Is it possible to create an API inside an iOS app to let 3rd party developers create app add-on that are downloadable for users inside the app?
I could not find any ressources for this topic online.
I think that's going to go against the app store guidelines if your intention is that the 3rd party developers write code, which will be downloaded and executed.
There's something of a grey area between legitimate uses and illegitimate uses of downloaded code:
Legitimate Example 1: Something like Hopscotch where children are creating simple fun shared games. These can be considered user generated content.
Legitimate Example 2: A game which is driven by scripts allows for the run-time downloading of patches to fix bugs in the scripts or even to add new levels. I believe such usages have been rejected occasionally in the past, but are generally accepted these days.
The relevant guideline from the guidelines is (emphasis mine):
2.5.2 Apps should be self-contained in their bundles, and may not read or write data outside the designated container area, nor may they download, install, or execute code, including other iOS, watchOS, Mac OS X, or tvOS apps.
And the all-encompassing:
We will reject apps for any content or behavior that we believe is over the line. What line, you ask? Well, as a Supreme Court Justice once said, "I'll know it when I see it". And we think that you will also know it when you cross it.
If your idea is to create some sort of app-store within your app-store app, then I would abandon it immediately, because that's going to be way over Apple's line.
That said, you haven't given much detail about your app, so I might have made wrong assumptions about where you're thinking of going.
Related
This may seem like an odd question, but I'm in the middle of creating an iOS app, and was wondering if there is a way to, in the future, roll out changes to the app without requiring all of the users to download an update.
I've noticed that Snapchat can do this with their filters - new filters are added regularly, without me updating the app.
I've read into 'Cloud code', something Parse had that apparently let you accomplish this. Obviously that's no longer an option.
Also, do Apple even allow this? Seeing as they need to approve every app before letting it onto the App Store, it would seem like they would need to approve any changes first too.
I've thought of strange things like storing a function in a database, then getting the app to download that function and run it - naturally if I were to now change the function in the database, it would change on all users devices instantly. Just how ridiculous is that idea? Thanks!
Depends on how much flexibility you want.
For example, Spotify does this for UI mainly - backend-driven UI, as they call it. They send a kind of layout from the server and convert it to a real iOS layout, based of a predefined mapping. You can find pretty more details in the Spotify's presentation.
Slides: http://www.slideshare.net/JohnSundell/backenddriven-native-uis
Video (more details): http://www.downvids.net/backend-driven-native-uis-john-sundell-and-diego-cristina-ca--777281.html
So actions can also be divided in similar pieces and abstractions, received from the server and interpreted - sort of scripting is needed. You can't compile and run arbitrary functions on the fly.
Hope it'll give you some ideas you can think of.
See Apple's app store review guidelines: https://developer.apple.com/app-store/review/guidelines/ section 2.4.5 (iv). Apps are not supposed to download code or resources to add functionality or significantly change the app from what Apple sees during the review process. And executable memory isn't even writable by sandboxed apps, which makes downloading compiled functions fairly useless. But downloading Javascript that complies with the above guidelines seems to be allowed.
At first, I will describe my use-case on why I need to add the framework at runtime on iOS.
Let's say I have an app on iOS device. The app requires some 3rd party frameworks to add some external features to it. Now, the features are many. So, the required number of frameworks will be many too. An user may not need lots of features. Just a small set of features. Plus, lots of framework will require a lot of space. The application will be huge in size.
For an example, an user needs only 1 feature. The application provides 100. So, all the other frameworks will definitely be unnecessary.
So, the solution would be to download the frameworks and the necessary files on demand from an online repository, link them on runtime and use them. This would mean the application size would be very small and not bulky with unnecessary stuff.
But does iOS provide that? I have to add an external framework and the necessary files that is not on the app use them on runtime.
Is this possible? Can anyone provide me with some resources on how I can do that?
I have seen some resources on SO and some other sites. The results are not so helpful.
"But does iOS provide that?" - no
"Is this possible?" - partially
"Can anyone provide me with some resources on how I can do that" - unfortunately no
More details: That is certainly not possible out of the box. And it will be firstly a pain in the ass to do achieve what you are trying and (far more important) secondly it will probably not be allowed by Apple. You are not allowed to load new "program code" into the app at runtime. For the simple reason that Apple cannot review what you are about to load. And it would a huge security risk for your users if your backend gets compromised.
What you can do is load resources like images, videos etc. on demand.
You might even be able to come up with some scripting behavior of your app. The server could theoretically deliver code to your app and then your app interprets that code. But that code would not be Swift but some other language that you have to define first. You see where this is going? That is possible while not really doable. And still it might get you rejected anyway.
In particular if you are talking about actual frameworks that include actual binary code is is not possible.
I'm developing a sophisticated engineering calculation iPad app for a specific customer (much like the poster of this question). I'm getting ready to release a first version to them, and trying to get this distribution issue sorted out. There are pretty much 5 methods covered throughout that post (general App Store, B2B, Enterprise, Ad Hoc, and stuff requiring Jail Breaking), and most of the searching I've done has shown those same 5 options. I've been looking into TestFlight to manage the Ad Hoc distribution for beta testing right now, but down the road I'm going to need to allow my customer to mass install my app onto as many devices as desired (hopefully without needing to involve me at all by that point). So in the end, none of these 5 options are really attractive as a final solution.
A colleague found this App Doc that mentions something call packagemaker, which sounds like it's a Terminal tool, but I can't really find much on it (the Apple Doc simply says "create the package as you normally do", which is pretty much meaningless since I normally don't - and everything else I can find is from around 2009, and I've learned that anything described in those old posts is almost certainly wrong by this point when it comes to the Apple world). From what I can tell, packagemaker should let me build an installer executable that I can send to my customer, and they could just run that installer and it would put the app on their device. This is how we currently distribute similar software for Windows environment devices, and it would be ideal for what I'm trying to do.
Does anyone know if packagemaker can be used as I've described to install iOS apps? If so, can anyone point me toward a beginner-orientated tutorial on how to use it?
Packagemaker sounds like overkill. The old tuts should still work though, it hasn't changed much (or at all) since 10.5. It's a separate download from Apple Developer center.
You could use testflight, though I presume you want to install B2B apps, and not ad-hoc apps for testing.
In that case just create a bash script that builds and/or deploys your app to a connected device with xcodebuild and related tools. If you're concerned that your end user will find Terminal use appalling, write a tiny Cocoa app to support & run the script, and put the script and your iPad app into the bundle.
For example the app's User Interface could query the user for paths, the device, and whatever is necessary to deploy your app and pass it to the bash script which you can run easily with NSTask.
This is certainly going to be many times faster than getting into PackageMaker, and figuring out how it works - cause it really doesn't work in a straightforward way, unfortunately.
OSX has functionality not available in the iOS SDK, and occasionally I'll find an answer like this one: no route.h on the iPhone SDK, that says "just copy the header file from the Simulator (or OSX) SDK and it'll work fine."
That seems suspect to me on multiple levels, including App Store approval, but then I read something like this from an Apple employee who says "if you're using sys/route.h declarations on iOS for an App Store app, please get in touch with me...", which sounds like tacit approval (or a ruse to catch misbehavior :)
Anyone know where the official line really is, and whether something like this is at all safe to do in production code? I'm particularly interested in lower-level BSD functions and whatnot for portability, as opposed to undocumented Objective-C methods.
I would not regard that comment on the Apple forums as tacit approval. It sounds like they're just trying to decide if some additional BSD APIs should be added for use in iOS apps.
In general, if there isn't an iOS header available for a function, then it's considered a Private API (for example, something under https://developer.apple.com/library/ios/). The official policy on Private APIs is that they're not to be used in apps distributed through the App Store.
Now, you can certainly use them for personal/hobby apps, or apps that you deploy in-house (Enterprise Distribution).
If the APIs you're talking about are APIs that are publically documented for OS X, and you can get them to work for iOS by copying headers, then they're probably part of the Darwin source base. It's very likely that they'll continue to function in future versions of iOS, although that's not guaranteed. Of course, nothing is guaranteed (really), as public APIs get deprecated, too.
Then, there's the issue that not all review checks are automated (I don't work for Apple, but can deduce this from things I've seen get through review). It isn't that unusual for apps to be approved with Private API usage, although if they get popular, Apple frequently pulls those apps from the store within a couple weeks. The review process isn't perfect.
So, my answer is that if you're submitting to the app store, don't expect copying in headers to work.
P.S. If you can tell us specifically which BSD function you're referring to, we might be able to give you a better answer.
P.P.S. The answer you link to about Route.h is from Grant Paul, who writes quite a bit of non-App Store software.
Additional useful information on Private APIs and the App Store
Apple never knows what header files you use - the danger of course is the definition from the copied file is close to, but not exactly, what iOS uses. If this file is important, or a few select files, post a question on Apple's internal forums and you will surely get an answer to such a question. Failing that, burn a DTS incident (you get two a year, I almost never use mine).
Apple's iOS developer guidelines state:
3.3.2 — An Application may not itself install or launch other executable code by any means, including without limitation through the use of a plug-in architecture, calling other frameworks, other APIs or otherwise. No interpreted code may be downloaded or used in an Application except for code that is interpreted and run by Apple’s Documented APIs and built-in interpreter(s).
Assuming that downloading data - like XML and images, or a game level description, for example - at run-time is allowed (as is my impression?), I am wondering where they draw the line between "data" and "code". Picture the scenario of an app that delivers interactive "presentations" to users (like a survey, for instance). Presentations are added continuously to the server and different presentations are made available to different users, so they cannot be part of the initial app download (which would be the whole point). They are described in XML format, but being interactive, they might contain conditional branching of this sort (shown in pseudo form to exemplify):
<options id="Gender">
<option value="1">Male</option>
<option value="2">Female</option>
</options>
<branches id="Gender">
<branch value="1">
<image src="Man" />
</branch>
<branch value="2">
<image src="Woman" />
</branch>
</branches>
When this XML is interpreted and "played" within the app, the above would be presented in two steps. First a selection screen is shown, where the user can click on either of the two choices ("Male" or "Female"). Next, an image will be [downloaded dynamically] and displayed based on the choice made in the previous step.
Now, from this, it's easy to imagine additional tags, describing further logic still. For example, a containing tag could be added:
<loop count="3">
<options... />
<branches... />
</loop>
The result here being that the selection screen / image screen pair would be sequentially presented three times over, of course.
Or imagine some format describing a level in a game. It is perhaps natural to view that as passive "data", but if it includes, say, several doorways that the user can go through and with various triggers, traps and points attached to them etc - isn't that the same as using a script (or, indeed, interpreted code) - to describe execution sequences, options and their conditional responses?
Assuming that the interpretation engine for the data is already present in the app and that such "presentations" can only be consumed (not created or edited) in the app, how would this fare against Apple's iOS guidelines? Doesn't XML basically constitute a scripting language in this sense (couldn't any program in an interpreted language be described in XML)?
Would it be OK if the proprietary scripting language (ref the XML used above) was strictly sandboxed (how can they tell?) and not given access to the operating system in any way (but able to download content - like a survey or a game level - dynamically as well as upload results - answers or scores - to the authoring server)?
Where does the line go?
Update as of WWDC 2017
Programming tools such as Codea mentioned below are now explicitly allowed to download code. The App Store Guidelines currently say (emphasis mine):
2.5.2 Apps should be self-contained in their bundles, and may not read or write data outside the designated container area, nor may they download, install, or execute code, including other apps. Apps designed to teach, develop, or test executable code may, in limited circumstances, download code provided that such code is not used for other purposes. Such apps must make the source code provided by the Application completely viewable and editable by the user.
There is also this tweet citing more details on the relaxed clauses.
Original
Does your interpreted download allow the user to write infinite loops or recursion?
Apple allow Javascript because they provide the interpreter and can kill your code. I have a feeling I've read that it's a 10 second limit but I couldn't find it on the site with a few minutes searching. (Yes, my self-imposed timeout for writing an answer kicked in.)
I think you're pretty safe if what you do is declarative and doesn't allow obvious looping in the interpreter.
I would also avoid the use of the word "interpreter" in any descriptions visible to Apple including public discussion. Maybe "parser" would be safer.
Codea have skated along the edge of these definitions with their Lua environment and cannot download code. They had to remove a feature for downloading new packages as ".codea" files.
Based on 3.3.2, they could reject an app for this. However, the scarier thing is that you could create the app, get it approved, have it be downloaded and used by many users, and then Apple could pull the app from the store.
Did you ever publish the app you described?
There's a major difference between the Guidelines and actual practice by the App Review team.
The current Guidelines state:
2.7 Apps that download code in any way or form will be rejected
2.8 Apps that install or launch other executable code will be rejected
So, the old ban on interpreted code is gone, and replaced by a ban on apps that could be considered to be IDEs or self-modifying.
However in practice there are a number of apps which do this, hence the difference between theoria and praxis.
You should take a look at what Apple has enabled in iOS7. It is now allowed to download and run JavaScript within your app.
I think what Apple means is your application should not depend in another module, compiled product or executable in order to work that will be downloaded from a website/server and that compiled add-on was not reviewed by Apple.
Basically when I asked something similar they told me something like: "If your application will download another executable compiled code that such a ftp downloader, key decryption tool or something of this kind that was not approved my Apple. You are available to download data or files (such as XML, HTML, PDF files, images) that does not represents an application.
The concept of the differences between 'code' and 'data' has been discussed on SO before.
Please see this answer: https://stackoverflow.com/a/642476/200696
From Apple's perspective, this ban prevents un-reviewed executable content from the app store. It would be trivial to create a program which is approved by Apple, and then downloads executable content that changes the pre-approved behavior.
All I can tell you is I've released products which use XML to script behavior within the app and Apple has always approved them.