optionally, you might want to use the #prepopulate! method to pre-populate fields and prepare the form for rendering.

Validation

After form submission, you need to validate the input.

class SongsController
  def create
    @form = SongForm.new(Song.new)

    #=> params: {song: {title: "Rio", length: "366"}}

    if @form.validate(params[:song])

The #validate method first updates the values of the form - the underlying model is still treated as immutuable and remains unchanged. It then runs all validations you provided in the form.

It's the only entry point for updating the form. This is per design, as separating writing and validation doesn't make sense for a form.

This allows rendering the form after validate with the data that has been submitted. However, don't get confused, the model's values are still the old, original values and are only changed after a #save or #sync operation.

Syncing Back

After validation, you have two choices: either call #save and let Reform sort out the rest. Or call #sync, which will write all the properties back to the model. In a nested form, this works recursively, of course.

It's then up to you what to do with the updated models - they're still unsaved.

Saving Forms

The easiest way to save the data is to call #save on the form.

if @form.validate(params[:song])
  @form.save  #=> populates album with incoming data
              #   by calling @form.album.title=.
else
  # handle validation errors.
end

This will sync the data to the model and then call album.save.

Sometimes, you need to do saving manually.

Default values

Reform allows default values to be provided for properties.

class AlbumForm < Reform::Form
  property :price_in_cents, default: 9_95
end

Saving Forms Manually

Calling #save with a block will provide a nested hash of the form's properties and values. This does not call #save on the models and allows you to implement the saving yourself.

The block parameter is a nested hash of the form input.

  @form.save do |hash|
    hash      #=> {title: "Greatest Hits"}
    Album.create(hash)
  end

You can always access the form's model. This is helpful when you were using populators to set up objects when validating.

  @form.save do |hash|
    album = @form.model

    album.update_attributes(hash[:album])
  end

Nesting

Reform provides support for nested objects. Let's say the Album model keeps some associations.

class Album < ActiveRecord::Base
  has_one  :artist
  has_many :songs
end

The implementation details do not really matter here, as long as your album exposes readers and writes like Album#artist and Album#songs, this allows you to define nested forms.

class AlbumForm < Reform::Form
  property :title
  validates :title, presence: true

  property :artist do
    property :full_name
    validates :full_name, presence: true
  end

  collection :songs do
    property :name
  end
end

You can also reuse an existing form from elsewhere using :form.

property :artist, form: ArtistForm

Nested Setup

Reform will wrap defined nested objects in their own forms. This happens automatically when instantiating the form.

album.songs #=> [<Song name:"Run To The Hills">]

form = AlbumForm.new(album)
form.songs[0] #=> <SongForm model: <Song name:"Run To The Hills">>
form.songs[0].name #=> "Run To The Hills"

Nested Rendering

When rendering a nested form you can use the form's readers to access the nested forms.

= text_field :title,         @form.title
= text_field "artist[name]", @form.artist.name

Or use something like #fields_for in a Rails environment.

= form_for @form do |f|
  = f.text_field :title

  = f.fields_for :artist do |a|
    = a.text_field :name

Nested Processing

validate will assign values to the nested forms. sync and save work analogue to the non-nested form, just in a recursive way.

The block form of #save would give you the following data.

@form.save do |nested|
  nested #=> {title:  "Greatest Hits",
         #    artist: {name: "Duran Duran"},
         #    songs: [{title: "Hungry Like The Wolf"},
         #            {title: "Last Chance On The Stairways"}]
         #   }
  end

The manual saving with block is not encouraged. You should rather check the Disposable docs to find out how to implement your manual tweak with the official API.

Populating Forms

Very often, you need to give Reform some information how to create or find nested objects when validateing. This directive is called populator and documented here.

Installation

Add this line to your Gemfile:

gem "reform"

Reform works fine with Rails 3.1-5.0. However, inheritance of validations with ActiveModel::Validations is broken in Rails 3.2 and 4.0.

Since Reform 2.2, you have to add the reform-rails gem to your Gemfile to automatically load ActiveModel/Rails files.

gem "reform-rails"

Since Reform 2.0 you need to specify which validation backend you want to use (unless you're in a Rails environment where ActiveModel will be used).

To use ActiveModel (not recommended because very out-dated).

require "reform/form/active_model/validations"
Reform::Form.class_eval do
  include Reform::Form::ActiveModel::Validations
end

To use dry-validation (recommended).

require "reform/form/dry"
Reform::Form.class_eval do
  feature Reform::Form::Dry
end

Put this in an initializer or on top of your script.

Compositions

Reform allows to map multiple models to one form. The complete documentation is here, however, this is how it works.

class AlbumForm < Reform::Form
  include Composition

  property :id,    on: :album
  property :title, on: :album
  property :songs, on: :cd
  property :cd_id, on: :cd, from: :id
end

When initializing a composition, you have to pass a hash that contains the composees.

AlbumForm.new(album: album, cd: CD.find(1))

More

Reform comes many more optional features, like hash fields, coercion, virtual fields, and so on. Check the full documentation here.

Reform is part of the Trailblazer project. Please buy my book to support the development and learn everything about Reform - there's two chapters dedicated to Reform!

Security And Strong_parameters

By explicitly defining the form layout using ::property there is no more need for protecting from unwanted input. strong_parameter or attr_accessible become obsolete. Reform will simply ignore undefined incoming parameters.

This is not Reform 1.x!

Temporary note: This is the README and API for Reform 2. On the public API, only a few tiny things have changed. Here are the Reform 1.2 docs.

Anyway, please upgrade and report problems and do not simply assume that we will magically find out what needs to get fixed. When in trouble, join us on Gitter.

Full documentation for Reform is available online, or support us and grab the Trailblazer book. There is an Upgrading Guide to help you migrate through versions.

Attributions!!!

Great thanks to Blake Education for giving us the freedom and time to develop this project in 2013 while working on their project.


Author: trailblazer
Source code: https://github.com/trailblazer/reform
License:  MIT license

#ruby  #ruby-on-rails

A Collection Of Swift Tips & Tricks That I've Shared on Twitter

⚠️ This list is no longer being updated. For my latest Swift tips, checkout the "Tips" section on Swift by Sundell.

Swift tips & tricks ⚡️

One of the things I really love about Swift is how I keep finding interesting ways to use it in various situations, and when I do - I usually share them on Twitter. Here's a collection of all the tips & tricks that I've shared so far. Each entry has a link to the original tweet, if you want to respond with some feedback or question, which is always super welcome! 🚀

Also make sure to check out all of my other Swift content:

#102 Making async tests faster and more stable

🚀 Here are some quick tips to make async tests faster & more stable:

  • 😴 Avoid sleep() - use expectations instead
  • ⏱ Use generous timeouts to avoid flakiness on CI
  • 🧐 Put all assertions at the end of each test, not inside closures
// BEFORE:

class MentionDetectorTests: XCTestCase {
    func testDetectingMention() {
        let detector = MentionDetector()
        let string = "This test was written by @johnsundell."

        detector.detectMentions(in: string) { mentions in
            XCTAssertEqual(mentions, ["johnsundell"])
        }
        
        sleep(2)
    }
}

// AFTER:

class MentionDetectorTests: XCTestCase {
    func testDetectingMention() {
        let detector = MentionDetector()
        let string = "This test was written by @johnsundell."

        var mentions: [String]?
        let expectation = self.expectation(description: #function)

        detector.detectMentions(in: string) {
            mentions = $0
            expectation.fulfill()
        }

        waitForExpectations(timeout: 10)
        XCTAssertEqual(mentions, ["johnsundell"])
    }
}

For more on async testing, check out "Unit testing asynchronous Swift code".

#101 Adding support for Apple Pencil double-taps

✍️ Adding support for the new Apple Pencil double-tap feature is super easy! All you have to do is to create a UIPencilInteraction, add it to a view, and implement one delegate method. Hopefully all pencil-compatible apps will soon adopt this.

let interaction = UIPencilInteraction()
interaction.delegate = self
view.addInteraction(interaction)

extension ViewController: UIPencilInteractionDelegate {
    func pencilInteractionDidTap(_ interaction: UIPencilInteraction) {
        // Handle pencil double-tap
    }
}

For more on using this and other iPad Pro features, check out "Building iPad Pro features in Swift".

#100 Combining values with functions

😎 Here's a cool function that combines a value with a function to return a closure that captures that value, so that it can be called without any arguments. Super useful when working with closure-based APIs and we want to use some of our properties without having to capture self.

func combine<A, B>(_ value: A, with closure: @escaping (A) -> B) -> () -> B {
    return { closure(value) }
}

// BEFORE:

class ProductViewController: UIViewController {
    override func viewDidLoad() {
        super.viewDidLoad()

        buyButton.handler = { [weak self] in
            guard let self = self else {
                return
            }
            
            self.productManager.startCheckout(for: self.product)
        }
    }
}

// AFTER:

class ProductViewController: UIViewController {
    override func viewDidLoad() {
        super.viewDidLoad()

        buyButton.handler = combine(product, with: productManager.startCheckout)
    }
}

#99 Dependency injection using functions

💉 When I'm only using a single function from a dependency, I love to inject that function as a closure, instead of having to create a protocol and inject the whole object. Makes dependency injection & testing super simple.

final class ArticleLoader {
    typealias Networking = (Endpoint) -> Future<Data>
    
    private let networking: Networking
    
    init(networking: @escaping Networking = URLSession.shared.load) {
        self.networking = networking
    }
    
    func loadLatest() -> Future<[Article]> {
        return networking(.latestArticles).decode()
    }
}

For more on this technique, check out "Simple Swift dependency injection with functions".

#98 Using a custom exception handler

💥 It's cool that you can easily assign a closure as a custom NSException handler. This is super useful when building things in Playgrounds - since you can't use breakpoints - so instead of just signal SIGABRT, you'll get the full exception description if something goes wrong.

NSSetUncaughtExceptionHandler { exception in
    print(exception)
}

#97 Using type aliases to give semantic meaning to primitives

❤️ I love that in Swift, we can use the type system to make our code so much more self-documenting - one way of doing so is to use type aliases to give the primitive types that we use a more semantic meaning.

extension List.Item {
    // Using type aliases, we can give semantic meaning to the
    // primitive types that we use, without having to introduce
    // wrapper types.
    typealias Index = Int
}

extension List {
    enum Mutation {
        // Our enum cases now become a lot more self-documenting,
        // without having to add additional parameter labels to
        // explain them.
        case add(Item, Item.Index)
        case update(Item, Item.Index)
        case remove(Item.Index)
    }
}

For more on self-documenting code, check out "Writing self-documenting Swift code".

#96 Specializing protocols using constraints

🤯 A little late night prototyping session reveals that protocol constraints can not only be applied to extensions - they can also be added to protocol definitions!

This is awesome, since it lets us easily define specialized protocols based on more generic ones.

protocol Component {
    associatedtype Container
    func add(to container: Container)
}

// Protocols that inherit from other protocols can include
// constraints to further specialize them.
protocol ViewComponent: Component where Container == UIView {
    associatedtype View: UIView
    var view: View { get }
}

extension ViewComponent {
    func add(to container: UIView) {
        container.addSubview(view)
    }
}

For more on specializing protocols, check out "Specializing protocols in Swift".

#95 Unwrapping an optional or throwing an error

📦 Here's a super handy extension on Swift's Optional type, which gives us a really nice API for easily unwrapping an optional, or throwing an error in case the value turned out to be nil:

extension Optional {
    func orThrow(_ errorExpression: @autoclosure () -> Error) throws -> Wrapped {
        switch self {
        case .some(let value):
            return value
        case .none:
            throw errorExpression()
        }
    }
}

let file = try loadFile(at: path).orThrow(MissingFileError())

For more ways that optionals can be extended, check out "Extending optionals in Swift".

#94 Testing code that uses static APIs

👩‍🔬 Testing code that uses static APIs can be really tricky, but there's a way that it can often be done - using Swift's first class function capabilities!

Instead of accessing that static API directly, we can inject the function we want to use, which enables us to mock it!

// BEFORE

class FriendsLoader {
    func loadFriends(then handler: @escaping (Result<[Friend]>) -> Void) {
        Networking.loadData(from: .friends) { result in
            ...
        }
    }
}

// AFTER

class FriendsLoader {
    typealias Handler<T> = (Result<T>) -> Void
    typealias DataLoadingFunction = (Endpoint, @escaping Handler<Data>) -> Void

    func loadFriends(using dataLoading: DataLoadingFunction = Networking.loadData,
                     then handler: @escaping Handler<[Friend]>) {
        dataLoading(.friends) { result in
            ...
        }
    }
}

// MOCKING IN TESTS

let dataLoading: FriendsLoader.DataLoadingFunction = { _, handler in
    handler(.success(mockData))
}

friendsLoader.loadFriends(using: dataLoading) { result in
    ...
}

#93 Matching multiple enum cases with associated values

🐾 Swift's pattern matching capabilities are so powerful! Two enum cases with associated values can even be matched and handled by the same switch case - which is super useful when handling state changes with similar data.

enum DownloadState {
    case inProgress(progress: Double)
    case paused(progress: Double)
    case cancelled
    case finished(Data)
}

func downloadStateDidChange(to state: DownloadState) {
    switch state {
    case .inProgress(let progress), .paused(let progress):
        updateProgressView(with: progress)
    case .cancelled:
        showCancelledMessage()
    case .finished(let data):
        process(data)
    }
}

#92 Multiline string literals

🅰 One really nice benefit of Swift multiline string literals - even for single lines of text - is that they don't require quotes to be escaped. Perfect when working with things like HTML, or creating a custom description for an object.

let html = highlighter.highlight("Array<String>")

XCTAssertEqual(html, """
<span class="type">Array</span>&lt;<span class="type">String</span>&gt;
""")

#91 Reducing sequences

💎 While it's very common in functional programming, the reduce function might be a bit of a hidden gem in Swift. It provides a super useful way to transform a sequence into a single value.

extension Sequence where Element: Equatable {
    func numberOfOccurrences(of target: Element) -> Int {
        return reduce(0) { result, element in
            guard element == target else {
                return result
            }

            return result + 1
        }
    }
}

You can read more about transforming collections in "Transforming collections in Swift".

#90 Avoiding manual Codable implementations

📦 When I use Codable in Swift, I want to avoid manual implementations as much as possible, even when there's a mismatch between my code structure and the JSON I'm decoding.

One way that can often be achieved is to use private data containers combined with computed properties.

struct User: Codable {
    let name: String
    let age: Int

    var homeTown: String { return originPlace.name }

    private let originPlace: Place
}

private extension User {
    struct Place: Codable {
        let name: String
    }
}

extension User {
    struct Container: Codable {
        let user: User
    }
}

#89 Using feature flags instead of feature branches

🚢 Instead of using feature branches, I merge almost all of my code directly into master - and then I use feature flags to conditionally enable features when they're ready. That way I can avoid merge conflicts and keep shipping!

extension ListViewController {
    func addSearchIfNeeded() {
        // Rather than having to keep maintaining a separate
        // feature branch for a new feature, we can use a flag
        // to conditionally turn it on.
        guard FeatureFlags.searchEnabled else {
            return
        }

        let resultsVC = SearchResultsViewController()
        let searchVC = UISearchController(
            searchResultsController: resultsVC
        )

        searchVC.searchResultsUpdater = resultsVC
        navigationItem.searchController = searchVC
    }
}

You can read more about feature flags in "Feature flags in Swift".

#88 Lightweight data hierarchies using tuples

💾 Here I'm using tuples to create a lightweight hierarchy for my data, giving me a nice structure without having to introduce any additional types.

struct CodeSegment {
    var tokens: (
        previous: String?,
        current: String
    )

    var delimiters: (
        previous: Character?
        next: Character?
    )
}

handle(segment.tokens.current)

You can read more about tuples in "Using tuples as lightweight types in Swift"

#87 The rule of threes

3️⃣ Whenever I have 3 properties or local variables that share the same prefix, I usually try to extract them into their own method or type. That way I can avoid massive types & methods, and also increase readability, without falling into a "premature optimization" trap.

Before

public func generate() throws {
    let contentFolder = try folder.subfolder(named: "content")

    let articleFolder = try contentFolder.subfolder(named: "posts")
    let articleProcessor = ContentProcessor(folder: articleFolder)
    let articles = try articleProcessor.process()

    ...
}

After

public func generate() throws {
    let contentFolder = try folder.subfolder(named: "content")
    let articles = try processArticles(in: contentFolder)
    ...
}

private func processArticles(in folder: Folder) throws -> [ContentItem] {
    let folder = try folder.subfolder(named: "posts")
    let processor = ContentProcessor(folder: folder)
    return try processor.process()
}

#86 Useful Codable extensions

👨‍🔧 Here's two extensions that I always add to the Encodable & Decodable protocols, which for me really make the Codable API nicer to use. By using type inference for decoding, a lot of boilerplate can be removed when the compiler is already able to infer the resulting type.

extension Encodable {
    func encoded() throws -> Data {
        return try JSONEncoder().encode(self)
    }
}

extension Data {
    func decoded<T: Decodable>() throws -> T {
        return try JSONDecoder().decode(T.self, from: self)
    }
}

let data = try user.encoded()

// By using a generic type in the decoded() method, the
// compiler can often infer the type we want to decode
// from the current context.
try userDidLogin(data.decoded())

// And if not, we can always supply the type, still making
// the call site read very nicely.
let otherUser = try data.decoded() as User

#85 Using shared UserDefaults suites

📦 UserDefaults is a lot more powerful than what it first might seem like. Not only can it store more complex values (like dates & dictionaries) and parse command line arguments - it also enables easy sharing of settings & lightweight data between apps in the same App Group.

let sharedDefaults = UserDefaults(suiteName: "my-app-group")!
let useDarkMode = sharedDefaults.bool(forKey: "dark-mode")

// This value is put into the shared suite.
sharedDefaults.set(true, forKey: "dark-mode")

// If you want to treat the shared settings as read-only (and add
// local overrides on top of them), you can simply add the shared
// suite to the standard UserDefaults.
let combinedDefaults = UserDefaults.standard
combinedDefaults.addSuite(named: "my-app-group")

// This value is a local override, not added to the shared suite.
combinedDefaults.set(true, forKey: "app-specific-override")

#84 Custom UIView backing layers

🎨 By overriding layerClass you can tell UIKit what CALayer class to use for a UIView's backing layer. That way you can reduce the amount of layers, and don't have to do any manual layout.

final class GradientView: UIView {
    override class var layerClass: AnyClass { return CAGradientLayer.self }

    var colors: (start: UIColor, end: UIColor)? {
        didSet { updateLayer() }
    }

    private func updateLayer() {
        let layer = self.layer as! CAGradientLayer
        layer.colors = colors.map { [$0.start.cgColor, $0.end.cgColor] }
    }
}

#83 Auto-Equatable enums with associated values

✅ That the compiler now automatically synthesizes Equatable conformances is such a huge upgrade for Swift! And the cool thing is that it works for all kinds of types - even for enums with associated values! Especially useful when using enums for verification in unit tests.

struct Article: Equatable {
    let title: String
    let text: String
}

struct User: Equatable {
    let name: String
    let age: Int
}

extension Navigator {
    enum Destination: Equatable {
        case profile(User)
        case article(Article)
    }
}

func testNavigatingToArticle() {
    let article = Article(title: "Title", text: "Text")
    controller.select(article)
    XCTAssertEqual(navigator.destinations, [.article(article)])
}

#82 Defaults for associated types

🤝 Associated types can have defaults in Swift - which is super useful for types that are not easily inferred (for example when they're not used for a specific instance method or property).

protocol Identifiable {
    associatedtype RawIdentifier: Codable = String

    var id: Identifier<Self> { get }
}

struct User: Identifiable {
    let id: Identifier<User>
    let name: String
}

struct Group: Identifiable {
    typealias RawIdentifier = Int

    let id: Identifier<Group>
    let name: String
}

#81 Creating a dedicated identifier type

🆔 If you want to avoid using plain strings as identifiers (which can increase both type safety & readability), it's really easy to create a custom Identifier type that feels just like a native Swift type, thanks to protocols!

More on this topic in "Type-safe identifiers in Swift".

struct Identifier: Hashable {
    let string: String
}

extension Identifier: ExpressibleByStringLiteral {
    init(stringLiteral value: String) {
        string = value
    }
}

extension Identifier: CustomStringConvertible {
    var description: String {
        return string
    }
}

extension Identifier: Codable {
    init(from decoder: Decoder) throws {
        let container = try decoder.singleValueContainer()
        string = try container.decode(String.self)
    }

    func encode(to encoder: Encoder) throws {
        var container = encoder.singleValueContainer()
        try container.encode(string)
    }
}

struct Article: Codable {
    let id: Identifier
    let title: String
}

let article = Article(id: "my-article", title: "Hello world!")

#80 Assigning optional tuple members to variables

🙌 A really cool thing about using tuples to model the internal state of a Swift type, is that you can unwrap an optional tuple's members directly into local variables.

Very useful in order to group multiple optional values together for easy unwrapping & handling.

class ImageTransformer {
    private var queue = [(image: UIImage, transform: Transform)]()

    private func processNext() {
        // When unwrapping an optional tuple, you can assign the members
        // directly to local variables.
        guard let (image, transform) = queue.first else {
            return
        }

        let context = Context()
        context.draw(image)
        context.apply(transform)
        ...
    }
}

#79 Struct convenience initializers

❤️ I love to structure my code using extensions in Swift. One big benefit of doing so when it comes to struct initializers, is that defining a convenience initializer doesn't remove the default one the compiler generates - best of both worlds!

struct Article {
    let date: Date
    var title: String
    var text: String
    var comments: [Comment]
}

extension Article {
    init(title: String, text: String) {
        self.init(date: Date(), title: title, text: text, comments: [])
    }
}

let articleA = Article(title: "Best Cupcake Recipe", text: "...")

let articleB = Article(
    date: Date(),
    title: "Best Cupcake Recipe",
    text: "...",
    comments: [
        Comment(user: currentUser, text: "Yep, can confirm!")
    ]
)

#78 Usages of throwing functions

🏈 A big benefit of using throwing functions for synchronous Swift APIs is that the caller can decide whether they want to treat the return value as optional (try?) or required (try).

func loadFile(named name: String) throws -> File {
    guard let url = urlForFile(named: name) else {
        throw File.Error.missing
    }

    do {
        let data = try Data(contentsOf: url)
        return File(url: url, data: data)
    } catch {
        throw File.Error.invalidData(error)
    }
}

let requiredFile = try loadFile(named: "AppConfig.json")

let optionalFile = try? loadFile(named: "UserSettings.json")

#77 Nested generic types

🐝 Types that are nested in generics automatically inherit their parent's generic types - which is super useful when defining accessory types (for things like states or outcomes).

struct Task<Input, Output> {
    typealias Closure = (Input) throws -> Output

    let closure: Closure
}

extension Task {
    enum Result {
        case success(Output)
        case failure(Error)
    }
}

#76 Equatable & Hashable structures

🤖 Now that the Swift compiler automatically synthesizes Equatable & Hashable conformances for value types, it's easier than ever to setup model structures with nested types that are all Equatable/Hashable!

typealias Value = Hashable & Codable

struct User: Value {
    var name: String
    var age: Int
    var lastLoginDate: Date?
    var settings: Settings
}

extension User {
    struct Settings: Value {
        var itemsPerPage: Int
        var theme: Theme
    }
}

extension User.Settings {
    enum Theme: String, Value {
        case light
        case dark
    }
}

You can read more about using nested types in Swift here.

#75 Conditional conformances

🎉 Swift 4.1 is here! One of the key features it brings is conditional conformances, which lets you have a type only conform to a protocol under certain constraints.

protocol UnboxTransformable {
    associatedtype RawValue

    static func transform(_ value: RawValue) throws -> Self?
}

extension Array: UnboxTransformable where Element: UnboxTransformable {
    typealias RawValue = [Element.RawValue]

    static func transform(_ value: RawValue) throws -> [Element]? {
        return try value.compactMap(Element.transform)
    }
}

I also have an article with lots of more info on conditional conformances here. Paul Hudson also has a great overview of all Swift 4.1 features here.

#74 Generic type aliases

🕵️‍♀️ A cool thing about Swift type aliases is that they can be generic! Combine that with tuples and you can easily define simple generic types.

typealias Pair<T> = (T, T)

extension Game {
    func calculateScore(for players: Pair<Player>) -> Int {
        ...
    }
}

You can read more about using tuples as lightweight types here.

#73 Parsing command line arguments using UserDefaults

☑️ A really cool "hidden" feature of UserDefaults is that it contains any arguments that were passed to the app at launch!

Super useful both in Swift command line tools & scripts, but also to temporarily override a value when debugging iOS apps.

let defaults = UserDefaults.standard
let query = defaults.string(forKey: "query")
let resultCount = defaults.integer(forKey: "results")

#72 Using the & operator

👏 Swift's & operator is awesome! Not only can you use it to compose protocols, you can compose other types too! Very useful if you want to hide concrete types & implementation details.

protocol LoadableFromURL {
    func load(from url: URL)
}

class ContentViewController: UIViewController, LoadableFromURL {
    func load(from url: URL) {
        ...
    }
}

class ViewControllerFactory {
    func makeContentViewController() -> UIViewController & LoadableFromURL {
        return ContentViewController()
    }
}

#71 Capturing multiple values in mocks

🤗 When capturing values in mocks, using an array (instead of just a single value) makes it easy to verify that only a certain number of values were passed.

Perfect for protecting against "over-calling" something.

class UserManagerTests: XCTestCase {
    func testObserversCalledWhenUserFirstLogsIn() {
        let manager = UserManager()

        let observer = ObserverMock()
        manager.addObserver(observer)

        // First login, observers should be notified
        let user = User(id: 123, name: "John")
        manager.userDidLogin(user)
        XCTAssertEqual(observer.users, [user])

        // If the same user logs in again, observers shouldn't be notified
        manager.userDidLogin(user)
        XCTAssertEqual(observer.users, [user])
    }
}

private extension UserManagerTests {
    class ObserverMock: UserManagerObserver {
        private(set) var users = [User]()

        func userDidChange(to user: User) {
            users.append(user)
        }
    }
}

#70 Reducing the need for mocks

👋 When writing tests, you don't always need to create mocks - you can create stubs using real instances of things like errors, URLs & UserDefaults.

Here's how to do that for some common tasks/object types in Swift:

// Create errors using NSError (#function can be used to reference the name of the test)
let error = NSError(domain: #function, code: 1, userInfo: nil)

// Create non-optional URLs using file paths
let url = URL(fileURLWithPath: "Some/URL")

// Reference the test bundle using Bundle(for:)
let bundle = Bundle(for: type(of: self))

// Create an explicit UserDefaults object (instead of having to use a mock)
let userDefaults = UserDefaults(suiteName: #function)

// Create queues to control/await concurrent operations
let queue = DispatchQueue(label: #function)

For when you actually do need mocking, check out "Mocking in Swift".

#69 Using "then" as an external parameter label for closures

⏱ I've started using "then" as an external parameter label for completion handlers. Makes the call site read really nicely (Because I do ❤️ conversational API design) regardless of whether trailing closure syntax is used or not.

protocol DataLoader {
    // Adding type aliases to protocols can be a great way to
    // reduce verbosity for parameter types.
    typealias Handler = (Result<Data>) -> Void
    associatedtype Endpoint

    func loadData(from endpoint: Endpoint, then handler: @escaping Handler)
}

loader.loadData(from: .messages) { result in
    ...
}

loader.loadData(from: .messages, then: { result in
    ...
})

#68 Combining lazily evaluated sequences with the builder pattern

😴 Combining lazily evaluated sequences with builder pattern-like properties can lead to some pretty sweet APIs for configurable sequences in Swift.

Also useful for queries & other things you "build up" and then execute.

// Extension adding builder pattern-like properties that return
// a new sequence value with the given configuration applied
extension FileSequence {
    var recursive: FileSequence {
        var sequence = self
        sequence.isRecursive = true
        return sequence
    }

    var includingHidden: FileSequence {
        var sequence = self
        sequence.includeHidden = true
        return sequence
    }
}

// BEFORE

let files = folder.makeFileSequence(recursive: true, includeHidden: true)

// AFTER

let files = folder.files.recursive.includingHidden

Want an intro to lazy sequences? Check out "Swift sequences: The art of being lazy".

#67 Faster & more stable UI tests

My top 3 tips for faster & more stable UI tests:

📱 Reset the app's state at the beginning of every test.

🆔 Use accessibility identifiers instead of UI strings.

⏱ Use expectations instead of waiting time.

func testOpeningArticle() {
    // Launch the app with an argument that tells it to reset its state
    let app = XCUIApplication()
    app.launchArguments.append("--uitesting")
    app.launch()
    
    // Check that the app is displaying an activity indicator
    let activityIndicator = app.activityIndicator.element
    XCTAssertTrue(activityIndicator.exists)
    
    // Wait for the loading indicator to disappear = content is ready
    expectation(for: NSPredicate(format: "exists == 0"),
                evaluatedWith: activityIndicator)
                
    // Use a generous timeout in case the network is slow
    waitForExpectations(timeout: 10)
    
    // Tap the cell for the first article
    app.tables.cells["Article.0"].tap()
    
    // Assert that a label with the accessibility identifier "Article.Title" exists
    let label = app.staticTexts["Article.Title"]
    XCTAssertTrue(label.exists)
}

#66 Accessing the clipboard from a Swift script

📋 It's super easy to access the contents of the clipboard from a Swift script. A big benefit of Swift scripting is being able to use Cocoa's powerful APIs for Mac apps.

import Cocoa

let clipboard = NSPasteboard.general.string(forType: .string)

#65 Using tuples for view state

🎯 Using Swift tuples for view state can be a super nice way to group multiple properties together and render them reactively using the layout system.

By using a tuple we don't have to either introduce a new type or make our view model-aware.

class TextView: UIView {
    var state: (title: String?, text: String?) {
        // By telling UIKit that our view needs layout and binding our
        // state in layoutSubviews, we can react to state changes without
        // doing unnecessary layout work.
        didSet { setNeedsLayout() }
    }

    private let titleLabel = UILabel()
    private let textLabel = UILabel()

    override func layoutSubviews() {
        super.layoutSubviews()

        titleLabel.text = state.title
        textLabel.text = state.text

        ...
    }
}

#64 Throwing tests and LocalizedError

⚾️ Swift tests can throw, which is super useful in order to avoid complicated logic or force unwrapping. By making errors conform to LocalizedError, you can also get a nice error message in Xcode if there's a failure.

class ImageCacheTests: XCTestCase {
    func testCachingAndLoadingImage() throws {
        let bundle = Bundle(for: type(of: self))
        let cache = ImageCache(bundle: bundle)
        
        // Bonus tip: You can easily load images from your test
        // bundle using this UIImage initializer
        let image = try require(UIImage(named: "sample", in: bundle, compatibleWith: nil))
        try cache.cache(image, forKey: "key")
        
        let cachedImage = try cache.image(forKey: "key")
        XCTAssertEqual(image, cachedImage)
    }
}

enum ImageCacheError {
    case emptyKey
    case dataConversionFailed
}

// When using throwing tests, making your errors conform to
// LocalizedError will render a much nicer error message in
// Xcode (per default only the error code is shown).
extension ImageCacheError: LocalizedError {
    var errorDescription: String? {
        switch self {
        case .emptyKey:
            return "An empty key was given"
        case .dataConversionFailed:
            return "Failed to convert the given image to Data"
        }
    }
}

For more information, and the implementation of the require method used above, check out "Avoiding force unwrapping in Swift unit tests".

#63 The difference between static and class properties

✍️ Unlike static properties, class properties can be overridden by subclasses (however, they can't be stored, only computed).

class TableViewCell: UITableViewCell {
    class var preferredHeight: CGFloat { return 60 }
}

class TallTableViewCell: TableViewCell {
    override class var preferredHeight: CGFloat { return 100 }
}

#62 Creating extensions with static factory methods

👨‍🎨 Creating extensions with static factory methods can be a great alternative to subclassing in Swift, especially for things like setting up UIViews, CALayers or other kinds of styling.

It also lets you remove a lot of styling & setup from your view controllers.

extension UILabel {
    static func makeForTitle() -> UILabel {
        let label = UILabel()
        label.font = .boldSystemFont(ofSize: 24)
        label.textColor = .darkGray
        label.adjustsFontSizeToFitWidth = true
        label.minimumScaleFactor = 0.75
        return label
    }

    static func makeForText() -> UILabel {
        let label = UILabel()
        label.font = .systemFont(ofSize: 16)
        label.textColor = .black
        label.numberOfLines = 0
        return label
    }
}

class ArticleViewController: UIViewController {
    lazy var titleLabel = UILabel.makeForTitle()
    lazy var textLabel = UILabel.makeForText()
}

#61 Child view controller auto-resizing

🧒 An awesome thing about child view controllers is that they're automatically resized to match their parent, making them a super nice solution for things like loading & error views.

class ListViewController: UIViewController {
    func loadItems() {
        let loadingViewController = LoadingViewController()
        add(loadingViewController)

        dataLoader.loadItems { [weak self] result in
            loadingViewController.remove()
            self?.handle(result)
        }
    }
}

For more about child view controller (including the add and remove methods used above), check out "Using child view controllers as plugins in Swift".

#60 Using zip

🤐 Using the zip function in Swift you can easily combine two sequences. Super useful when using two sequences to do some work, since zip takes care of all the bounds-checking.

func render(titles: [String]) {
    for (label, text) in zip(titleLabels, titles) {
        print(text)
        label.text = text
    }
}

#59 Defining custom option sets

🎛 The awesome thing about option sets in Swift is that they can automatically either be passed as a single member or as a set. Even cooler is that you can easily define your own option sets as well, perfect for options and other non-exclusive values.

// Option sets are awesome, because you can easily pass them
// both using dot syntax and array literal syntax, like when
// using the UIView animation API:
UIView.animate(withDuration: 0.3,
               delay: 0,
               options: .allowUserInteraction,
               animations: animations)

UIView.animate(withDuration: 0.3,
               delay: 0,
               options: [.allowUserInteraction, .layoutSubviews],
               animations: animations)

// The cool thing is that you can easily define your own option
// sets as well, by defining a struct that has an Int rawValue,
// that will be used as a bit mask.
extension Cache {
    struct Options: OptionSet {
        static let saveToDisk = Options(rawValue: 1)
        static let clearOnMemoryWarning = Options(rawValue: 1 << 1)
        static let clearDaily = Options(rawValue: 1 << 2)

        let rawValue: Int
    }
}

// We can now use Cache.Options just like UIViewAnimationOptions:
Cache(options: .saveToDisk)
Cache(options: [.saveToDisk, .clearDaily])

#58 Using the where clause with associated types

🙌 Using the where clause when designing protocol-oriented APIs in Swift can let your implementations (or others' if it's open source) have a lot more freedom, especially when it comes to collections.

See "Using generic type constraints in Swift 4" for more info.

public protocol PathFinderMap {
    associatedtype Node
    // Using the 'where' clause for associated types, we can
    // ensure that a type meets certain requirements (in this
    // case that it's a sequence with Node elements).
    associatedtype NodeSequence: Sequence where NodeSequence.Element == Node

    // Instead of using a concrete type (like [Node]) here, we
    // give implementors of this protocol more freedom while
    // still meeting our requirements. For example, one
    // implementation might use Set<Node>.
    func neighbors(of node: Node) -> NodeSequence
}

#57 Using first class functions when iterating over a dictionary

👨‍🍳 Combine first class functions in Swift with the fact that Dictionary elements are (Key, Value) tuples and you can build yourself some pretty awesome functional chains when iterating over a Dictionary.

func makeActor(at coordinate: Coordinate, for building: Building) -> Actor {
    let actor = Actor()
    actor.position = coordinate.point
    actor.animation = building.animation
    return actor
}

func render(_ buildings: [Coordinate : Building]) {
    buildings.map(makeActor).forEach(add)
}

#56 Calling instance methods as static functions

😎 In Swift, you can call any instance method as a static function and it will return a closure representing that method. This is how running tests using SPM on Linux works.

More about this topic in my blog post "First class functions in Swift".

// This produces a '() -> Void' closure which is a reference to the
// given view's 'removeFromSuperview' method.
let closure = UIView.removeFromSuperview(view)

// We can now call it just like we would any other closure, and it
// will run 'view.removeFromSuperview()'
closure()

// This is how running tests using the Swift Package Manager on Linux
// works, you return your test functions as closures:
extension UserManagerTests {
    static var allTests = [
        ("testLoggingIn", testLoggingIn),
        ("testLoggingOut", testLoggingOut),
        ("testUserPermissions", testUserPermissions)
    ]
}

#55 Dropping suffixes from method names to support multiple arguments

👏 One really nice benefit of dropping suffixes from method names (and just using verbs, when possible) is that it becomes super easy to support both single and multiple arguments, and it works really well semantically.

extension UIView {
    func add(_ subviews: UIView...) {
        subviews.forEach(addSubview)
    }
}

view.add(button)
view.add(label)

// By dropping the "Subview" suffix from the method name, both
// single and multiple arguments work really well semantically.
view.add(button, label)

#54 Constraining protocols to classes to ensure mutability

👽 Using the AnyObject (or class) constraint on protocols is not only useful when defining delegates (or other weak references), but also when you always want instances to be mutable without copying.

// By constraining a protocol with 'AnyObject' it can only be adopted
// by classes, which means all instances will always be mutable, and
// that it's the original instance (not a copy) that will be mutated.
protocol DataContainer: AnyObject {
    var data: Data? { get set }
}

class UserSettingsManager {
    private var settings: Settings
    private let dataContainer: DataContainer

    // Since DataContainer is a protocol, we an easily mock it in
    // tests if we use dependency injection
    init(settings: Settings, dataContainer: DataContainer) {
        self.settings = settings
        self.dataContainer = dataContainer
    }

    func saveSettings() throws {
        let data = try settings.serialize()

        // We can now assign properties on an instance of our protocol
        // because the compiler knows it's always going to be a class
        dataContainer.data = data
    }
}

#53 String-based enums in string interpolation

🍣 Even if you define a custom raw value for a string-based enum in Swift, the full case name will be used in string interpolation.

Super useful when using separate raw values for JSON, while still wanting to use the full case name in other contexts.

extension Building {
    // This enum has custom raw values that are used when decoding
    // a value, for example from JSON.
    enum Kind: String {
        case castle = "C"
        case town = "T"
        case barracks = "B"
        case goldMine = "G"
        case camp = "CA"
        case blacksmith = "BL"
    }

    var animation: Animation {
        return Animation(
            // When used in string interpolation, the full case name is still used.
            // For 'castle' this will be 'buildings/castle'.
            name: "buildings/\(kind)",
            frameCount: frameCount,
            frameDuration: frameDuration
        )
    }
}

#52 Expressively comparing a value with a list of candidates

👨‍🔬 Continuing to experiment with expressive ways of comparing a value with a list of candidates in Swift. Adding an extension on Equatable is probably my favorite approach so far.

extension Equatable {
    func isAny(of candidates: Self...) -> Bool {
        return candidates.contains(self)
    }
}

let isHorizontal = direction.isAny(of: .left, .right)

See tip #35 for my previous experiment.

#51 UIView bounds and transforms

📐 A really interesting side-effect of a UIView's bounds being its rect within its own coordinate system is that transforms don't affect it at all. That's why it's usually a better fit than frame when doing layout calculations of subviews.

let view = UIView()
view.frame.size = CGSize(width: 100, height: 100)
view.transform = CGAffineTransform(scaleX: 2, y: 2)

print(view.frame) // (-50.0, -50.0, 200.0, 200.0)
print(view.bounds) // (0.0, 0.0, 100.0, 100.0)

#50 UIKit default arguments

👏 It's awesome that many UIKit APIs with completion handlers and other optional parameters import into Swift with default arguments (even though they are written in Objective-C). Getting rid of all those nil arguments is so nice!

// BEFORE: All parameters are specified, just like in Objective-C

viewController.present(modalViewController, animated: true, completion: nil)

modalViewController.dismiss(animated: true, completion: nil)

viewController.transition(from: loadingViewController,
                          to: contentViewController,
                          duration: 0.3,
                          options: [],
                          animations: animations,
                          completion: nil)

// AFTER: Since many UIKit APIs with completion handlers and other
// optional parameters import into Swift with default arguments,
// we can make our calls shorter

viewController.present(modalViewController, animated: true)

modalViewController.dismiss(animated: true)

viewController.transition(from: loadingViewController,
                          to: contentViewController,
                          duration: 0.3,
                          animations: animations)

#49 Avoiding Massive View Controllers

✂️ Avoiding Massive View Controllers is all about finding the right levels of abstraction and splitting things up.

My personal rule of thumb is that as soon as I have 3 methods or properties that have the same prefix, I break them out into their own type.

// BEFORE

class LoginViewController: UIViewController {
    private lazy var signUpLabel = UILabel()
    private lazy var signUpImageView = UIImageView()
    private lazy var signUpButton = UIButton()
}

// AFTER

class LoginViewController: UIViewController {
    private lazy var signUpView = SignUpView()
}

class SignUpView: UIView {
    private lazy var label = UILabel()
    private lazy var imageView = UIImageView()
    private lazy var button = UIButton()
}

#48 Extending optionals

❤️ I love the fact that optionals are enums in Swift - it makes it so easy to extend them with convenience APIs for certain types. Especially useful when doing things like data validation on optional values.

func validateTextFields() -> Bool {
    guard !usernameTextField.text.isNilOrEmpty else {
        return false
    }

    ...

    return true
}

// Since all optionals are actual enum values in Swift, we can easily
// extend them for certain types, to add our own convenience APIs

extension Optional where Wrapped == String {
    var isNilOrEmpty: Bool {
        switch self {
        case let string?:
            return string.isEmpty
        case nil:
            return true
        }
    }
}

// Since strings are now Collections in Swift 4, you can even
// add this property to all optional collections:

extension Optional where Wrapped: Collection {
    var isNilOrEmpty: Bool {
        switch self {
        case let collection?:
            return collection.isEmpty
        case nil:
            return true
        }
    }
}

#47 Using where with for-loops

🗺 Using the where keyword can be a super nice way to quickly apply a filter in a for-loop in Swift. You can of course use map, filter and forEach, or guard, but for simple loops I think this is very expressive and nice.

func archiveMarkedPosts() {
    for post in posts where post.isMarked {
        archive(post)
    }
}

func healAllies() {
    for player in players where player.isAllied(to: currentPlayer) {
        player.heal()
    }
}

#46 Variable shadowing

👻 Variable shadowing can be super useful in Swift, especially when you want to create a local copy of a parameter value in order to use it as state within a closure.

init(repeatMode: RepeatMode, closure: @escaping () -> UpdateOutcome) {
    // Shadow the argument with a local, mutable copy
    var repeatMode = repeatMode
    
    self.closure = {
        // With shadowing, there's no risk of accidentially
        // referring to the immutable version
        switch repeatMode {
        case .forever:
            break
        case .times(let count):
            guard count > 0 else {
                return .finished
            }
            
            // We can now capture the mutable version and use
            // it for state in a closure
            repeatMode = .times(count - 1)
        }
        
        return closure()
    }
}

#45 Using dot syntax for static properties and initializers

✒️ Dot syntax is one of my favorite features of Swift. What's really cool is that it's not only for enums, any static method or property can be used with dot syntax - even initializers! Perfect for convenience APIs and default parameters.

public enum RepeatMode {
    case times(Int)
    case forever
}

public extension RepeatMode {
    static var never: RepeatMode {
        return .times(0)
    }

    static var once: RepeatMode {
        return .times(1)
    }
}

view.perform(animation, repeated: .once)

// To make default parameters more compact, you can even use init with dot syntax

class ImageLoader {
    init(cache: Cache = .init(), decoder: ImageDecoder = .init()) {
        ...
    }
}

#44 Calling functions as closures with a tuple as parameters

🚀 One really cool aspect of Swift having first class functions is that you can pass any function (or even initializer) as a closure, and even call it with a tuple containing its parameters!

// This function lets us treat any "normal" function or method as
// a closure and run it with a tuple that contains its parameters
func call<Input, Output>(_ function: (Input) -> Output, with input: Input) -> Output {
    return function(input)
}

class ViewFactory {
    func makeHeaderView() -> HeaderView {
        // We can now pass an initializer as a closure, and a tuple
        // containing its parameters
        return call(HeaderView.init, with: loadTextStyles())
    }
    
    private func loadTextStyles() -> (font: UIFont, color: UIColor) {
        return (theme.font, theme.textColor)
    }
}

class HeaderView {
    init(font: UIFont, textColor: UIColor) {
        ...
    }
}

#43 Enabling static dependency injection

💉 If you've been struggling to test code that uses static APIs, here's a technique you can use to enable static dependency injection without having to modify any call sites:

// Before: Almost impossible to test due to the use of singletons

class Analytics {
    static func log(_ event: Event) {
        Database.shared.save(event)
        
        let dictionary = event.serialize()
        NetworkManager.shared.post(dictionary, to: eventURL)
    }
}

// After: Much easier to test, since we can inject mocks as arguments

class Analytics {
    static func log(_ event: Event,
                    database: Database = .shared,
                    networkManager: NetworkManager = .shared) {
        database.save(event)
        
        let dictionary = event.serialize()
        networkManager.post(dictionary, to: eventURL)
    }
}

#42 Type inference for lazy properties in Swift 4

🎉 In Swift 4, type inference works for lazy properties and you don't need to explicitly refer to self!

// Swift 3

class PurchaseView: UIView {
    private lazy var buyButton: UIButton = self.makeBuyButton()
    
    private func makeBuyButton() -> UIButton {
        let button = UIButton()
        button.setTitle("Buy", for: .normal)
        button.setTitleColor(.blue, for: .normal)
        return button
    }
}

// Swift 4

class PurchaseView: UIView {
    private lazy var buyButton = makeBuyButton()
    
    private func makeBuyButton() -> UIButton {
        let button = UIButton()
        button.setTitle("Buy", for: .normal)
        button.setTitleColor(.blue, for: .normal)
        return button
    }
}

#41 Converting Swift errors to NSError

😎 You can turn any Swift Error into an NSError, which is super useful when pattern matching with a code 👍. Also, switching on optionals is pretty cool!

let task = urlSession.dataTask(with: url) { data, _, error in
    switch error {
    case .some(let error as NSError) where error.code == NSURLErrorNotConnectedToInternet:
        presenter.showOfflineView()
    case .some(let error):
        presenter.showGenericErrorView()
    case .none:
        presenter.renderContent(from: data)
    }
}

task.resume()

Also make sure to check out Kostas Kremizas' tip about how you can pattern match directly against a member of URLError.

#40 Making UIImage macOS compatible

🖥 Here's an easy way to make iOS model code that uses UIImage macOS compatible - like me and Gui Rambo discussed on the Swift by Sundell Podcast.

// Either put this in a separate file that you only include in your macOS target or wrap the code in #if os(macOS) / #endif

import Cocoa

// Step 1: Typealias UIImage to NSImage
typealias UIImage = NSImage

// Step 2: You might want to add these APIs that UIImage has but NSImage doesn't.
extension NSImage {
    var cgImage: CGImage? {
        var proposedRect = CGRect(origin: .zero, size: size)

        return cgImage(forProposedRect: &proposedRect,
                       context: nil,
                       hints: nil)
    }

    convenience init?(named name: String) {
        self.init(named: Name(name))
    }
}

// Step 3: Profit - you can now make your model code that uses UIImage cross-platform!
struct User {
    let name: String
    let profileImage: UIImage
}

#39 Internally mutable protocol-oriented APIs

🤖 You can easily define a protocol-oriented API that can only be mutated internally, by using an internal protocol that extends a public one.

// Declare a public protocol that acts as your immutable API
public protocol ModelHolder {
    associatedtype Model
    var model: Model { get }
}

// Declare an extended, internal protocol that provides a mutable API
internal protocol MutableModelHolder: ModelHolder {
    var model: Model { get set }
}

// You can now implement the requirements using 'public internal(set)'
public class UserHolder: MutableModelHolder {
    public internal(set) var model: User

    internal init(model: User) {
        self.model = model
    }
}

#38 Switching on a set

🎛 You can switch on a set using array literals as cases in Swift! Can be really useful to avoid many if/else if statements.

class RoadTile: Tile {
    var connectedDirections = Set<Direction>()

    func render() {
        switch connectedDirections {
        case [.up, .down]:
            image = UIImage(named: "road-vertical")
        case [.left, .right]:
            image = UIImage(named: "road-horizontal")
        default:
            image = UIImage(named: "road")
        }
    }
}

#37 Adding the current locale to cache keys

🌍 When caching localized content in an app, it's a good idea to add the current locale to all keys, to prevent bugs when switching languages.

func cache(_ content: Content, forKey key: String) throws {
    let data = try wrap(content) as Data
    let key = localize(key: key)
    try storage.store(data, forKey: key)
}

func loadCachedContent(forKey key: String) -> Content? {
    let key = localize(key: key)
    let data = storage.loadData(forKey: key)
    return data.flatMap { try? unbox(data: $0) }
}

private func localize(key: String) -> String {
    return key + "-" + Bundle.main.preferredLocalizations[0]
}

#36 Setting up tests to avoid retain cycles with weak references

🚳 Here's an easy way to setup a test to avoid accidental retain cycles with object relationships (like weak delegates & observers) in Swift:

func testDelegateNotRetained() {
    // Assign the delegate (weak) and also retain it using a local var
    var delegate: Delegate? = DelegateMock()
    controller.delegate = delegate
    XCTAssertNotNil(controller.delegate)
    
    // Release the local var, which should also release the weak reference
    delegate = nil
    XCTAssertNil(controller.delegate)
}

#35 Expressively matching a value against a list of candidates

👨‍🔬 Playing around with an expressive way to check if a value matches any of a list of candidates in Swift:

// Instead of multiple conditions like this:

if string == "One" || string == "Two" || string == "Three" {

}

// You can now do:

if string == any(of: "One", "Two", "Three") {

}

You can find a gist with the implementation here.

#34 Organizing code using extensions

👪 APIs in a Swift extension automatically inherit its access control level, making it a neat way to organize public, internal & private APIs.

public extension Animation {
    init(textureNamed textureName: String) {
        frames = [Texture(name: textureName)]
    }
    
    init(texturesNamed textureNames: [String], frameDuration: TimeInterval = 1) {
        frames = textureNames.map(Texture.init)
        self.frameDuration = frameDuration
    }
    
    init(image: Image) {
        frames = [Texture(image: image)]
    }
}

internal extension Animation {
    func loadFrameImages() -> [Image] {
        return frames.map { $0.loadImageIfNeeded() }
    }
}

#33 Using map to transform an optional into a Result type

🗺 Using map you can transform an optional value into an optional Result type by simply passing in the enum case.

enum Result<Value> {
    case value(Value)
    case error(Error)
}

class Promise<Value> {
    private var result: Result<Value>?
    
    init(value: Value? = nil) {
        result = value.map(Result.value)
    }
}

#32 Assigning to self in struct initializers

👌 It's so nice that you can assign directly to self in struct initializers in Swift. Very useful when adding conformance to protocols.

extension Bool: AnswerConvertible {
    public init(input: String) throws {
        switch input.lowercased() {
        case "y", "yes", "👍":
            self = true
        default:
            self = false
        }
    }
}

#31 Recursively calling closures as inline functions

☎️ Defining Swift closures as inline functions enables you to recursively call them, which is super useful in things like custom sequences.

class Database {
    func records(matching query: Query) -> AnySequence<Record> {
        var recordIterator = loadRecords().makeIterator()
        
        func iterate() -> Record? {
            guard let nextRecord = recordIterator.next() else {
                return nil
            }
            
            guard nextRecord.matches(query) else {
                // Since the closure is an inline function, it can be recursively called,
                // in this case in order to advance to the next item.
                return iterate()
            }
            
            return nextRecord
        }
        
        // AnySequence/AnyIterator are part of the standard library and provide an easy way
        // to define custom sequences using closures.
        return AnySequence { AnyIterator(iterate) }
    }
}

Rob Napier points out that using the above might cause crashes if used on a large databaset, since Swift has no guaranteed Tail Call Optimization (TCO).

Slava Pestov also points out that another benefit of inline functions vs closures is that they can have their own generic parameter list.

#30 Passing self to required Objective-C dependencies

🏖 Using lazy properties in Swift, you can pass self to required Objective-C dependencies without having to use force-unwrapped optionals.

class DataLoader: NSObject {
    lazy var urlSession: URLSession = self.makeURLSession()
    
    private func makeURLSession() -> URLSession {
        return URLSession(configuration: .default, delegate: self, delegateQueue: .main)
    }
}

class Renderer {
    lazy var displayLink: CADisplayLink = self.makeDisplayLink()
    
    private func makeDisplayLink() -> CADisplayLink {
        return CADisplayLink(target: self, selector: #selector(screenDidRefresh))
    }
}

#29 Making weak or lazy properties readonly

👓 If you have a property in Swift that needs to be weak or lazy, you can still make it readonly by using private(set).

class Node {
    private(set) weak var parent: Node?
    private(set) lazy var children = [Node]()

    func add(child: Node) {
        children.append(child)
        child.parent = self
    }
}

#28 Defining static URLs using string literals

🌏 Tired of using URL(string: "url")! for static URLs? Make URL conform to ExpressibleByStringLiteral and you can now simply use "url" instead.

extension URL: ExpressibleByStringLiteral {
    // By using 'StaticString' we disable string interpolation, for safety
    public init(stringLiteral value: StaticString) {
        self = URL(string: "\(value)").require(hint: "Invalid URL string literal: \(value)")
    }
}

// We can now define URLs using static string literals 🎉
let url: URL = "https://www.swiftbysundell.com"
let task = URLSession.shared.dataTask(with: "https://www.swiftbysundell.com")

// In Swift 3 or earlier, you also have to implement 2 additional initializers
extension URL {
    public init(extendedGraphemeClusterLiteral value: StaticString) {
        self.init(stringLiteral: value)
    }

    public init(unicodeScalarLiteral value: StaticString) {
        self.init(stringLiteral: value)
    }
}

To find the extension that adds the require() method on Optional that I use above, check out Require.

#27 Manipulating points, sizes and frames using math operators

✚ I'm always careful with operator overloading, but for manipulating things like sizes, points & frames I find them super useful.

extension CGSize {
    static func *(lhs: CGSize, rhs: CGFloat) -> CGSize {
        return CGSize(width: lhs.width * rhs, height: lhs.height * rhs)
    }
}

button.frame.size = image.size * 2

If you like the above idea, check out CGOperators, which contains math operator overloads for all Core Graphics' vector types.

#26 Using closure types in generic constraints

🔗 You can use closure types in generic constraints in Swift. Enables nice APIs for handling sequences of closures.

extension Sequence where Element == () -> Void {
    func callAll() {
        forEach { $0() }
    }
}

extension Sequence where Element == () -> String {
    func joinedResults(separator: String) -> String {
        return map { $0() }.joined(separator: separator)
    }
}

callbacks.callAll()
let names = nameProviders.joinedResults(separator: ", ")

(If you're using Swift 3, you have to change Element to Iterator.Element)

#25 Using associated enum values to avoid state-specific optionals

🎉 Using associated enum values is a super nice way to encapsulate mutually exclusive state info (and avoiding state-specific optionals).

// BEFORE: Lots of state-specific, optional properties

class Player {
    var isWaitingForMatchMaking: Bool
    var invitingUser: User?
    var numberOfLives: Int
    var playerDefeatedBy: Player?
    var roundDefeatedIn: Int?
}

// AFTER: All state-specific information is encapsulated in enum cases

class Player {
    enum State {
        case waitingForMatchMaking
        case waitingForInviteResponse(from: User)
        case active(numberOfLives: Int)
        case defeated(by: Player, roundNumber: Int)
    }
    
    var state: State
}

#24 Using enums for async result types

👍 I really like using enums for all async result types, even boolean ones. Self-documenting, and makes the call site a lot nicer to read too!

protocol PushNotificationService {
    // Before
    func enablePushNotifications(completionHandler: @escaping (Bool) -> Void)
    
    // After
    func enablePushNotifications(completionHandler: @escaping (PushNotificationStatus) -> Void)
}

enum PushNotificationStatus {
    case enabled
    case disabled
}

service.enablePushNotifications { status in
    if status == .enabled {
        enableNotificationsButton.removeFromSuperview()
    }
}

#23 Working on async code in a playground

🏃 Want to work on your async code in a Swift Playground? Just set needsIndefiniteExecution to true to keep it running:

import PlaygroundSupport

PlaygroundPage.current.needsIndefiniteExecution = true

DispatchQueue.main.asyncAfter(deadline: .now() + 3) {
    let greeting = "Hello after 3 seconds"
    print(greeting)
}

To stop the playground from executing, simply call PlaygroundPage.current.finishExecution().

#22 Overriding self with a weak reference

💦 Avoid memory leaks when accidentially refering to self in closures by overriding it locally with a weak reference:

Swift >= 4.2

dataLoader.loadData(from: url) { [weak self] result in
    guard let self = self else { 
        return 
    }

    self.cache(result)
    
    ...

Swift < 4.2

dataLoader.loadData(from: url) { [weak self] result in
    guard let `self` = self else {
        return
    }

    self.cache(result)
    
    ...

Note that the reason the above currently works is because of a compiler bug (which I hope gets turned into a properly supported feature soon).

#21 Using DispatchWorkItem

🕓 Using dispatch work items you can easily cancel a delayed asynchronous GCD task if you no longer need it:

let workItem = DispatchWorkItem {
    // Your async code goes in here
}

// Execute the work item after 1 second
DispatchQueue.main.asyncAfter(deadline: .now() + 1, execute: workItem)

// You can cancel the work item if you no longer need it
workItem.cancel()

#20 Combining a sequence of functions

➕ While working on a new Swift developer tool (to be open sourced soon 😉), I came up with a pretty neat way of organizing its sequence of operations, by combining their functions into a closure:

internal func +<A, B, C>(lhs: @escaping (A) throws -> B,
                         rhs: @escaping (B) throws -> C) -> (A) throws -> C {
    return { try rhs(lhs($0)) }
}

public func run() throws {
    try (determineTarget + build + analyze + output)()
}

If you're familiar with the functional programming world, you might know the above technique as the pipe operator (thanks to Alexey Demedreckiy for pointing this out!)

#19 Chaining optionals with map() and flatMap()

🗺 Using map() and flatMap() on optionals you can chain multiple operations without having to use lengthy if lets or guards:

// BEFORE

guard let string = argument(at: 1) else {
    return
}

guard let url = URL(string: string) else {
    return
}

handle(url)

// AFTER

argument(at: 1).flatMap(URL.init).map(handle)

#18 Using self-executing closures for lazy properties

🚀 Using self-executing closures is a great way to encapsulate lazy property initialization:

class StoreViewController: UIViewController {
    private lazy var collectionView: UICollectionView = {
        let layout = UICollectionViewFlowLayout()
        let view = UICollectionView(frame: self.view.bounds, collectionViewLayout: layout)
        view.delegate = self
        view.dataSource = self
        return view
    }()
    
    override func viewDidLoad() {
        super.viewDidLoad()
        view.addSubview(collectionView)
    }
}

#17 Speeding up Swift package tests

⚡️ You can speed up your Swift package tests using the --parallel flag. For Marathon, the tests execute 3 times faster that way!

swift test --parallel

#16 Avoiding mocking UserDefaults

🛠 Struggling with mocking UserDefaults in a test? The good news is: you don't need mocking - just create a real instance:

class LoginTests: XCTestCase {
    private var userDefaults: UserDefaults!
    private var manager: LoginManager!
    
    override func setUp() {
        super.setup()
        
        userDefaults = UserDefaults(suiteName: #file)
        userDefaults.removePersistentDomain(forName: #file)
        
        manager = LoginManager(userDefaults: userDefaults)
    }
}

#15 Using variadic parameters

👍 Using variadic parameters in Swift, you can create some really nice APIs that take a list of objects without having to use an array:

extension Canvas {
    func add(_ shapes: Shape...) {
        shapes.forEach(add)
    }
}

let circle = Circle(center: CGPoint(x: 5, y: 5), radius: 5)
let lineA = Line(start: .zero, end: CGPoint(x: 10, y: 10))
let lineB = Line(start: CGPoint(x: 0, y: 10), end: CGPoint(x: 10, y: 0))

let canvas = Canvas()
canvas.add(circle, lineA, lineB)
canvas.render()

#14 Referring to enum cases with associated values as closures

😮 Just like you can refer to a Swift function as a closure, you can do the same thing with enum cases with associated values:

enum UnboxPath {
    case key(String)
    case keyPath(String)
}

struct UserSchema {
    static let name = key("name")
    static let age = key("age")
    static let posts = key("posts")
    
    private static let key = UnboxPath.key
}

#13 Using the === operator to compare objects by instance

📈 The === operator lets you check if two objects are the same instance. Very useful when verifying that an array contains an instance in a test:

protocol InstanceEquatable: class, Equatable {}

extension InstanceEquatable {
    static func ==(lhs: Self, rhs: Self) -> Bool {
        return lhs === rhs
    }
}

extension Enemy: InstanceEquatable {}

func testDestroyingEnemy() {
    player.attack(enemy)
    XCTAssertTrue(player.destroyedEnemies.contains(enemy))
}

#12 Calling initializers with dot syntax and passing them as closures

😎 Cool thing about Swift initializers: you can call them using dot syntax and pass them as closures! Perfect for mocking dates in tests.

class Logger {
    private let storage: LogStorage
    private let dateProvider: () -> Date
    
    init(storage: LogStorage = .init(), dateProvider: @escaping () -> Date = Date.init) {
        self.storage = storage
        self.dateProvider = dateProvider
    }
    
    func log(event: Event) {
        storage.store(event: event, date: dateProvider())
    }
}

#11 Structuring UI tests as extensions on XCUIApplication

📱 Most of my UI testing logic is now categories on XCUIApplication. Makes the test cases really easy to read:

func testLoggingInAndOut() {
    XCTAssertFalse(app.userIsLoggedIn)
    
    app.launch()
    app.login()
    XCTAssertTrue(app.userIsLoggedIn)
    
    app.logout()
    XCTAssertFalse(app.userIsLoggedIn)
}

func testDisplayingCategories() {
    XCTAssertFalse(app.isDisplayingCategories)
    
    app.launch()
    app.login()
    app.goToCategories()
    XCTAssertTrue(app.isDisplayingCategories)
}

#10 Avoiding default cases in switch statements

🙂 It’s a good idea to avoid “default” cases when switching on Swift enums - it’ll “force you” to update your logic when a new case is added:

enum State {
    case loggedIn
    case loggedOut
    case onboarding
}

func handle(_ state: State) {
    switch state {
    case .loggedIn:
        showMainUI()
    case .loggedOut:
        showLoginUI()
    // Compiler error: Switch must be exhaustive
    }
}

#9 Using the guard statement in many different scopes

💂 It's really cool that you can use Swift's 'guard' statement to exit out of pretty much any scope, not only return from functions:

// You can use the 'guard' statement to...

for string in strings {
    // ...continue an iteration
    guard shouldProcess(string) else {
        continue
    }
    
    // ...or break it
    guard !shouldBreak(for: string) else {
        break
    }
    
    // ...or return
    guard !shouldReturn(for: string) else {
        return
    }
    
    // ..or throw an error
    guard string.isValid else {
        throw StringError.invalid(string)
    }
    
    // ...or exit the program
    guard !shouldExit(for: string) else {
        exit(1)
    }
}

#8 Passing functions & operators as closures

❤️ Love how you can pass functions & operators as closures in Swift. For example, it makes the syntax for sorting arrays really nice!

let array = [3, 9, 1, 4, 6, 2]
let sorted = array.sorted(by: <)

#7 Using #function for UserDefaults key consistency

🗝 Here's a neat little trick I use to get UserDefault key consistency in Swift (#function expands to the property name in getters/setters). Just remember to write a good suite of tests that'll guard you against bugs when changing property names.

extension UserDefaults {
    var onboardingCompleted: Bool {
        get { return bool(forKey: #function) }
        set { set(newValue, forKey: #function) }
    }
}

#6 Using a name already taken by the standard library

📛 Want to use a name already taken by the standard library for a nested type? No problem - just use Swift. to disambiguate:

extension Command {
    enum Error: Swift.Error {
        case missing
        case invalid(String)
    }
}

#5 Using Wrap to implement Equatable

📦 Playing around with using Wrap to implement Equatable for any type, primarily for testing:

protocol AutoEquatable: Equatable {}

extension AutoEquatable {
    static func ==(lhs: Self, rhs: Self) -> Bool {
        let lhsData = try! wrap(lhs) as Data
        let rhsData = try! wrap(rhs) as Data
        return lhsData == rhsData
    }
}

#4 Using typealiases to reduce the length of method signatures

📏 One thing that I find really useful in Swift is to use typealiases to reduce the length of method signatures in generic types:

public class PathFinder<Object: PathFinderObject> {
    public typealias Map = Object.Map
    public typealias Node = Map.Node
    public typealias Path = PathFinderPath<Object>
    
    public static func possiblePaths(for object: Object, at rootNode: Node, on map: Map) -> Path.Sequence {
        return .init(object: object, rootNode: rootNode, map: map)
    }
}

#3 Referencing either external or internal parameter name when writing docs

📖 You can reference either the external or internal parameter label when writing Swift docs - and they get parsed the same:

// EITHER:

class Foo {
    /**
    *   - parameter string: A string
    */
    func bar(with string: String) {}
}

// OR:

class Foo {
    /**
    *   - parameter with: A string
    */
    func bar(with string: String) {}
}

#2 Using auto closures

👍 Finding more and more uses for auto closures in Swift. Can enable some pretty nice APIs:

extension Dictionary {
    mutating func value(for key: Key, orAdd valueClosure: @autoclosure () -> Value) -> Value {
        if let value = self[key] {
            return value
        }
        
        let value = valueClosure()
        self[key] = value
        return value
    }
}

#1 Namespacing with nested types

🚀 I’ve started to become a really big fan of nested types in Swift. Love the additional namespacing it gives you!

public struct Map {
    public struct Model {
        public let size: Size
        public let theme: Theme
        public var terrain: [Position : Terrain.Model]
        public var units: [Position : Unit.Model]
        public var buildings: [Position : Building.Model]
    }
    
    public enum Direction {
        case up
        case right
        case down
        case left
    }
    
    public struct Position {
        public var x: Int
        public var y: Int
    }
    
    public enum Size: String {
        case small = "S"
        case medium = "M"
        case large = "L"
        case extraLarge = "XL"
    }
}

Download Details:

Author: JohnSundell
Source code: https://github.com/JohnSundell/SwiftTips

License: MIT license
#swift 

Aayush Singh

Aayush Singh

1607579145

Keras Tutorial For Beginners | What is Keras | Keras Sequential Model | Keras Training

In this video on Keras, you will understand what is Keras and why do we need it, how to compose different models in Keras like the Sequential model and functional model, and later on how to define the inputs, how to connect layers over, and finally hands-on demo.
Why Keras is important

Keras is an Open Source Neural Network library written in Python that runs on top of Theano or Tensorflow. It is designed to be modular, fast, and easy to use. Keras is very quick to make a network model. If you want to make a simple network model with a few lines, Keras can help you with that.

Call Our Course Advisors IND: +91-7022374614 US: 1-800-216-8930 (Toll-Free) sales@intellipaat.com
Link: https://www.youtube.com/watch?v=nS1J-2uoKto

#keras tutorial for beginners #what is keras #keras sequential model #keras training

Keras Models - Types and Examples

A model is the basic data structure of Keras. Keras models define how to organize layers. In this article, we will discuss Keras Models and its two types with examples. We will also learn about Model subclassing through which we can create our own fully-customizable models.

Types of Keras Models

Models in keras are available in two types:

  • Keras Sequential Model
  • Keras Functional API

#keras tutorials #functional api in keras #keras models #models in keras