when is official nested prefabs coming out?

How come they never fix one of the most important and basic feature of Unity? I just want to make a very small change to some button, and I need to edit 1000s of buttons already placed because prefab links are broken!

 

If they didn't intend to fix it, it wouldn't be on the roadmap. They are aware of the issue, they had something like a working version, but the alpha or beta testers gave the feedback that it is not good enough or that it is not what they are looking for.
Meanwhile it seems that they know what is needed, also thanks to the official thread that was started:
http://forum.unity3d.com/threads/official-prefab-improvements.253427/

 

How many years does it take to fix a simple bug? We don't need a new uber prefab system, we just want nested prefabs to not break connections. And then they can think about the future...

 

Even though it is annoying from a user point of view and it is clearly a usability deficiency, it is not a bug, because it was meant to be like that. And you can be sure, if it was simple to resolve that issue, it would have been done long ago!

 

Yes, I understand that. At the same time, I can't help myself thinking that they are taking simple matter too complicated. And hoping some day soon some super hero in unity team will put an end to this suffering.

 

summer 2016 has just ended. no signs of nested prefabs so far. not even a word from Unity for a year. It would be nice to hear some clarification on the state of one of the most wanted features. Even "development is stopped for an undefined period for _that_ reason" is better than no feedback at all.

 

https://unity3d.com/unity/roadmap

 

"In-progress, timelines long or uncertain"

That sounds like definitely not 2016 or 2017. Well, better than to rush it and release it unstable I guess.

 

Isn't the nested prefabs feature just main prefabs that make instances of other prefabs on start? A script can easily do that.

I can see the advantages, to visualize the nested prefab on editor, and e.g. update a nested rock, and watch all rocks update in the main prefab ... where right now updating a rock in a prefab would update the main prefab, but does nothing on the other rocks in that prefab.

But I sense it can also have disadvantages, troublesome to manage. Possible accidents when applying changes on prefabs or nested prefabs. You'd certainly have to pay a lot more attention as to what prefab you're really doing changes on. And if you're working in teams? Even worse.

 

Well according to Unity's own employee, we are all colossally stupid for holding them accountable to their own words, and for wanting what we need (rather than what they would rather do instead).

Unbelievable disrespect toward their own userbase. I cannot believe I read this.

Just shows how little they think of us colossally stupid lowly peons.

 

its not better to rush it, its better to do it right. The rush it and release it unstable mentality is the reason why half of the engine barely worked in 5.1 and 5.2

 

Based on his Unite talk about scriptable objects, he seems to be one of the most chilled out persons I've ever seen. His comment seems understandable and I'm not the least bit offended by it.
Considering how harsh the community sometimes reacts to things, I think Unity employees by and large are incredibly patient and reasonable.

That's similar to how I'm currently working around this. If I want a prefab in a prefab, I put a placeholder there with a script that instantiates the other prefab. Works fine so far.

 

Sure, he is probably a fine person; someone who you'd be glad to have as a neighbor.

That still doesn't change the disrespectful perspective many have about their own audiences.

I fully understand. And I actually would be likely to agree with him; colossal stupidity is inherent in the human species.

But come on... a company isn't suppose to say that to their own users... that's in itself colossal stupidity.

Imagine if Taco Bell was caught on video denigrating their consumers, "Some of these idiots eat this S*** 3 times a week!"

 

Personally I'll always prefer a "rude" but honest answer over meaninglessly "polite" corporate bullshit that is just one textbook phrase after another. I think being "real" makes employees much more relatable and human, even if you are having a disagreement with them. I'm more likely to get angry at a "Thank you for taking time to give your feedback on _X_. We are always eager to listen to our customers and constantly trying to improve our service. I regret to inform you that we currently have no resources to address the issue, because it doesn't affect a significant percentage of the userbase. Thank you for understanding and have a wonderful day.". I'd rather get an honest "Sorry, but we can't take care of every special-snowflake usecase under the sun. You need to solve this yourself.". I respect no-bullshit bluntness, especially on big-company-level, because it is so rare there. To me talking bluntly doesn't imply having any contempt. I would not appreciate contempt, on that we can agree.

This may or may not be a cultural difference. We Germans aren't exactly known for being very polite, so I might be used to a different "baseline" in this regard.

 

Superpig wasn't a Unity Employee when he said this.

 

I dont think they will ever add nested prefabs

 

well the problem is more difficult than many think it is, mostly about how you serialize data on the prefabs.
Say you got 1 prefab for a window in your UI, then prefabs for each element, if you make changes to each button, where does that data get written to? The parent window, or each buttons prefab. Their is going to have to be some user driven way to define and which nesting level properties get saved.

 

I tend to agree with this. It's an extremely "nice-to-have", but seeing how many good games have shipped using Unity just says that it's not an absolute necessity for completing a game. I'm also of the mindset that if it's taking this long to complete, it's an extremely difficult problem, prone to breaking whatever work you've already done. At this point, if it ever gets released, I'd at least wait for a large adoption rate before even considering using it. And for any developer who believes their project is "held up" because of this feature, I'm afraid you may not ever finish your game. The challenges of game dev are sometimes, and maybe even often, finding sound workarounds to march forward to the finish.

 

This.
With prefabs and nesting them, and affecting and being tied to potentially many, many assets in project, there is a risk of being able to wreck a whole lot of stuff at once. We/I built a sort of nested prefab (more like style sheets) solution for our UI. (Granted, this is overlaying existing Unity/editor and not core/low level implementation) Updating prefabs, meant having to instantiate, check and sort dependencies and references and then applying them, then pushing the changes back. When something went wrong, it went wrong silently and in a lot of places. It became a magic button to destroy hours of work. Also tracking what should and shouldn't be affected by an update, got convoluted fast. It works great now, but it is very specific to UI, and more of a styles/template solution than a multipurpose nested prefab solution.

If they do add an internal solution, I would much rather it take longer and be more robust and tested. Or maybe require that a VCS be in place for the project before it can be used.

 

I'm pretty sure most of us agree that releasing it as a broken feature because it was rushed is bad. But we would really be glad to hear a word about what's the current state of the feature. Is it in development? Is it in the backlog waiting for a product manager to remember this is a feature people want? If this is in development, why is it taking so long? Is there a specific problem that's causing the development of this feature to take forever?

 

Nope.

Nested prefab means that you can have prefab B and then you can have prefab A which includes prefab B as a child. You can modify prefab B, and changes will propagate to the object attached to prefab A, at the same time A will continue function as a single object. Currently this kind of thing is not possible. Practical application for nested prefabs is things like dungeon kits. So you can have a "corridor section" prefab, then make a nested "corridor with a torch and some clutter" prefab based on it, etc.

Eitherway, whoever wanted nested prefabs, asset store has package called "PrefabEvolution", that implements them. It is a bit quirky, but works most of the time.

 

Maybe they just need to strip out the unneeded functionality of a C++ compiler or some similar compiler and use that to create nested prefabs? Or would having a mini-compiler at runtime to evaluate prefabs be too expensive using memory and computations?

 

I don't see how this is supposed to be related to nested prefabs. The point of nested prefab is that you make a prefab onto object, then put it (as a part) into another prefab, then modify original object, and changes propagate into every prefab where this object was used. Currently this is not possible, Object "A" attached to Object "B" becomes a part of OBject B when you turn object B into a prefab. It is not about memory/computations.

Trying to roll custom C++ compiler is suicide, by the way.

 

Well from your description of a nested prefab the inheritance and polymorphism logic of a C++ compiler adopted for nested prefabs might do the trick. Of course I have no informed depth of how prefabs actually work in the Unity Engine.

Uh, creating a custom C++ compiler is dangerous to who? Or how is creating a compiler dangerous to the creator of the compiler? It's just automating that much more danger out of reach.

As far as the Unity attempts to create nested prefabs I'm guessing it must be more difficult than the C++ implementation of inheritance and polymorphism for Unity staff as they managed to create IL2CPP for multiple platforms but not a nested prefab solution. They've got have more than a couple of compiler experts working for them.

 

The scope of C++ is too huge to create a custom compiler. When C++11 update got released it took several years to implement it all, even for someone like microsoft. That's why it is a bad idea. It might make sense to create some preprocessor, like Meta Object Compiler usd in Qt 5. But custom compiler is out of the question. It is "suicide", because the project is too huge, bigger than making a custom game engine from scratch.

I think that the idea of trying to stick polymorphism into prefabs (which is what used in UE4) is bad, mostly because it is counter-intuitive.

Nested prefabs in UNreal 4 work like s***. When you have an object with blueprinted objects attached to it, attached objects do not show up in blueprint editor, well, because they don't exist at this point. It is inferior to prefab evolution.

 

LOL, well you just solved the mystery why Unity is not finished the nested prefab solution yet. They are co-opting the C++ compiler logic for that and it will in fact take 11 years. What have they, 5 or 6 more years left?

Joking, I was just wondering, I'll leave the informed and expert Unity staff to do the prefab work. I've written a compiler once as a term semester assignment and no, it wasn't easy, and that was even was a compiler for a dumbed down language.

 

Umm.... have you even read what I posted before? C++ compiler logic has nothing to do with nested prefabs.

There's working nested prefab solution on asset store. No C++ involved.
https://www.assetstore.unity3d.com/en/#!/content/17557

Seriously...

 

Yes I did but you didn't read what i wrote. Who says there only one equation with a solution set of 0 or that I must agree with your preferred method.

Seriously...

 

(-_-)

Well, I can't help you with that. You're trying to push discussion towards C++/polymorphism for some reason, I don't know why and I don't care why.

Try UE4. Their approach is that "a prefab is a class". In practice it sucks compared to unity prefabs and compared to prefabevolution prefabs too. Unity's approach is that prefab is not a class, but data. Which in practice works much better.

I'm done here.

 

Ouch! Now that's some serious loss of faith in Unity as a company. (Justifiably, of course. I'm just pointing out that this is a real thing. There is a serious disconnect between the Community & the Company.)

Who is the Unity community manager / rep? Or is there one? I'm on the forum "enough" to notice one, but I don't see anyone but mods. Mods who I'm sure are unpaid (not part of the company).

Unity should hire a community manager. And if one already exists, then perhaps get them to do their job to help restore the faith of the community (not just forum, but all users) to the company?

 

We have several Community Managers! They are @SaraCecilia, @Buhlaine @aliceingameland and @LeonhardP!

 

Staph! This is a place of lurve!

 

Off topic posts have been removed. The topic is nested prefabs. Keep on topic.

 

How reliable is it? I worked on a project using it a while ago and found that prefabs using it could easily get corrupted.

 

On the updated roadmap Nested Prefabs is one of the few things that hasn't been touched in the past week. Would be really nice to hear something official about it, even if it was "we need to update the serialization system first, so it's not going to happen in 5.6".

 

Nested prefabs would be super useful for UIs, but what I do now is store all my various HUDs in different scenes (one for the standard HUD, one for the time attack mode, etc) and load them all in additively, which works pretty well for me, since with the new SceneManager you can see which scene each object belongs to.

 

As far as I know occasionally it hiccups and creates component duplicates. That has been a year ago. Not sure about its current state.

 

I hadn't noticed the updated roadmap, that page looks like complete crap now.. well done unity, oversized fonts, overbloated padding, and a ton of vertical scrolling, the previous 3column layout out for a 1column of vertical scrolling tablet crap. Oh well just another page of Unity website turned rubbish that I won't bother with.

And it seems they still don't bother putting a time stamp on it anywhere to show when it was last updated or what had changed in roadmap since the last update.. that would be useful info to have when checking it every so often.

 

Why not just have linked prefabs.

If you use a prefab when you make another prefab it is linked to the original.

If you change the original prefab Unity alerts you to the other prefabs that it has links to similar to the dialogue you get when you install assets. So you can selectively update linked prefab elements.

View the prefabs that have links to a prefab and drop links or re-root a subset of prefabs by cloning and changing an existing prefab creating a new tree of links.

Something as simple as this could provide what most people need for nested prefabs.

 

Since it's been 2 years since I last asked about this, and seeing as how there is no progress here. I have just one very simple serious question for Unity -

Do you see prefabs and scenes as the most core building block for Unity games?

Thanks.

 

I haven't had a go at this yet as I haven't started a fresh project in a while, but I suspect that the newly available multi-scene support will actually do a bunch of stuff that I was previously using Prefabs for.

For instance, instead of having my game's HUD as a prefab that each scene uses, it can be a scene of its own. Then there's just a script in scenes that need it which additively loads the HUD scene. That HUD scene can internally also load sub-scenes of its own, and any of the above can also use prefabs. Of course any actual prefabs can still only be one layer deep, but for me this actually addresses my number one case of wanting nexted prefabs - which is having pre-created GUI components that can contain other re-used, pre-created GUI components.

That aside, I too wonder why it's taken this long to get support for nested prefabs of some kind.

 

I actually recently did this, got so fking tired of editing the ugui stuff in the scene view with rest of the main scene objects all in the background/foreground getting in the way of the canvas.. and disabling/enabling that stuff was just stupid micromanagement.

Ofc having it as a seperate scene hasn't made things all that much better along witht he extra time spent making it work this way. Now I can go and open the ugui scene in its own scene file to work on.. but yknow I still want to be in the main scene ideally where everything else is. I just wanted some better fking options to disable (ie hide/unhide) all the other objects except the ones in other scene files while in the multiscene hierarchy or a ugui canvas..these could be simply fking rmb context menu options for multiescene in heirarhy for all care. So I'm back to square one, I still work on the ugui stuff while in the main scene and I still have other scene objects in the way sometimes. Only now I do have the option to just go edit the ugui scene file directly without the rest of the S*** in the way.

Yknow what would be like a genious fking idea.. a dedicated scene like tab for doing ugui stuff in 2d.. complete with decent grid snapping for components and all the rest of good stuff. As this method of seperate scene for ugui has only added more code to main scene to make sure its loading the ugui scene etc.

Anyway multiscene has have got nothing to do with replacing a core nested prefab workflow.. Where I see that being useful just in ugui alone is styling ugui components, and having those changes propagate without the headache of either writing code just to get every button hover the same fking color etc just because I wanted to change one of them I have to manually change the rest. Seeing as ugui has no global theming type setup like imgui has with its messy guistyles thing.

But I guess this stuff needs a few years for them to think about its implementation and umm and arr over it.... even though it somewhat exists as an asset, I simply won't touch it because prefabs are so damn core to Unity, why tf would I want to buy an asset for it when it should be native and in every build of Unity going forwards.

 

You don't find layers useful for this?

I completely agree that it's all micromanagement that our tools would ideally handle for us. I wonder how difficult it would be to make a view derived from the Scene view, with automatic filters to only show stuff in your UI layer(s). Again, I agree it'd be better if it's native, but one step at a time...

 

I don't understand this at all, and consider it quite backwards logic.

If it should be native to Unity because it's such a big deal, and you want to have it, then you'd want to buy the asset because the only other option is to wait for another 5+ years for Unity to decide not to add it themselves. So why wouldn't all of this reasoning be THE REASON you WANT to buy the asset?

TLDR: Why? It's as simple as it sounds: Because you want to use the feature and can't wait an eternity for Unity.

 

"You don't find layers useful for this?"
Not really.. do you know of modifier key that you can hold while clicking on a layer.. and all the layers BUT the one you clicked on gets hidden instead option? No I don't... therefore not really much of a solution as you're still left with unhiding/hiding everything the long way. Ofc not really much a hard task for someone at unity to fix in a day with a minor feature improvement to supporting doing exactly that..yknow for somoene who knows the editor code and its engine.

Or even having an context menu option in the hierarchy menu option to hide all gameobjects except any of the objects attached to the selected root canvas, that too would help.. none of these minor little things that I'm sure someone at unity probably wouldn't take that long to add support for. But are never done, probably bloat the editor up or something, better too add some gooble fakeborg build support yknow whatever.

"If it should be native to Unity because it's such a big deal, and you want to have it, then you'd want to buy the asset because the only other option is to wait for another 5+ years for Unity to decide not to add it themselves. So why wouldn't all of this reasoning be THE REASON you WANT to buy the asset?"

Have you not noticed how messed up prefabs can be, especially in version upgrades or cleaning up projects. Even exported unitypackages can be a whole nuther area of potential breakdown for prefabs. Just those things alone are enough reason put me off buying an asset that revolves around prefabs from the third party side, its not being improved from inside out or with native engine code adjusted to make it work better. Improving these things by making the kind of core changes that need to be done going forwards properly.. perhaps its more complicated tackling it from that side of things.. ofc its not the only area that should be improved from the core side, not some c# layer ontop.. such as the numerous hierarchy enhancing assets that I would find useful, except they all suffer from performance crapness pretty quickly when the number of gameobjects in the hierarchy piles and cause responsive ness to lag because c#, when those would be areas for Unity to improve the editor functionality properly.

Imo Unity seemed to have dodged it and put it off for the longest time, I believe it got mentioned at one of the Unite's that they've finally made some headway on it (nested prefabs that is).. who knows.. doesn't seem like it, but I'll be the first to actually bother installing a preview build for that to see how is.

Fyi before you come back at how unthinking my reasoning is.. consider these are the nested prefab type assets available...

https://www.assetstore.unity3d.com/en/#!/content/17557
https://www.assetstore.unity3d.com/en/#!/content/72007

Now I ain't knocking them, might be useful for someone to try out and risk a few months on it and maybe it works out. Overall they have clearly put more of an effort to try improve this area than the Unity has so far.. However I've asked myself before when looking at these assets almost last year... do I seriously want to jump into using a third party asset that is inevitability going to be so core to how I build a project around.. Do I really want that kind of third party asset given all things I've seen go awry with prefabs and abandoned assets, unity breaks certain third party assets that hook in lesser used api stuff all the time in updates, which they don't exactly jump in to rescue if the authors not fixing them anytime soon. And this isn't even taking into account how I feel about either of those assets implementation of nested prefabs. So no, I will just work around it and that is what I've done. When it comes to editor assets I like to know I can throw them out at any point and not have my project crippled by doing so, you can get that with Unity already sometimes

So I think building something around nested prefabs would likely cause a lot more problems should anything go wrong, so having it added natively by Unity is somewhat of a guarantee they maintain it (at least, God forbid they continue improving something *cough* ugui related stuff) along with updates etc.. Maybe different for building a game around an editor & game asset. Nested prefabs aren't like terrain assets etc you know what I mean.

Besides I'm the kinda guy who holds the weight of 1 negative reviews above 10 positive ones anyday these days. Off topic ... have you seen printer reviews lately.. I think anyone giving a modern printer a 10/10 rating should be shot -with ink

"Oh it printed beautiful picture the first time I plugged it into winblows 10, after communicating over my network with hq. All teh brand new chipped ink cartridges hadn't even said they'd run out of ink yet after 1 print . 10/10 would get raped by printer industry again." - some noob.

 

Here's the thing with Prefabs and editor extensions. I've seen nested prefab work arounds. I built a scene based prefab-like system to handle the major pain points with big groups of objects. The thing is that scenes and prefabs are so low level in Unity that you can't implement the nested prefab feature even close to what they show in the Unite talk. If they actually let me implement this to a workable state then I would have no complaints.

 

Pretty sure layers are not the right tool for the job, because there are only 32 of them.

HOwever it is quite easy to write a context menu which will toggle object's "active" flag. This way, you'll be able to toggle then via hiearachy window without jumping back and forth between hierarchy and inspector.