Official: How Can We Serve You Better?

While agreeing with some parts on that, I don't think that "managing reports" is stealing their time. Putting reports into those 5 categories is presumably completely automatic. Then again, they changed their behaviour about bugs, so maybe they will come to an almost bug-free version

Edit: this was an answer to Tiles

 

So non-deterministic bugs are never fixed? Baffling.

Edit:
You should at least have some karma system then. So that when somebody who has 100% karma for his bug reports reports a bug it is looked at with priority. Even if it is a non-deterministic bug.

 

We have a new team, sustained engineering, who are fixing bugs and getting them into patch releases. The aim is to have a patch release every other week. This team, separate from the Devs, is finding bugs to fix. One of the ways we identify bugs is how often it's reported and voted upon on the issue tracker. Think of this as a small hit squad. It's not the only way Unity fix bugs. It's a new initiative to help get fixes out more quickly. You are welcome to quote me, of course, but in reality, 5 of us have spent around 3 hours going through a list and picking bugs for sustained engineering to fix. I'm participating in this thread from a support perspective. I know that only small numbers of Unity devs are joining in this bug-fixing discussion. Thomas can probably share up the number of bugs that the dev team have fixed for 4.5. The complaint from the community is not enough bugs that affect people making games get fixed. Sustained engineering is a solution for that. The community told us that they didn't have visibility over bugs, so we added the issue tracker. By all means beat us up for making new features ahead of fixing bugs. Thomas and me have personally stepped up to get that sorted. It's going to take time for the new team to make a difference, but that's what we're aiming to do through the rest of the year and beyond.

What's the issue tracker link for the "gravity for gamepad" problem? Not sure I know what that is.

 

In a situation like this you should fly an engineer to the client and see the bug first hand on their machine, taking your bag of code and tools with you. It's what other industries do. Your baked bean cans are jamming up your production line labelling machine we sold you? It doesn't do it on ours and we can't replicate it so either you are wrong or the problem doesn't really exist, either way we spent some time looking for it and have given up! Sorry Unity it's unacceptable. It doesn't even matter if it is a OSX problem or not. I've been in this situation with software and we find a workaround, not just let the customer stew.

 

I can understand why some bugs need to be reproducible to debug, but I've sent issues in the past which I've been unable to reproduce in a small project file (and cannot send my entire project file for practicality) but I've given information which links, in my mind, to a very clear cause. For example, I recently submitted a bug report where a FindWithTag search during runtime can somehow find GameObjects outside of the scene (it would return items from the project folder if they happened to be highlighted in the inspector when running the game.) It may not have been simple for me to find the exact circumstances to make it reproducible for you, but I believe that such a bug is clear cut for a developer to know where to look. I always believed these would at least be logged for further investigation... But you're saying that unless I've given you instructions 100% for how to reproduce it, it'll be ignored? Or are bugs taken on a case-by-case basis? Should I not bother reporting bugs in future unless I can give you specific reproducible instructions? I'm not trying to appear facetious, just curious to know what happens to bugs without repro instructions.

 

Looks like I'm late to the party. I'm not sure how this thread slipped threw my eyes for such a long time. Anyways, hope you guys are still eagerly listening to us complain.

I'm a full time student, that uses Unity mostly on the weekends trying to get a product ready for sale on the asset store. I'm the problem user of Unity. Let me explain:

Over the past years that I've used Unity seriously, you have milked a grand total of $18 dollars from me(Sixty dollar asset store purchase) over three years of use! I'm the type of user that you need to milk more cash from... But unfortunately you don't give me a reasonable option to do so. I simply can't afford to make a purchase of $750(student commercial version) for really anything and definitely not on something that hasn't had any monetary return yet. The $75 dollar a month subscription is more than my "emergency" budget for the whole month. Simply, not something I can afford at the moment with full-time school. I'm serious about using Unity and want to get more advanced features; however, I think you need to offer a ($200-$300) option of Unity. What this would contains can only be determined by UT. However, that's about the only option that would milk more money out of me, and I want to assume there are a lot other users in the same boat(One with wine corks filling holes ).

I would like to retouch upon the complaint of UT not forwarding enough information. I know you guys are working on fixing this, but it just can't be stressed enough and I'll be quick about it. I need to know whats going on behind those lovely doors.

At times it seems as if you guys jump onto the train of big new flashy features that take ages to implement and that's great. I like them too. However, you seem to focus too much on certain new features and let other areas of Unity slowly rot into oblivion. *cough* terrain */cough* *cough* does anyone even remember that tree maker?? */cough* *cough* shuriken */cough*

Unity Answer is the new S*** fest, We've had meta debates on there about ways to fix it, but nothing ever happens. Something needs done as it is only getting worse. See the links below for more information.
LINK
LINK

I don't want to seem like a singing soap box, repeating all the others; however I want to +1 on one more.

Low hanging fruit. This seems like something that should really have been address a long time ago IMHO. Put the newer guys on it and run a QA team threw it every month or so. It'll really teach the new guys the code base and get them some experience fixing/implementing small things before their debut on the big show projects. These are really small things that are easy to do, and make a lot of difference. Worth > work...


That's all the came to mind for now. I might make a follow up post if I think of anything else that might not have been covered yet; as I'll admit I only read the first few pages. I've really liked some of the changes lately in UT and hope they keep coming. You guys had me worried for a while when all the leads started jumping ship and everything went silent.

Best Regards!

 

I honestly can't tell if you're joking.

 

I'm also interested in this. For example, a couple of times we ran into the "all transitions out of Any State get corrupted" in Mecanim. Twice over 6 months or something. We don't have a repro, but we know it's there, and to us it's a high-risk bug precisely because we don't know what causes it.

Does that mean that Unity aren't looking into it?

Also, when a bug is already known, should we report it anyway (considering that we often can't attach a repro case)? I typically don't, because I understand that there's a QA team somewhere who has to sift through all the dupes, and I don't want to increase their workload without giving them some concrete data to help. But from the sounds of it, I should be reporting it because if I don't then it's less likely to get acted upon at your end?

I'm not fussed one way or the other, but I'd like to know so that I can behave in the most useful manner that's practical on my end. If hitting the report button, removing the project file and filling in a description of what just happened is useful then by all means I'll start doing it. By the same token, if that's a nuisance on your end then it's better if I continue as I already do.

Also, I have to echo Essential's concern about "No repro, no bug fix." Are you saying that you don't look at it if we can't repro in the report, or if you can't repro after investigation? If a bunch of people report something but can't repro it, do you guys spend time investigating or do you wait for someone to send you a repro case? Because honestly, I don't have time at work to fiddle around looking for repro steps and, to put it bluntly, isn't that a part of the QA team's job?

 

The Unity licence/pricing model offers a Free limited version or a Expensive Pro version plus expensive Add ons (iOS, Android etc).

This results in:
- Free users (majority of users) that are frustrated at not having all the bells and whistles.
- Pro users (minority of users) having to pay a premium to subsidise development.

The Unreal Engine licence/price model shares the load and gives every user full featured use at a bargain cost with no additional cost for mobile. Only if you publish something do you pay the 5% royalty. If you use it for Archviz there are no royalties. If you stop your subscription, you can still use the current build you have.

The outcome of this could result in UE gaining $20 per month subs from many Unity Free users choosing to jump ship. Money that Unity never saw from these users. Unity will still have its loyal base, but at $20 per month with no feature limitations and all users contributing and being treated equal, the UE route makes a very enticing offer.

I am a long term Pro + iOS/Android Pro user and have pre-ordered Unity 5.0 (that's a lot of years worth of UE subscriptions ). I never use to be a fan of subscriptions, however UE seem to have got the balance right for the user and I like the fact they do not stop you using the current build if your sub is cancelled. However I believe a 12 month subscription commitment should be required to stop abuse of this offer.

Unity makes a lot of revenue from the Asset Store. Such revenue comes from all users, why not get licence revenue from all users equally too.

 

Adding a minimum commitment period is not a good idea when the competition does not have one.

 

Exactly. It's about giving value. My UE subscription is still active, because each month I have received real value from Epic, with updates and Marketplace content. In the same time I've heard vague promises and excuses from Unity while being presumptuously asked to preorder.

 

It's what we had to do in my industry prior to going indie. The point I'm trying to poorly make is when unity can't find a bug and fix it for whatever reason, they are not supposed to just give up and put it in the "Too Hard" bin.

 

I liked the idea of a watermarked pro version for free users. At least I could make my game with pro features and kickstart the funds to buy all the licensing to get rid of the watermarks.

 

You could kickstart to get the funds to remove the watermark. Or you could spend a one-time fee of $20 for an engine that does not have them from the beginning and use the kickstart funds to pay for actual content.

 

You can serve me better by not censoring this forum for a start. Why was my poll on what engine plans people had closed?

 

It appears to have been closed on a technicality. Namely the moderator agreed with others that the poll did not acknowledge other engines. Which would be a problem if Unity and Unreal were not the only engines actually being discussed on this forum.

CryTek's knee-jerk reaction to Epic's new licensing scheme certainly did not get them any meaningful attention.

 

Not sure how UE are going to stop people from signing up for one month, pay their $20, get current build then sign up in 12 months for another $20 to get their upgrade . Hence why I stated the 12 month period.

 

Guess that would help encourage people to continue with the monthly subs.

 

Epic keeps saying they have no problem with this, although most people posting on the subject (both here and there) seem content keeping their subscription active for now. I guess it's all about users finding enough value month to month. I really think this (subscription) approach would benefit UT as well.

 

Why would they stop people from doing that? They have drawn enough attention to it that it is very clearly considered a feature in their license.

Even if they do unsubscribe though, the royalties still apply. Epic is very likely counting on those royalties for the majority of their income.

 

The UE licence model is a great deal for users and I totally agree with your last sentence, hence why I posted my original comments.

 

There is a very simple solution.Have the devs working on engine look at the bug reports, like UE4 does now. The engine developer doesn't need a repro case all the time, just a clear description is enough to trigger the Aha, I missed that. They can request more information if they are needed.Some of them are pretty obvious e.g I reported when multithreading is on, NGUI will render incorrectly on mobile device, not in editor. So this type of bug should not get out in first place, given the through unit testing / procedures you guys are touting to do. I don't know effective QA is going to be effective at filtering issues if they aren't the developers.

Unreal 3 was built by 10 engineers, Unity, you say you have 160 engineers, come on.

 

I don't get this too.. how is possible for Unreal to do so many things with less peoples. And the price justification for Unity is that they have 400 employers, but in same time they do less and less.

 

contractors are not employees.... we use to hide actual man hours with this. ez-peezy

 

I have the same question - is UT saying that we should be posting bug reports, even if we can already see the same issue in the Issue Tracker? I've deliberately not sent bug reports several times because I know that someone else has already put them into the system. Should we now be raising bug reports anyway?

For example, most Mac users of Unity Pro cannot see GPU stats in the Profiler. This issue is registered in Issue Tracker in at least three different bug reports. If everyone experiencing this creates a bug report, it will give UT a better idea of the impact of the issue, but doesn't it just flood the system with junk and make it even more difficult to manage and track?

Surely an issue where a paid-for Pro feature is completely non-functional should be prioritised above minor fixes in Unity Free?

 

I agree
Though in this case its just a matter of 'your hardware + OSX do not offer the information' because they lack the required feature to offer these informations. Unity taps into a specific feature of the gpu, exposed by the drivers, to offer this feature.

In such cases UT probably could save itself time on closing by linking to the related tech sheet from Software / HW creator and where possible link the bug raising sheet of said tech to the corresponding bug report sheet so you can raise it with them directly and add some weight to bugs raised by UT already.

 

I am not sure who is inflating or deflating the employers numbers, but I remember I've seen somewhere Shiva3D got under 20 employers, Eshentel and Leadwerks was one man engines. Even if they hired peoples, i doubt they got over 20 and they still do a very nice job and good products.

Secondly, at that level where Unity or Epic plays, do you think you can afford to do any tricks like that with employers? I doubt that.

 

The reason they're doing this, and the reason they can do this, is because they're going to be making their true money from royalties, not sub costs.

 

OMG vote on the bug to increase it's chances of being fixed! It's like some kind of comedy show.

 

Maybe we misunderstand each other here, but no, it is not nonsense. I talked about the bug reports, not the bugs. Bug reports != bugs, because some reports comes in multiple times. (Errr. Now you made me unsure. I really hope that it is not 16.000 open bugs. That would be really nasty. )

And for bug reports you do exactly that by your current system.You simply produce those many bug reports, wanted or not. Your current system is a system that feeds itself with work. A system that floods yourself with not needed reports. Workload as you call it below.

And you do double work here. First you let some people filter the reports. Then you finally let the developers have a look at. Means every bug is minimum touched by two people.

See, that is what i talk about. And this statement stands in contrary to the statement that the bug with the most reports gets fixed

I am sure most of the users doesn't even know that there is an issue tracker. The "send bug report" item in Unity itself does not lead to the tracker. It has also no hint towards the tracker. It sends a report, directly to Unity. It sends all your workload. It is wanted here that the user reports the incident immediately. Without the way to check if the bug already exists.

Yeah, let's make a quiz out of it instead fixing the bugs. Doesn't fix a single bug, and costs even more time and manpower to all participants. But is at least entertaining


The title of this thread is how can we serve you better. And one answer is: please solve the bug dilemma. It has added over the versions long enough now.

The picture you paint here when it comes to bugs in Unity is no nice one. Reminds me more at a hospital after the apocalypse, where the not so hard cases and the lost cases gets simply ignored. You admit that you cannot fix all bugs anymore. Which shows that there is a big problem somewhere.

Yes, of course, there are bugs that are unfixable for the one or another reason. Every developer knows that kind of rascals. And i don't even think that you do a bad job with bug fixing at all. That is not the question. But too much makes it through nowadays, and stays unfixed for the next five years or even more. That's the problem when you priorize bugs. We have 100 (i have no clue how close i really am here) low priority bugs from version 2 hanging around. Plus 100 low priority bugs from Version 3. Plus 100 low priority bugs from Version 4. And now Version 5 is coming.

One possible solution from my angle of view is to add at least a link to the issue tracker in the Unity bug report module. Which could reduce the number of sent reports quite a bit. I personally would also remove the q+a layer to check if a bug is a bug. The user has reported it, he sees this as a bug already. And a developer may have immediately a clue how to fix things where the q+a employee is even unsure what it is about.

I would also highly suggest to have a look at other companies and how they have solved their bug dilemma. For example, Blender has an open bug tracker, and Microsoft has the knowledge base. Both systems are transparent to some degree, without the need to vote for a bug to get it fixed. And at least the Report Bug link in Blender leads directly to the bug tracker. Never reported a bug to Microsoft though. So i dunno.

 

I know it's the internet and all that but are you this cynical in real life, too?
Call me naive or whatever but I don't think that mere cynical comments are helpful to the discussion. Do you have a better idea? Post it!

I'm sure that this is pretty much true. I know I had seen the bugtracker once or twice but pretty much immeadiately forgot about it a few hours later. Maybe there should be at least a link in the bug-reporter if it really is that important for regular users.

 

1) It very much depends on the people involved. E.g. read about John Carmack, the code wizard of ID Software (e.g. the book "Masters of Doom"). I assume what this single guy did in 3 months more than once could not have been matched by a team of 100 developers in a year. Because of the breakthroughs he made that nobody in the team would have made except if the team had a genius like Carmack too.

2) Teamwork in a 10 man team is much more efficient than in a 100 man team. Read "mythical man-month", assigning more people to a late project makes it even more late normally. I have been in software engineering for a while, and in most cases I have seen a small agile team beats the S*** out of any large team almost always. It's is not only communication overhead, but also diffusion of responsibility, accountability, missing global picture and much more.

3) This:

So unity is the OpenSSL of game engines? Kind of shocking, but matches point 1) and 2) perfectly. Monolithical software is hard to maintain, hard to change and often has lots of bugs. That is why nobody develops software like that.
And telling it is for performance is cheap. The AAA engines I know about the architecture have very nice modular design and still are very performing. Modular design has no large impact on performance if done right. If done wrong you can have performance problems, but that should never be an excuse to do a monolithic design, never.

 

That's something we do with enterprise support customers. We'll send them a diff or a experimental build, and if that works for them, will get the fix into the next release. Sustained engineering will just get these fixes into patches so the fix-release cycle is shortened.

 

I would really love to know the real actual internal state of Unity and whether it really needs fundamental changes or not. I am by no means a software developer (as stated numerous times) but I know that I really do love Unity and its approach to development and licensing for the users. Yes - I really do think that an affordable perpetual license in the price range median of 3D software and that a "no roaylties" model are very fair and for the customers. But I'm digressing again ...

I know that a lot of large software projects had core code rewrites that were necessary over the years. Which meant that there were incompatibilities but in the end the results were pretty much always worth it. I can't say that I took part in the actual development of a large-scale project with Unity. As an artist you often just to the fleshing out and "make beautiful" part and then be done with it except there are problems in the model. Still I always had that diffuse feeling that Unity might be problematic in actually large productions.

I would really love to know whether that's the case right now and how many changes are really necessary or even possible or yet even: how many will be introduced in Unity5 already? Is it that necessary like some people here make it sound?
Also - could an integration of the AssetServer for no extra cost in Pro maybe at least improve Unity's image as being a better teamplayer software?

 

No I'm not being cynical, but replying to the comment that if you want a certain bug fixed you have to vote for it to increase it's odds of being fixed. I never made that statement up, it came from a unity dev replying on here. Unity is a software developer and yes that comes with bugs. They don't put them in on purpose but it is very much their responsibility to fix them, and I mean all of them. I'm not asking them to fix my bugs, just the ones in their engine.

you shouldn't have to vote up a bug or rely on lots of people having the same bug before you fix it. You shouldn't cherry pick what bugs you fix. If I'm sounding like a stuck record it's because I can't believe a serious software developer can have a bug fixing policy like this.

 

In my opinion, upgrade model is also the culprit for the amount of outstanding bugs that Unity has. You'd think that Unity, with 400 people in it, would be in a better position to deal with the bugs than say, Epic. Problem is, most of those people are working on new and flashy features because that is what sells Unity upgrades. Bug fixes don't make good headlines or sit well with the cool announcement videos.

Few examples, there are very frequent bugs in the editor that makes it a pain to use. If scene becomes crowded, it randomly stops saving it. Gives out an error that it can't save, but doesn't tell the reason why. We have to either save the scene with a new name, or wait about 2 minutes, move something on scene, and try to save it, hoping that it works. Post processing effects randomly disable themselves, CC LUT image effect unloads the LUT texture when assemblies load. You have no idea why it is happening, other than the scene being somewhat crowded. Trying to import big packages with a big project folder completely effs up the Windows session, offloading everything into desktop, services like svchost get uncloseable windows (only a full restart solves it). We get these issues regularly with different configurations, but can't report them because we have no idea why they are happening, but they happen.

It is great that there is a new team for bugfixes, but when there are bugs from 2.x releases still present (and annoying), I don't really have that much confidence that it'll make a significant change. The fact that we have to vote for the bugs only makes matters worse. I'll believe it when I see it, is the phrase here I think.

 

I agree. If I understand correctly, a bug in the issue tracker is a bug that has been reproduced. It should get fixed no matter what. The idea that you can gauge the impact of a bug based on the number of votes is ludicrous. You have no idea how many users are affected by the bug but do not know that the bug is already known and that they should vote on it.
Sure, you can guess how many users a bug impacts and of course decide how serious the bug is. And sure there is triage - I too mark severe bugs for fixing before the smaller merely annoying ones.

But in the end - the goal SHOULD be to fix them all. The message you are sending here is that a single bug report from a single team will most likely not be acted upon unless it is a major bug. That is really worrying!

 

Oh, it's unfortunately not so uncommon to do cherry picking at bugs as you may think. The whole game industry works that way. Just fix what really needs a fix. Time is money. And deadlines have killed more than one needed bug fix in the past. There is rarely a commercial game that is really stable and bug free when it gets released. Most of them gets released in a late alpha or early beta stage.

For Unity the situation is a bit different though. It is not a game. It is what gets used to make a game. So it is better super stable and as bug free as possible. So that the bugs of the engine doesn't add to the game bugs too.

And the situation is different because Unity has reached a state where the not fixed bugs starts to hurt. Where functionality is affected. Where the workflow slows down because of the needed workarounds. I just hate it to restart Unity regularly to get my edit boxes in the properties tab editable again.

 

Well, guys, you have to understand, some bugs are not so easy to fix, and I give some right to Unity guys in that case.

But the amount of bugs and fact that them do not get fixed, is just because Unity is not doing any games at all. They do some tutorials from time to time and that's all. I am super sure if they was hit by their own bugs some of them was super fixed super fast. I hope you got the idea. They fix them only because users needs them fixed, not because they need them fixed for them.

 

No.

Just... no.

 

That is totally shocking, especially if you really do have 16000 bugs in backlog. Why are you even asking us how you can "serve us better", with such a giant elephant already in the room?

 

16.000 bug reports, not 16.000 bugs. I hope