For a long time, an overwhelmingly popular view among game developers and publishers has been that offering Linux builds would involve too much work, because they had either tried it briefly or heard from other devs who had tried it, and found that their problem reports massively increased. Their conclusion was often that Linux causes too many bugs to be supportable. As a gamer, I was of course disappointed every time I read this.
More importantly, as a developer, I couldn’t help noticing ways in which this reasoning seemed flawed. I always felt that it was either poorly informed or not completely honest.
So, when this refreshingly different perspective from a game developer surfaced on social media, it warmed my heart. I thought the rest of you might find it interesting.
That was a few years ago. I imagine the influx of gamers using Linux since then (since it’s easier now) might mean a smaller portion of our group has the technical skills described in that post, but I think it still applies. I hope it also gives us something to aspire to when interacting with the people who make the games we play.
Whenever I use Proton/Wine and I run into a bug, I have now resorted to claiming I am using Windows.
A lot of extra work for just 5.8% of extra units, right?
Wrong. Bugs exist whenever you know about them, or not.
Do you know how many of these 400 bug reports were actually platform-specific? 3. Literally only 3 things were problems that came out just on Linux. The rest of them were affecting everyone - the thing is, the Linux community is exceptionally well trained in reporting bugs. That is just the open-source way. This 5.8% of players found 38% of all the bugs that affected everyone. Just like having your own 700-person strong QA team. That was not 38% extra work for me, that was just free QA!
Wrong. Bugs exist whenever you know about them, or not.
In a perfect world? I fully agree.
In a world where I am limited by hours in a day and how many engineers I have on staff? A bug that nobody knows about is not a bug. That is obviously playing with fire because there is a big difference between “if you unequip and reequip a flashlight over and over it will make you invisible” and “if you mash these four numbers at once then the ATM will wipe its cameras and start spewing benjies”. But for the purposes of adding new features/maintaining developer sanity? Yeah…
I dunno. This comes up a lot. I am going to ignore the circle jerk of “linux users are smarter and make better bug reports and also have bigger dicks” because… either people are slipping their hands down their pants or they know that is nonsense.
But I think it DOES ignore the reality that adding actual support for a new platform does drastically increase the testing and build/deployment overheads which are usually the realest of costs anyway. And… truth be told, I think the standard of “Don’t break Proton support. Fix things as they come up” really is the best of both worlds.
In a world where I am limited by hours in a day and how many engineers I have on staff? A bug that nobody knows about is not a bug.
But people know about it, so much that they are reporting it, you don’t know about it, but everyone else does.
That is obviously playing with fire
Exactly, without the report you wouldn’t know what type of bug it is that affects people.
“linux users are smarter and make better bug reports and also have bigger dicks”
No one claimed any such thing, have you actually read the article? He claims Linux users are just more used to making bug reports, so they keep doing that on games the same way they would on any other piece of software. It’s about the mentality than intelligence, most people experience a bug, curse/laugh and carry on. Let me ask you, have you ever reported a bug in a game you played? I’m sure you’ve experienced many, but have you ever actually reported one?
But I think it DOES ignore the reality that adding actual support for a new platform does drastically increase the testing and build/deployment overheads which are usually the realest of costs anyway.
That is true, which is why the majority of games released for Linux are indie, since only indie developers have the necessary funds to carry such big overhead… But being serious, yes, there’s some overhead in setting a Linux build, but it’s usually one of the easiest to make, most games are already doing Windows/Xbox/Playstation/Switch adding an extra pipeline there should be much simpler than you’d expect.
Fix things as they come up" really is the best of both worlds.
How would you know things came up without bug reports?
No, this is absolutely wrong. Most people don’t report bugs, they just stop using the product and move on.
Yes, time is limited, so you won’t be able to fix every bug, so you need to prioritize. How can you prioritize something nobody reports?
FOSS has a culture of reporting everything, but not necessarily fixing everything. Look at release notes and you’ll see bugs getting fixed that were reported 20 years ago. Does that mean they fixed all the recent bugs and they’re finally getting to the harder ones? No, some bugs reported today won’t get fixed for another 20 years. All it means is it finally got prioritized or someone really wanted to fix it.
Linux gamers don’t expect every bug report to result in a patch right away, they just want to add it to the pile. Maybe a dev sees it and fixes it along with higher priority fixes, idk, but it can’t happen unless it’s reported.
adding actual support for a new platform does drastically increase the testing and build/deployment overheads
Sure. Most Linux users don’t ask for a native build, they just want it to work on their platform. But even Proton support requires some level of testing.
Whether you go with Proton or a native build, the takeaway here should be that you’ll get a lot of free QA and detailed bug reports, so even if you don’t break even on Linux sales, you should come out ahead with that added QA.
You noticed that my dick is bigger than yours?
I am going to ignore the circle jerk of “linux users are smarter and make better bug reports and also have bigger dicks” because… either people are slipping their hands down their pants or they know that is nonsense.
Fuck off asshole
Determining that a bug isn’t a priority is fine and doesn’t require burying heads in the sand.
Its not burying your head in the sand.
If it is isolated or people just don’t care? Then… it kind of doesn’t actually matter. You scan the forums and optimally have community managers/PR people to do the same to keep an eye out for “This was weird?” style comments but you mostly focus on the stuff that naturally rises to the top or that you identify as an issue.
The more bug reports you have? That is engineer time spent assessing what is and isn’t a priority. And the sad reality is that it is a LOT easier to say “we have our five thousandth number one priority” rather than to say something doesn’t matter. Because if stuff does go down? Suddenly you are on record saying the most important thing ever (whether it is a critical vulnerability or just something people fixate on) didn’t matter and you can bet everyone will throw you under the bus.
As a developer? I want every bug reported to me because I genuinely do want to make the best product I can. That said… if you don’t care enough about reporting a bug or it isn’t reproducible enough to matter… I am not going to complain about getting some extra time to work on things that actually interest me. Which may very well be trying to reproduce that “weird behavior” myself because it sounds like it could be bad.
And… as someone managing a project/team of developers? I can watch in real time as people become more and more drained as every single day is fixing all the “this would be low impact if we were allowed to call it low impact” bugs. And that person who clearly was bored and searching for the corneriest of corner cases (the bug that “nobody knows about”)… that causes significant psychic damage to the person who reads the report and has to fix it.
Its not burying your head in the sand.
It is, just because people haven’t reported it doesn’t mean they haven’t experienced it. Maybe 90% of the people experienced that bug, but only the ones on Linux reported it. It had to be a very big number so that statistically less than 6% of the population experienced it enough to report it. Think about it, what are the chances someone specifically would get a generalized bug? If it’s 1% the chance that that 1% happens to be within the 6% of Linux users is very slim, for that to happen 400 times it’s inconceivable, those bugs were widespread, just not reported.
If it is isolated or people just don’t care? Then… it kind of doesn’t actually matter.
Again, you’re making an assumption, the bugs were probably not isolated, and we don’t know what they were so maybe they were big deals, just unreported big deals.
You scan the forums and optimally have community managers/PR people to do the same to keep an eye out for “This was weird?” style comments but you mostly focus on the stuff that naturally rises to the top or that you identify as an issue.
So you’re saying getting a bug with reproducible steps is worse than having to hire people to search the internet for posts and then pay engineers money to try to reproduce, so that you can finally have the same thing you would have gotten for free? Dude, sometimes people say “the game crashed, piece of shit” and that’s all the info you get in a forum, whereas a bug report is more akin to “When talking to NPC X the game crashed, here’s the stack trace, here’s my save file right before, I’ve confirmed that going and talking to X immediately triggers the issue”, but you do you, hire a community manager full time to read posts in case someone says the “the game crashed”, then pay a QA to sit on their hands until such report comes and then spend months to try to reproduce the issue, to finally get the same bug report that some random person would have given you for free.
The more bug reports you have? That is engineer time spent assessing what is and isn’t a priority.
No, engineers fix the bugs, project managers asses whether a bug is or isn’t a priority, or you thought their job was just to guide you through scrum practices?
And the sad reality is that it is a LOT easier to say “we have our five thousandth number one priority” rather than to say something doesn’t matter.
All you have to say is “your bug has been reported, we will look into it”.
And the sad reality is that it is a LOT easier to say “we have our five thousandth number one priority” rather than to say something doesn’t matter.
It sounds like you’ve worked in a pretty bad development environment in relation to bugs in the past
our five thousandth number one priority
if we were allowed to call it low impact
I think I found the problem. Have you never heard of triage?
You scan the forums and optimally have community managers/PR people to do the same to keep an eye out for “This was weird?” style comments
This takes a hell of a lot more time than just reading a bug report and adding it to the list.
From your comment, I doubt you’ve ever done any software development before.
Buddy you’re talking about bugs you know about but calling them bugs nobody knows about.
A bug that nobody knows about is not a bug.
Did you read the post? It says, “This 5.8% of players found 38% of all the bugs that affected everyone.”
Bugs that affect everyone are not bugs that nobody knows about.
And beyond this specific game, let’s remember that it’s very common for players to experience bugs without filing bug reports, but still complain (either to you or publicly) about your broken game. So you won’t have identified these bugs, but they will still be out there affecting player experience. If you don’t care about that, consider that they will also affect word of mouth and reviews of your game, and therefore your sales.
We’re all limited by hours in a day. That makes this all the more important: A bug fixed once is a bug that doesn’t consume support time (and budget) ever again.
A bug that nobody knows about is not a bug.
Thats not at all what’s going on here. These are bugs that are affecting everyone but weren’t previously being reported. The users know, even if their response is to bang their head against the keyboard, frighten the cat, and vow to never give you more money instead of filing a report.
Good bug reports are gold.
I am going to ignore…
And then you just get really weird. You OK?
As a dev with many years of experience, a bug no one knows about is ticking timebomb waiting to blow up when you have the least amount of time to deal with it.
I’d much rather have it captured and known where I can try and find time to fix it then have it blow up in my face.
There is a world of difference between a bug that doesn’t get reported because its impact is minimal and a bug that doesn’t get reported because people can’t be bothered to make the report and just live with it. The latter category is where the general complacency of 94.2% of players makes a negative impact.
It also leads to people leaving negative reviews on steam. It’s a lot easier to leave a bad review than to report bugs.
But I think it DOES ignore the reality that adding actual support for a new platform does drastically increase the testing and build/deployment overheads which are usually the realest of costs anyway.
John Carmack thinks it’s a worthwhile tradeoff.
Quake III had an official Linux retail release back in the day. It apparently broke even on id’s costs. Carmack said it was worthwhile, anyway, because code that works on multiple platforms tends to be better code. It makes fewer assumptions about the underlying system.
I fucked around once in the open source release of Q3, and yeah, it’s really good code. I had an idea for a game that would need specific joystick support, and with no experience with the code base and limited gamedev knowledge at all, I found exactly the place to change and made a working build within an hour. Carmack isn’t just good at optimization, he’s good at clean, organized code.
And, mostly starting in the Quake 3 era, iD was very much an engine company that sold video games to pay the bills. And the team in that era very much believed in the ethos of Open Source development (hence releasing the engine under a GPL when they were “done with it” as it were).
But they also were effectively just targeting PC. I THINK the Dreamcast port might have been first party but many of the console releases were done through third party studios handling the porting process. Which gets back to “hours in the day” and the realities of project management.
Oh yes, just targeting the PC, the magical fairytale console that only has one hardware configuration that never changes and is so notoriously easy to develop for.
What are you basing this opinion off of? Do you have any evidence that would lead you to think this way?
Also steam deck being a thing now
I remember reading this post.
3 platform specific bugs for Linux, the rest were general bugs and getting repeatable steps and actual data on the problem, not just the game freezes or the game doesn’t boot.
In addition to the Dev’s experience of the Linux community being more likely to collect information and submit quality bug reports (even assisting in bug diagnosis after the report), many developers just won’t test in Linux and fix bugs, while doing that regularly for Windows. So, their under-tested version is buggier.
That’s awesome, OP!
Also
That was a few years ago.
Figured b/c https://theverge.com/news/757538/reddit-internet-archive-wayback-machine-block-limit
Well, the old way of Linux development was actually more work for developers of proprietary applications, especially games, regardless of that specific post having been shortsighted. Whereas enterprisey applications could just target RHEL and call it a day, end user stuff like games did suffer from a broader and ever evolving set of systems.
It’s very different now. Linux-native games just need to target whatever the latest Steam Linux Runtime is at that point (it’s basically a stripped down Debian Stable container, SLR 3 is based on Debian 11, SLR will probably be based on Debian 13).
With the exception of NVidia, all GPU vendors build their drivers on Mesa, so they are more homogeneous than Windows.
So just AMD and Intel?
So just AMD and Intel?
I think the big Linux-supporting ARM vendors as well but I did not check. Apple does not support Linux, so they don’t factor in.
… sort of.
The old model was definitely hell and there is a reason basically no studios supported Linux builds.
These days? Yes, you can go a long way with targeting the SLR. A buddy who does game dev for his day job describes it as “targeting 3/4 of a platform” or being like targeting two generations of the same console. In theory, your middleware handles everything. In practice, you have another platform for your testers to evaluate RCs on and you still find weird corner case weirdness.
But the issue is also… that lets you target the Steam Linux Runtime. What about other storefronts? And the people who tend to care the most about actually making Linux builds are the same ones who aren’t fully comfortable with the idea that “SteamOS == Linux” as it were.
So it becomes that discussion of whether the added testing and development burden is actually worth… still not actually being all that great ideologically.
What about other storefronts?
Which other storefronts? Flatpaks can target whatever the closest runtime is there. The Steam Linux Runtimes are just as FOSS as Debian.
And the people who tend to care the most about actually making Linux builds are the same ones who aren’t fully comfortable with the idea that “SteamOS == Linux” as it were.
Steam Linux Runtimes are NOT SteamOS. SteamOS is Arch-based, the runtimes are based on Debian Stable.
At that point they arent fucking even using a premade engine if they are that uncomfortable with the idea.
Which means if they are building everything from scratch it doesn’t matter anyways.
With the exception of NVidia, all GPU vendors build their drivers on Mesa, so they are more homogeneous than Windows.
This sentence is written in some form of alien language that looks like human writing but clearly isn’t.
Alternately, it may or may not be a particularly playful segment of Alice in Wonderland.
I’d assume that if it was 30% of sales it’d still be around 38% of bug reports.
And in what world are more big reports a bad thing. Except for spam?
I think the implication is that linux has far more bugs, so it’s not worth supporting it for such a small audience. That’s when more bug reports are bad.
This post is raising that only 3 of the 400 linux user bug reports were actually linux related, so it’s not that linux as a platform has far more bugs, but that linux users are much better at reporting bugs.
I remember this post! I ended up buying the game afterward, and although I don’t have a huge number of hours on it, I did find it pretty good.
