Week One Adventures

[Jivaii]

3 minutes ago, beetitan said:

I think the steam community needs to see more activity like this. Right now the game is rating mixed with only 50 percent positive ratings which is heartbreaking. Many of us are excited about the future of Kerbal but hearing more from the devs may restore trust in the project and dev team. I would be lying if I said there has been a loss of trust out there after the launch of the game. Perhaps an address video to the community that empathizes the difficulties many are experiencing would be better received than just a bug list. 

Sadly, I doubt the majority of the users from Steam will even come over here to see this. So they'll continue to be semi-rabid and chomping at the bit.

[coyotesfrontier]

3 hours ago, Jivaii said:

Sadly, I doubt the majority of the users from Steam will even come over here to see this. So they'll continue to be semi-rabid and chomping at the bit.

Steam community forums are basically a 4chan board. It's unreasonable to expect any sort of civilized discussion out of them, ever.

[jebycheek]

Since the game release is certainly a disaster,this fix better be good or the damage to the reputation will be almost Irreversible...

I think the 1.0version will takes 3 to 5 years to complete I'm right? One Way or Another I wish you guys can finish your goals and maintain KSP2 to become eventually what we hoped.

Edited March 6, 2023 by jebycheek

[RocketBoy1641]

On 3/3/2023 at 4:57 PM, hidearimjosh said:

Do the words "Fixed:" mean that you've already solved the problem or that you will be working on them? the wording is a little confusing to me.

Depends on if you tell your boss "done", "fixed", "finished" without it being that way and expect to keep your job.  I trust that Nate saying "Fixed" means QA has verified through sufficient iterations.

[BlooSkies]

So at least part of the problem with the Steam reviews, I would imagine, is that the Steam storefront loudly proclaims that the last update you've provided was on the 24th. Obviously that isn't the case, provided people care enough to come here, but that relies on an expectation for the Steam user to first go to the forums and then see the pinned topic over there AND click the link in that topic that links here. Important updates about the state of the game like this should really be crossposted on the Steam announcement feed.

[Reusables]

On 3/5/2023 at 4:18 AM, ShadowZone said:

How do you automate a game like KSP2 that gives players immense freedom to build and fly ships?

You can catch most bugs using unit testing and property testing, then integration testing will finish the job.

Most game devs do not care, though, because the audience mostly do not care about bugs..

On 3/5/2023 at 4:18 AM, ShadowZone said:

But you can easily script tests for web products, since HTML, JavaScript etc are standardized and widely understood. A custom built UI on the other hand? I don't know how that should work...

I would say it is even easier. If you are constrained by web or GUI frameworks, you need to follow their way. I don’t think being widely understood as in JS way is helping, JS is more of unconstrained.

In custom implementation, you can arrange it to make it testable. E.g. while testing physical button press is not viable, you can insert hooks to simulate it.

So why is this not usually done? Again, because they did not care.

[Periple]

3 hours ago, jebycheek said:

I think the 1.0version will takes 3 to 5 years to complete I'm right?

Not that long. It'll take a few months to bugfix and optimize what we have now. The really messy things not well-supported in the engine are already implemented, they just need to be stabilized, optimized, and finished up. The systems on the roadmap are all plain-vanilla gamedev stuff, there shouldn't be any unknown unknowns there.

I think it'll hit 1.0 in 2024. There's even an outside chance it'll make it this year, depending on how complete the stuff on the roadmap is. 

52 minutes ago, Reusables said:

You can catch most bugs using unit testing and property testing, then integration testing will finish the job.

If only it was that easy. 

54 minutes ago, Reusables said:

So why is this not usually done? Again, because they did not care.

This is absolutely and categorically not true.

A part of the real reason is that people who actually know how to set up and maintain really good build pipelines and integration test suites are scarcer than you think. Most developers have other specialties. Programmers aren't build engineers, and a really good build engineer is hard to find. So a lot of the time things are sub-optimal because the build pipelines were set up by people who'd really rather be doing something else and don't have the requisite skills.

A better part of the real reason specifically regarding integration tests is that it's a mistake to set them up too early in the project's lifecycle. Early on you want to iterate fast and refactor a lot. Integration tests are pretty cumbersome to make and they tend to be fragile. That means that when the product is undergoing rapid changes, the integration tests will be continuously breaking, and you'll have to spend a lot of effort maintaining them. This has two pernicious consequences: 

1. It's time not spent iterating on the product.
2. It's a disincentive to refactor. "Okay this bit is a bit hairy and I'd like to clean it up, but if I do that, the integration tests will break and I'll have to spend a half an hour putting them back together so **** it, I'll just hard-code it." 

The upshot is that if you do this too early, you'll end up with a codebase that's WORSE than you would otherwise have -- your integration tests will have locked in a suboptimal architecture that's then too expensive to change.

Finally, integration tests (and unit tests for that matter!) can only control regression. A test won't catch a bug that you don't know about. It'll just catch when old things break unexpectedly.

I also think that KSP2 is especially handicapped for integration tests because most of the test tooling available for Unity makes a lot of assumptions that simply don't hold -- like, you're controlling a "character" who's in a "scene" doing stuff like "interacting with entities." To be able to even make proper integration tests for KSP2 you'd need to do a lot of work completely from scratch!

Integration tests should go in when the product is stable, and even then you should think carefully about just how comprehensive you want them to be. They're not a silver bullet and they're especially difficult for something like a game, which consists of a lot of systems that interact in all kinds of complex ways. 

[Vexillar]

Yep... what @Periplesays.

It's one thing to create and maintain test plans for business software (which, after all is where the professional IT skills and processes are defined... if not always used...) where the use-cases are fairly predictable and well defined in the requirements set.

It's quite another for an open-world game where there are almost infinite combinations of possible events and situations. That's precisely why EA is a thing. And the nearer to a release date, the more danger there is from the D'Oh! moments where a quick fix breaks something even bigger. Sadly, the real world constrains the developers on the decision as to when to release... '

[Reusables]

1 hour ago, Periple said:

If only it was that easy

It is hard, I would not object that. But it is what should have been done.

1 hour ago, Periple said:

A part of the real reason is that people who actually know how to set up and maintain really good build pipelines and integration test suites are scarcer than you think. Most developers have other specialties. Programmers aren't build engineers, and a really good build engineer is hard to find. So a lot of the time things are sub-optimal because the build pipelines were set up by people who'd really rather be doing something else and don't have the requisite skills.

I mean, there are many people who do not know to do that, but many software engineers could at least do basic unit tests. Gamedevs usually cannot. And there is why: game companies **do not bother** with thorough testing, so gamedevs would not bother with it as well - why create additional workload? This might not have happened if game players were considerate towards bugs, but yea that would be hard ask.

By the way, there is a reason I put integration test to the last. It is the weakest test and used just to see if components are working together reasonably. I do not know why you talk extensively about that part.

[Periple]

Just now, Reusables said:

many software engineers could at least do basic unit tests. Gamedevs usually cannot

This used to be true but isn't so much anymore, at least not in what you'd call "professionally managed" studios. I can't remember the last time I talked to a game programmer with a few years of experience at an established studio who wasn't familiar with the practice at least. 

2 minutes ago, Reusables said:

By the way, there is a reason I put integration test to the last. It is the weakest test and used just to see if components are working together reasonably. I do not know why you talk extensively about that part.

Because I don't see any reason to think KSP2's code doesn't have unit tests in place. The failures we're seeing mostly look like the kind of stuff that comes up in integration or just wouldn't be covered in unit tests because it didn't occur to the programmer to test for it. Unit tests don't catch bugs, they prevent regression -- they'll only catch whatever you thought of when you wrote them.

[Reusables]

16 minutes ago, Vexillar said:

It's quite another for an open-world game where there are almost infinite combinations of possible events and situations.

This is exactly what modular code and unit testing can tame. Property testing and fuzzy testing will be even more helpful. But gamedevs love to mess with global states, so.. that (global states) is not amenable to testing.

Also business applications often have wildly incomplete specifications. Feature request could flood in frequently as well. I do not think it would be as well-defined as you assume.

1 minute ago, Periple said:

I can't remember the last time I talked to a game programmer with a few years of experience at an established studio who wasn't familiar with the practice at least. 

Interesting that it used to be true. Might explain the trend of increasingly buggy games.

2 minutes ago, Periple said:

The failures we're seeing mostly look like the kind of stuff that comes up in integration or just wouldn't be covered in unit tests because it didn't occur to the programmer to test for it.

I do not see what you are talking about here, unit testing can be used a priori to catch bugs. Poorly written unit tests would be incapable of that, but if well-written, it will prevent wide variety of bugs.

If manually specifying test case is insufficient, you can employ automatic testing like property tests and fuzzy tests as well. Those are usually smart enough to find the problematic cases quite early.

[Periple]

@Reusables, where you work, are the early builds of new software you work on bug-free, well-performing, and entirely robust? Do you start out with fully mature DevOps pipelines, tooling, and test frameworks? 

The closest to this ideal that I've personally seen are companies that have spent years building really robust CICD pipelines for software that they've also been developing for years, with a tech stack that's also mature and stable that they know inside and out. 

Even a company that has all that in place will struggle if, for example, a key part of their tech stack becomes obsolete and isn't supported any more and they have to change it. Even though they know exactly what it takes.

It's really not as simple as you make it sound. 

Of course it could be I'm wrong and you're just really good at what you do! In that case, you have a really lucrative consulting career available for you, helping set this kind of thing up for game studios!

[Vexillar]

1 hour ago, Reusables said:

gamedevs love to mess with global states

That would explain a lot... oh dear... 

[Mushylog]

On 3/4/2023 at 6:17 PM, Temeriki said:

Ksp1 early access:  Hey guys heres our buggy game at a reasonable price, please break it and let us know how you broke it so we can try to fix it, we will most likely break it more trying to fix it and repeat over and over while we all have a blast so please be patient.  Since you helped us fund the game super early were gonna give you all future dlc for free cause you guys are awesome and we love our community.  Hey modders, heres some early release update candidates so you can try to get your mods current when a new update drops so everyone can have as much fun as possible.

Ksp2 early access:  Heres a AAA priced pile of poop, youll get updates when we think their ready, so just shut up and pay us to QA our releases whenever we get around to them cause despite us saying we wanted to recreate the amazingness that was ksp1 community engadgement we have no plans on actually doing so.  Hey modders, theres no sdk and no documentation, well get around to it whenever, deal with it.



Something something money changes people I guess.

 

That's a very narrow, pessimistic view of KSP2 and its team. Modding the game isn't impossible, it's just not supported yet. It is possible, though harder without the active help of the devs, right? I heard Scott Manley say that. Also there already are mods of KSP2 early access!

So... ok, the problem is the price of the early access game?

Nobody forces anybody to pay 50 bucks for the early access game, complaining about the price just makes me question your ability to restrain yourself and not submit to compulsions. It's hard, right now Red Dead Online is down 50% of its price and I want to buy it really bad but I won't because I'm planning on moving out. That is restraint; it sucks but that's life, man!

You are in control of how you choose to spend your money - if the team behind a video game does not seem trustworthy, do not pay for their product, especially if it's early access; or get a refund! You'll leave a negative review and that will be it for your part - that's fine - the company will get bad reputation and lose a lot of money if they really are eViL and money hungry.

I'm having a lot of fun despite the bugs, and I put my faith in their hands to deliver what they promise in the roadmap - even though we have no idea how long it's going to take. These 50 bucks I spent, it's on me and nobody else, because I know what the risks are: I'm still going to spend hundreds of hours on this game, just like I did for KSP1 (1150+ hours for 30 EUROS or so, maybe 50 with the DLC I bought years later, it's a great investment!), so if you look at it in terms of [how much you spent] times [how much time you spent having fun]... I mean it is a win for me! And I already spent 51 hours in KSP2, so it's less than €1 per hour, I still have a lot to do in the game! In its current state! And there's going to be more to come in the future (IF the developers deliver, yes)!!

[caipi]

On 3/3/2023 at 8:40 PM, Nate Simpson said:

• Fixed: Animation stutters when EVA Kerbal is running

 

I can't even go on EVA. I first discovered it when I landed with the MK1-3 Pod on the Mün. I then tried it in space, still nothing. Then I tried almost every pod on the VAB, including the rover EAS seat. Nope. I read that a few other people had the same issue. Not sure what causes this though.

[Moon Goddess]

10 hours ago, beetitan said:

I think the steam community needs to see more activity like this. Right now the game is rating mixed with only 50 percent positive ratings which is heartbreaking. Many of us are excited about the future of Kerbal but hearing more from the devs may restore trust in the project and dev team. I would be lying if I said there has been a loss of trust out there after the launch of the game. Perhaps an address video to the community that empathizes the difficulties many are experiencing would be better received than just a bug list. 

 Seeing stuff like this really shouldn't change the Steam reviews.   Steam makes it very clear that your reviews about early access games should be about the game NOW not promises for the future.    Trust should have no effect on steam reviews if people are unhappy with the current state of the game they should give it a negative review on Steam not give it a positive cuz it might get better.

[qwery123]

Is the crash during loading of games (where it freezes at "Pumping Sim Once" due to textures not loading properly) fixed?

[mcwaffles2003]

On 3/3/2023 at 6:56 PM, linecrafter said:

Hilarious bug, was fun for some time but its time to let go  

I got really excited when it finally happened to me. 

[MarcAbaddon]

My latest bug with EVA was that on exit my poor Kerbal was on a very different orbit than the craft my Kerbal exited by several 100 kilometers. The two orbits intersected in a point about 90 degrees away but with a relative speed of 250 m/s.

Edited March 6, 2023 by MarcAbaddon

[Carraux]

On 3/3/2023 at 11:28 PM, Domonian said:

Any fix for decouplers getting stuck on stuff? I can't get a rover to the Mun because I'm not very good at building rockets and my "solutions" all end up having a decoupler get stuck on an engine bell or something and causing issues. 

I have this problem too when decoupler and engine are in the same stage. Try to put them into separate stages.

Or decouple manually (right click)

[wpetula]

It would be cool if these weekly adventures doubled as loading bars for the next update! For every bug fixed or feature implemented on the to-do list, the devs could add a  emoji next to that item. We could then watch development take place in real time.

Edited March 6, 2023 by wpetula

[pandaman]

10 minutes ago, wpetula said:

It would be cool if these weekly adventures doubled as loading bars for the next update! For every bug fixed or feature implemented on the to-do list, the devs could add a  emoji next to that item. We could then watch development take place in real time.

But all the features Nate listed are already fixed.

[MechBFP]

4 hours ago, Moon Goddess said:

 Seeing stuff like this really shouldn't change the Steam reviews.   Steam makes it very clear that your reviews about early access games should be about the game NOW not promises for the future.    Trust should have no effect on steam reviews if people are unhappy with the current state of the game they should give it a negative review on Steam not give it a positive cuz it might get better.

Steam should technically be archiving reviews then during major revisions as EA progresses if that is actually how they view it. But of course they don’t, so clearly their actions don’t meet their words. 

[W2D2]

2 hours ago, Carraux said:

I have this problem too when decoupler and engine are in the same stage. Try to put them into separate stages.

Or decouple manually (right click)

I think the bug is highly likely the priority order of part actions.

 Burning engine, then decoupling -> explosions

 Decoupling, then burning engine -> happy staging

[RockyTV]

On 3/5/2023 at 6:09 PM, Bej Kerman said:

You'll stop asking questions intended to frame KSP 2 as a complete product, then we can discuss.

Well, it technically is if we look at its price, $49.99 SRP makes it qualify as a complete product, I guess.

17 hours ago, beetitan said:

I think the steam community needs to see more activity like this. Right now the game is rating mixed with only 50 percent positive ratings which is heartbreaking. Many of us are excited about the future of Kerbal but hearing more from the devs may restore trust in the project and dev team. I would be lying if I said there has been a loss of trust out there after the launch of the game. Perhaps an address video to the community that empathizes the difficulties many are experiencing would be better received than just a bug list. 

If by hearing more you mean frequent patches then yes, otherwise just coming to the forums once in a week and saying "hey! we added more fixes to our list of fixed issues for the next patch which will be released sometime soon!" doesn't really do anything to get people to restore their trust in the devs.

Also, for those that argued with me and pretty much said it's a dumb thing to demand a more-or-less working, decent product in EA:

I realized this type of releasing half-baked products is the industry standard now and not just for games. I was talking to a few coworkers of mine today and they told me that my company does the exact same thing I'm complaining about for IG and KSP2. They promise a ton of things and then they have to deliver. The problem is, it's not me, you or casual, random players. Other huge companies depend on the products the company I work for is promising/developing, so things get more complicated.

So, yeah, the industry itself has a problem where there are lots of promise and lack of delivery. We're talking about a EA product here so we can take it a bit easier, but there were SEVERAL AAA games released recently that were literally half-baked games because they promised too much to create hype, get people to buy the game, then they could slowly polish the game whilst also trying to maintain the hype up so people won't lose interest in the game.

Which doesn't seem to be happening with KSP2, honestly.