I'm working on an iOS app in Swift. The other app I created was one I did in Objective C and release around this time in 2014. Storyboards seem to have made UI stuff both easier and more complicated at the same time, so I'm trying to figure out current best practice in terms of view development.
Size classes and constraints seem like almost a necessary timesaver at this point for multiple screen sizes. Back in 2014, this was less the case, and programmatically keeping track of UI layouts as CGRect code made programmatically mucking with UI layout much simpler and better for code reuse vs creating a whole new view controller for just adding new UI elements to much the same view. Doing the same thing with constraint code seems less appealing, but also necessary if I want more code reuse there.
So I'm wondering what the current practice is here as I'm just thinking in terms of code reuse. Programmatic constraints just seem less elegant than storyboard defined ones, but I'm not sure they're the end all be all for UI code since they seem problematic for programmatically updating the UI on the fly.
Is the best strategy at this point to enclose everything in layout preserving superviews and keep most of this storyboard centric or would it still make sense to do swift programmatic code for these layouts since I'd have to for iPad and iPhone specific changes anyway? On that subject would it still make sense to split drastically different UIs into multiple storyboards (e.g. 2 different iPad and iPhone storyboards as that was a default at one point)?
Thanks in advance for answers to this. Device specific stuff just seems to not always be code reuse, but I just want reuse to be relatively simple I guess. Otherwise I'm just creating more swift classes than I strictly need.
This is inherently opinion based. Everything from no Interface Builder to only StoryBoards is likely being used in production applications, and you can make just about anything work.
My personal tendency is to use Storyboards for everything other than TableView/CollectionView cells. I find it removes almost all interface boilerplate code from my classes and makes it so I only need to handle the interface between my ViewControllers. Here are the guidelines I generally try and follow: (again...opinion)
Use multiple storyboards organized in a useful way:
Large storyboard become hard to maintain and performance while editing suffers noticeably. We have the option to use StoryboardReferences, so might as well use them.
The more Scenes and ViewControllers, the better. (within reason)
This makes things more maintainable and re-usable. E.g. a header that could be used in multiple Scenes. It can be a bit annoying to have to use containerVCs and deal with segues everywhere, but I rarely regret separating something into it's own ViewController.
Don't use separate interface files for different sizes.
It's more code to maintain and will force you to swap ViewControllers in-out if you want to support resizing on iPad (or any future devices). Not to mention going forward it's clear Apple is assuming you're using size classes rather than swapping ViewControllers and is clearly where the future is headed in Apple platforms development
When you need animations, try and narrow it down to changing a single constraint.constant
This greatly simplifies your code and usually avoids having to deal with size-classes anywhere other than the storyboards. It's not always possible for complicated animations, but you can do a surprising amount if you're willing to mess around. It makes it possible to narrow down a toggling action to 2 lines of code, which is quite nice. Using StackViews can also help with this a lot.
The other thing to focus on is avoiding as much of the stringly-typed nature of IB as possible. This is much easier in Swift and there are some decent solutions using string-backed enums and extensions, but specifics are probably out of the scope of this question.
This question already has answers here:
Adding Views. Storyboard VS. Programmatically [closed]
(2 answers)
Closed 6 years ago.
Is it more efficient to add views and subviews programmatically than using a storyboard to add them ?
What is be the best practice ? Does the best practice depend on the situation ? If so, when is it preferable to use the one method over the other ?
Storyboard vs Code comes up an awful lot. And I think looking at it as which is better often starts flame wars. So instead I'll list out pros and cons for each, then you can hopefully make an informed decision.
This is by no means an exhaustive list. I'd love to see edits or comments adding additional bullet points.
Storyboard Pros
Storyboards are great for getting a concept working quickly.
They're great for seeing your actual app and previewing it on multiple devices.
They make it really easy to customize appearances and explore whats possible.
They're fantastic for beginners as they remove one barrier and let you focus on code for your app rather than some boiler plate UI.
Storyboard Cons
Storyboards and Xibs are not great for merging. Its possible to read the XML and make decisions about merge conflicts, but its certainly not enjoyable.
Overtime as you customize your app and build the custom parts that make it special, Storyboards can't keep up. This means that the benefit of seeing and previewing your app as you see it on device slowly loses its value as more and more of your app is done in code out of necessity.
Storyboards don't have a huge performance hit for actual users, but they definitely slow Xcode down on even the fastest machines.
You can't customize everything in Interface Builder that you can in code, so its inevitable you'll have some code and some Storyboard customization. Later when you want to change something, you'll have to check at least two places for the right place to change it.
If you're using the same fonts and colors throughout the app, its easy to change in code in one place. In Storyboard you'll need to change it for every single label or view you've set up. You can easily set fonts and colors in code while using Storyboard for the rest of your layout, but over time you'll find your storyboards represent less and less of your actual app.
While its possible to have pixel perfect designs in Storyboard, drag and drop isn't as precise as entering specific numbers. Storyboard does support entering those numbers, but you need to navigate to multiple places and if you drag it later the numbers are all reset.
Reusing views typically involves copy and pasting them. These causes issues later when you need to make changes.
Showing and hiding views at certain times is doable, but again leaves your storyboard in an unrealistic state compared to your actual app.
Programmatic Pros
All your code is in one place making it easier to debug or change later.
Subclasses and custom properties are no different from first class views and properties. Your customizations will be easy to manage.
Views that are reused throughout the app only need to be created once.
Merge conflicts are often easy to understand and make decisions.
Dependency Injection is one of the safest ways to instantiate UIViewControllers and ensure necessary information is passed in, its not possible with Storyboards leading to less straight forward and more bug prone code.
Programmatic Cons
The additional complexity can be a stumbling block for beginners.
Writing a few lines of code can take longer than clicking a few checkboxes.
When working with layouts, you'll need to visualize it in your head or run your app to see what your layout looks like as you make changes. This can be difficult or slower for some.
Storyboard exposes checkboxes and buttons for many customizable properties like background color, font, etc. This makes it easy to see at a glance what you can customize. (Though as mentioned above its not everything) In Code, you'll need to look at documentation or headers to understand what you can customize.
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 7 years ago.
Improve this question
So I'm new to iOS development and I found it easier to write the views all programmatically. So my views have UIViews, ScrollViews, UIButton, UILabel all created and positioned programmatically. (So I never used AutoLayouts).
I've now pretty much finished my app and want to make the iPad views, and realized maybe it was a bad idea to do it like this.
Is this bad practice or should I really be using auto layouts as much as I can?
If it's ok to do it how I'm doing it right now, what is the correct way to add different views for iOS and iPad? I've seen this answer below on how to find the device, is a simple if else statement sufficient? - iOS: How to determine the current iPhone/device model in Swift?
I am using programatic views in a live app and its awesome. A bunch of people I know us this as well.
Here is a little algorithm I use to choose between the two methods:
Are you building a fast app for a client or a hobby? Use storyboard with autolayout.
Are you building an open source project that will be used by many people? Use programatic UI
Are you building an app for the long run? (1+ year) Use programatic UI
Its also harder to make an app thats supposed to be rotated without autolayout. Because doing that with code takes much more work than autolayout. Most good apps dont use this feature anyways so I don't see much problem.
A good tip is never to use constants while writing programatic UI.
If you are going to make a button thats 100px in width, do not type in 100px anywhere in the code. Instead figure out the screen sizes and place the main views according to screen sizes. Then place the subviews or secondary views according to the position of the main views. If you do this correctly you will have more powerful multidevice layout support than autolayout.
Here is a little library I wrote, please inspect and play with the code on how I place the view: https://github.com/goktugyil/CozyLoadingActivity/blob/master/CozyLoadingActivity.swift
Also here is a good article I like about this:
http://www.toptal.com/ios/ios-user-interfaces-storyboards-vs-nibs-vs-custom-code
This is only fine, If you have enough time, patience, good skill on calculations and relationship configurations between different UI elements.
However, using Auto Layout is pretty useful and low time consuming than manual calculations.
We can easily create a dynamic and versatile interface that responds appropriately to changes in screen size, device orientation, and localization with minimal effort.
Read Adopting Auto Layout,to implement the Auto Layout in your existing application
TL;DR
It depends
Longer version
Clearly one size does not fit all. AutoLayout is pretty powerful both in Interface Builder and code (be it visual language format, or simple constraint setting), but sometimes it just seems like you're "rubbing your right ear with your left hand" - that's when adding views programmatically comes in. But beware of not having it different in each view controller - you don't want to introduce too much complexity in your project, right?
I personally like using AutoLayout as much as I can and whenever I can't use it anymore, or the StoryBoard View would get too messed up with millions of constraints, I try to separate views into containers - have the container be resized by AutoLayout and have the subviews handled by code.
Example would be a custom Media Player - maybe I want to have two stripes below and above the video - I could have the video and two UIView extended stripes handled by the AutoLayout. But the subviews (controls) in the stripes themselves would be added by code. It gives me the control over my code but still it doesn't introduce too much complexity.
First of all - if you want to develop for iOS you have to learn Autolayout. There are already a lot of different devices with different resolutions and maybe will be even more in future.
Secondary - if you want to work effectively with IB you have to read guide / watch tutorial videos and have some practice. It could be difficult to start but then you will realize that IB is powerful, fast and often the best way to develop GUI. Often, but not always!
Code advantages:
Easy to copy-paste and reuse GUI. It could be critical if you have
several similar views or want to reuse some old code.
Easy to resolve merge conflicts and check commits.
Easier to make styles - like the same font for all labels depending on the country.
More powerful (there are things that could not be done with IB) so you have to use it sometimes.
IB advantages:
You can see your GUI during development for different resolutions/localizations so you do not have to compile and run a project on different devices/simulators to check is GUI ok or not. Also IB will show you warnings if you forget some Autolayout constraints or there are conflicts. Saves a lot of time if you have a complex GUI with non-trivial Autolayout constraints.
It is much easier to understand someones else code if it is developed in IB. Especially important for complex GUI - not so easy to find a required label or button in a few hundreds lines of code.
Small bonus - if you want to use a custom control developed via code you can make it IBInspectable and use it without problems in IB
Just to summarize - if you do not need IB benefits (for example GUI is quite simple and does not use Autolayout) developing GUI via code could be easier and faster. But in case you have to support different resolutions and/or you have hundreds lines of GUI code in each view controller I strongly recommend to try IB.
I'm in the process of developing an iPad-only survey-app using MonoTouch. With monotouch.dialog (mt.d) I found that building these interfaces can come quickly, which is awesome.
However... I also found that mt.d only does about 80% of what I want. Makes me wonder: should I invest in extending mt.d to my needs or should I choose something differently over mt.d?
Some of my requirements:
Radiogroups without transitions: I like the options to be
presented right away (there's more than enough space on the iPad
screen)
A rating UI control, such as
http://www.cocoacontrols.com/platforms/ios/controls/dyrateview
Mixed radiogroups: like 3 predefined elements and a fourth which
allows for manually added content
What are your thoughts on this? Can this be done easily (I'm a trained programmer, but quite new to both C# and iOS development)? Do you guys know of any online repositories of custom UI components with C#/MonoTouch bindings?
Thanks a lot!
This is of course a subjective opinion, but my take on it is that if you believe you can do your UI in UITableView (which MonoTouch.Dialog is based on), then you should go for MonoTouch.Dialog. If UITableView will not fit your needs, you should look for a different approach. MonoTouch.Dialog is quite flexible, and open-source, so if you need anything to be different you can just use the source code and modify it at will.
Are there any guidelines on when to use storyboards in an iOS project and when to use XIBs? what are the pros and cons of each and what situations do they each suit?
Near as I can tell it's not that clean to use storyboard segues when you have view controllers being pushed by dynamic UI elements (Like map pins).
Update 1/12/2016: It's 2016 and I still prefer laying out my UIs in code and not in Storyboards. That being said, Storyboards have come a long way. I have removed all the points from this post that simply do not apply anymore in 2016.
Update 4/24/2015: Interestingly Apple doesn't even use Storyboards in their recently open-sourced ResearchKit as Peter Steinberger has noticed (under the subheading "Interface Builder").
Update 6/10/2014: As expected, Apple keeps improving Storyboards and Xcode. Some of the points that applied to iOS 7 and below don't apply to iOS 8 anymore (and are now marked as such). So while Storyboards inherently still have flaws, I revise my advice from don't use to selectively use where it makes sense.
Even now that iOS 9 is out, I would advise against to use caution when deciding whether to use Storyboards. Here are my reasons:
Storyboards fail at runtime, not at compile time: You have a typo in a segue name or connected it wrong in your storyboard? It will blow up at runtime. You use a custom UIViewController subclass that doesn't exist anymore in your storyboard? It will blow up at runtime. If you do such things in code, you will catch them early on, during compile time. Update: My new tool StoryboardLint mostly solves this problem.
Storyboards get confusing fast: As your project grows, your storyboard gets increasingly more difficult to navigate. Also, if multiple view controllers have multiple segues to multiple other view controllers, your storyboard quickly starts to look like a bowl of spaghetti and you'll find yourself zooming in and out and scrolling all over the place to find the view controller you are looking for and to find out what segue points where. Update: This problem can mostly be solved by splitting your Storyboard up into multiple Storyboards, as described in this article by Pilky and this article by Robert Brown.
Storyboards make working in a team harder: Because you usually only have one huge storyboard file for your project, having multiple developers regularly making changes to that one file can be a headache: Changes need to be merged and conflicts resolved. When a conflict occurs, it is hard to tell how to resolve it: Xcode generates the storyboard XML file and it was not really designed with the goal in mind that a human would have to read, let alone edit it.
Storyboards make code reviews hard or nearly impossible: Peer code reviews are a great thing to do on your team. However, when you make changes to a storyboard, it is almost impossible to review these changes with a different developer. All you can pull up is a diff of a huge XML file. Deciphering what really changed and if those changes are correct or if they broke something is really hard.
Storyboards hinder code reuse: In my iOS projects, I usually create a class that contains all the colors and fonts and margins and insets that I use throughout the app to give it a consistent look and feel: It's a one line change if I have to adjust any of those values for the whole app. If you set such values in the storyboard, you duplicate them and will need to find every single occurrence when you want to change them. Chances are high that you miss one, because there's no search and replace in storyboards.
Storyboards require constant context switches: I find myself working and navigating much faster in code than in storyboards. When your app uses storyboards, you constantly switch your context: "Oh, I want a tap on this table view cell to load a different view controller. I now have to open up the storyboard, find the right view controller, create a new segue to the other view controller (that I also have to find), give the segue a name, remember that name (I can't use constants or variables in storyboards), switch back to code and hope I don't mistype the name of that segue for my prepareForSegue method. How I wish I could just type those 3 lines of code right here where I am!" No, it's not fun. Switching between code and storyboard (and between keyboard and mouse) gets old fast and slows you down.
Storyboards are hard to refactor: When you refactor your code, you have to make sure it still matches what your storyboard expects. When you move things around in your storyboard, you will only find out at runtime if it still works with your code. It feels to me as if I have to keep two worlds in sync. It feels brittle and discourages change in my humble opinion.
Storyboards are less flexible: In code, you can basically do anything you want! With storyboards you are limited to a subset of what you can do in code. Especially when you want to do some advanced things with animations and transitions you will find yourself "fighting the storyboard" to get it to work.
Storyboards don't let you change the type of special view controllers: You want to change a UITableViewController into a UICollectionViewController? Or into a plain UIViewController? Not possible in a Storyboard. You have to delete the old view controller and create a new one and re-connect all the segues. It's much easier to do such a change in code.
Storyboards add two extra liabilities to your project: (1) The Storyboard Editor tool that generates the storyboard XML and (2) the runtime component that parses the XML and creates UI and controller objects from it. Both parts can have bugs that you can't fix.
Storyboards don't allow you to add a subview to a UIImageView: Who knows why.
Storyboards don't allow you to enable Auto Layout for individual View(-Controller)s: By checking/unchecking the Auto Layout option in a Storyboard, the change is applied to ALL controllers in the Storyboard. (Thanks to Sava Mazăre for this point!)
Storyboards have a higher risk of breaking backwards compatibility: Xcode sometimes changes the Storyboard file format and doesn't guarantee in any way that you will be able to open Storyboard files that you create today a few years or even months from now. (Thanks to thoughtadvances for this point. See the original comment)
Storyboards can make your code more complex: When you create your view controllers in code, you can create custom init methods, for example initWithCustomer:. That way, you can make the customer inside of your view controller immutable and make sure that this view controller cannot be created without a customer object. This is not possible when using Storyboards. You will have to wait for the prepareForSegue:sender: method to be called and then you will have to set the customer property on your view controller, which means you have to make this property mutable and you will have to allow for the view controller to be created without a customer object. In my experience this can greatly complicate your code and makes it harder to reason about the flow of your app. Update 9/9/16: Chris Dzombak wrote a great article about this problem.
It's McDonald's: To say it in Steve Jobs' words about Microsoft: It's McDonald's (video)!
These are my reasons for why I really don't like working with storyboards. Some of these reasons also apply to XIBs. On the storyboard-based projects that I've worked on, they have cost me much more time than they have saved and they made things more complicated instead of easier.
When I create my UI and application flow in code, I am much more in control of what is going on, it is easier to debug, it is easier to spot mistakes early on, it is easier to explain my changes to other developers and it is easier to support iPhone and iPad.
However, I do agree that laying out all of your UI in code might not be a one-size-fits-all solution for every project. If your iPad UI differs greatly from your iPhone UI in certain places, it might make sense to create a XIB for just those areas.
A lot of the problems outlined above could be fixed by Apple and I hope that that's what they will do.
Just my two cents.
Update: In Xcode 5, Apple took away the option to create a project without a Storyboard. I've written a small script that ports Xcode 4's templates (with Storyboard-opt-out option) to Xcode 5: https://github.com/jfahrenkrug/Xcode4templates
I have used XIBs extensively and completed two projects using Storyboards. My learnings are:
Storyboards are nice for apps with a small to medium number of screens and relatively straightforward navigation between views.
If you have lots of views and lots of cross-navigation between them the Storyboard view gets confusing and too much work to keep clean.
For a large project with multiple developers I would not use Storyboards because you have a single file for your UI and cannot easily work in parallel.
It might be worth for large apps to split up into multiple storyboard files but I have not tried that. This answer shows how to do segues between storyboards.
You still need XIBs: In both of my Storyboard projects I had to use XIBs for custom table cells.
I think Storyboards are a step in the right direction for UI implementation and hope Apple will extend them in future iOS versions. They need to resolve the "single file" issue though, otherwise they won't be attractive for larger projects.
If I start a small size app and can afford iOS5 only compatibility, I would use Storyboards. For all other cases I stick to XIBs.
Storyboards were created to help developers visualize their application and the flow of the application. It is alot like having a bunch of xib but in a single file.
There is a question similar to this located What is the difference between a .xib file and a .storyboard?.
You can also create custom transitions via code that will change dynamically if needed, much like you can with .xibs.
PROS:
You can mock up flow of an application without writing much, if any code.
Much easier to see your transitions between screens and your application flow.
Can also use .xibs if needed with storyboards.
CONS:
Only works with iOS 5+. Does not work with iOS4.
Can get cluttered easily if you have a very view intensive application.
There really isn't a right / wrong when to use one or the other, it is just a matter of preference and what iOS versions you are wanting to use.
I will just state 4 simple reasons why you should use storyboards, especially in a productive environment where you have to work in a team of product owners, product managers, UX designers, etc.
Apple has GREATLY improved working with Storyboards. And they encourage you to work with them. Which means they will not break your existing projects with updates, they will ensure that storyboards are future proof for newer XCode/iOS versions.
More visible results in less time for the product owners and managers, even during the creation phase. You can even use the storyboard itself as a screenflow diagram and discuss it in meetings.
Even after an app is done (and that's generally where its life-cycle begins) – in the future it will be faster and easier to apply small adjustments. And these could very well change multiple aspects of your layout at the same time, which you probably want to see in a WYSIWYG manner. The alternative would be hand-writing UI changes in code and switching back and forth between the IDE and the simulator to test it out, each time waiting for compile & build.
Non-developers can be taught to set up layouts in storyboards and create the necessary hooks for the developers (IBOutlets and IBActions). That's a very big plus because it lets the devs focus on the logic and the UX designers apply their changes in a visual manner, without having to write any code at all.
I won't write up any CONS, since Johannes has already listed probably all the viable ones in his answer. And most of them are definitely not viable, especially not with XCode6's major improvements.
I don't think there is a right answer for your question, it's just a matter of personal experience and what you feel more confortable with.
In my opinion, Storyboards are a great thing. It's true, it's really hard to find out why your app is misteriously crashing at runtime, but after some time and experience you'll realize it's always related to some IBOutlet missing somewhere and you'll be easily able to fix it.
The only real issue is working in team under version control with storyboards, in the early stages of development it could be a real mess. But after that first stage, UI updates that completely changes the storyboard are very rare, and in most cases you end up with conflicts in the very last parts of the xml, which are segue references that usually autofix themselves when you re-open the storyboard. In our team work we prefered to deal with this instead of heavy view-controllers with tons of view code.
I've read many comments againts auto-layout. With XCode5 it got really improved, It's really good even for autorotating layouts. In some case you'll have to do something in code, but you can simply outlet the constraint you need to edit and, at that point, do what you need in your code. Even animate them.
I also think that most of the people who dislike storyboards didn't fully try to understand the power of a custom manual segue, where you can totally customize (in a single file) the way you transition from a way to another and also (with some tricks) even reuse a previously loaded view controller by just updating it's view contents instead of fully reload the whole thing.
At the end you can really do the same things as in code, but I think you have a better separation of concerns with storyboards, but I agree that in many things they lack of features (fonts, image as color background, ecc...).
I am not using StoryBoard or XIBs in my any of the app.. but creating everything programmatically.
∆ Benefits :
√ You can create any complex kind of UI or transition animations for UIView's.
√ Support all iOS versions. No need to worry about < iOS 5.
√ *Your app would support all iPhone/iPod/iPad devices within your code.
√ You're always updated as you know the code that'll always work.
√ *Will work on any (new) device launched – No need to change in code.
√ Everything is upto you. At certain place you want to change something – No need to look into storyboard or xib. Just search for it in particular class.
√ Last but not the list – You'll never forget that, how to manage everything programmatically. This is the best thing as you know a control very deep then anyone.
I've never find a problem by not using SB or XIBs as I'm good with this.
* if you've set UIKit's object frames according to screen size.
P.S. If you've still not done this thing – you may faced difficulty (or may feel boring) but once you get familiar with this – its really a Candy for you.
If you are about to care about Storyboard performance, watch WWDC 2015 Session 407
Build Time
When interface builder is compiling a storyboard it's doing two things
first, it's trying to maximize the performance of your application and
secondly it's also minimizing the number of nib files created.
If I have a view controller with a view and a bunch of sub views,
interface builder, the build time is going to create a nib file for
the view controller and create a nib file for the view.
By having separate nib files for both the view controller and the
view, this means the view hierarchy can be loaded on demand.
Run Time
When you allocate a storyboard instance using UI storyboard, API,
initially all you are allocating memory for is the UI storyboard
instance itself.
No view controllers no views yet.
When you instantiate your initial view controller it will load the nib
for that initial view controller but, again, no view hierarchy has
been loaded yet until someone actually asks for it.
I have been working on a reasonably sized project (>20 scenes in storyboard parlance), and have come across many limitations and have to repeatedly go to documentation and google searches for doing things.
The UI is all in one file. Even if you create multiple storyboards, you still have many scenes/screens in each storyboard. This is a problem in medium-large teams.
Secondly, they do not play well with custom Container Controllers which embed other container controllers etc. We're using MFSlideMenu in a Tabbed application and the scene has a table. This is almost impossible to do with a storyboard. Having spent days, I've resorted to doing the XIB way where there is complete control.
The IDE does not allow to select controls in zoomed-out state. So, in a large project, the zoom-out is mostly to get a high level view and nothing more.
I would use storyboards for smaller applications with small team sizes and use XIB approach for medium-large teams/projects.
If you want to reuse some UI in multiple view controllers then you should use XIBs