Do you think Linux is catching up?

Ah, that's great news then! I had heard something mentioned about middleware portability before somewhere but never looked into it myself.

I think optimally that platform support availability is ultimately something a developer really needs to decide during the planning phases of a project optimally, but also that even if a developer is planning only on supporting a single platform that there is a huge benefit of focusing all planning development to be portable from day 1 and not rely on any non-portable middleware nor platform specific features, and if any platform specific features are to be used they should be relegated to a custom portability layer where similar features on other platforms could eventually be dropped in place.

For example, instead of calling OS specific audio interfaces, one should abstract that by using OpenAL or similar. Using SDL or similar for graphics setup abstraction, and the list goes on and on. Valve has a huge amount of developer videos and other resources to impress these things on developers minds as putting a little thought into these decisions at the start of a project not only can save a developer's work effort porting to other platforms later on, but in fact can speed up project development even on a single platform.

Virtually all code I write is developed specifically for Linux but I almost religiously code in a portable fashion from day 1 even if I've no specific plans to make my software run on other platforms at the moment or ever. If this way of coding actually took a lot of extra effort or added complexity to a project there could theoretically be a reason to not spend time writing portable code, but I've never found it to cause extra work but rather more often to make LESS work overall. Certain individual things may take more work or less work, but overall I find that portable design and software development ends up being a win-win.

Another thing I've learned along the way, is that even if you do not officially plan to release your software on other platforms such as Mac or Linux and you only plan to target Windows, you can still develop it on a cross section of those platforms as often a bug is triggered more easily on one platform/architecture than on another, so other platforms can end up finding bugs during development more easily even if the finished product only ships on say Windows/x86_64. As an example, I used to discover a lot of programming bugs in both my own and other software simply by building it on Alpha. Some of that was due to code turning out to not be 64bit clean as Alpha is a 64bit platform, while other issues were related to code alignment which is mostly a performance problem on x86 whereas it was a violation on Alpha. Making the code 32bit/64bit clean ended up improving the quality and stability of the code on both platforms even if the final results used were limited to just 32bit x86. Other platforms used in the mix for building/testing included 32bit/64bit sparc, PPC, s390, s390x, ia64 platforms, and even occasionally MIPS and ARM later on. Another issue that had to be dealt with occasionally was endian related issues on PPC/MIPS platforms. The extra effort spent cleaning up bugs found like this on other architectures and OSs made the code so much more robust overall, IMHO saving a lot of effort off of the long run, and actually making it much easier to support other platforms from the mix later on if need be.

Game developers come mostly from a different world though I think. Most probably get their start in a Windows centric environment, and I think end up hard coding a lot of stuff to the Windows platform, directly using OS vendor supplied APIs rather than an open abstraction layer, and developing primarily on and for a single architecture such as 32bit x86 or 64bit AMD64 or perhaps both of those but nothing else, and that is where a lot of bugs and problems end up hidden that only show up when one actually tries to port the code later to another platform and has to deal with OS specific and or CPU architecture design choices biting them in the ass.

I've no idea if any of this is specifically the case for CDPR's games however, but I think generically speaking the majority of games developed out there would largely fit into the mould I've described above.

Ideally the best game ports to Linux for example, are ones that are not ports at all, but where the code was designed to be cross platform portable from the start and shared across all platforms using abstraction layers to hide any necessary OS/arch specific bits. As far as games are concerned though, Valve has a lot of resources covering topics like this for developers. If only more devs would take advantage of it, we'd find more games available for Linux and of equal quality to the Windows releases. Ditto for Mac. The need to make a "port" of a game as some kind of monumental effort would more or less vanish. :)

Having said all of that though, there can still very much be effort required even with the most portable code ever written. One still would have to test on all of the supported OS release + hardware combinations, driver variations etc. and have the resources to put into all of that both financially and manpower. One would have to be able to justify the assignment of those resources based on bang for buck in terms of revenue expected in return versus what assigning those resources to other projects might bring in in terms of ROI also. I think this is where things fail on Linux and/or Mac sometimes, is that it isn't that a company does not necessarily have the resources to do a Linux or Mac release of their game or other software, it's that they often perceive that from a business standpoint of having finite resources they can assign the cash and developers to working on their next game or program aimed at the mainstream platforms and end up with a higher ROI from the investment of resources on the next project than what they'd get from using resources to port existing software to more niche platforms that make up the smallest amount of their potential market.

Personally I think this latter reason is the larger reason why there are not more Linux releases rather than a company hating a platform or being lazy or anything like that. It comes down to allocation of resources and ROI in the end IMHO.

Hopefully for TW3 though, CDPR will see value in making it available to other platforms at some point.

Sure, Linux is catching up fast: https://mesamatrix.net

Where are latest OpenGL and Vulkan on OS X?

Not porting a game to a platform that only represent a fraction of the market is not necessarily a "major planning mistake" or even a "lack of professional foresight but most probably simply a sound and calculated business decision.

If you consider that the extra sales you will get by doing a port to another platform is not worth the cost and effort to do it then you don't do it; It's basic business 101.


It's like it was for Linux games on GoG before; CDPR will release Linux version TW3 and/or CP2077 when/if they consider it is profitable for them to do so, not before.

You can call it a "business decision" but there is nothing calcualted or sound about it, when said management later comes to developers and says "hey folks, let's port it to Linux now", and developers need to scratch their heads and tell the management "wake up call guys, it's too late and not feasible now without huge rewrite of major parts of the engine". Nothing about it screams sound and calculated. More like lack of foresight and planning, plus erratic ad hock decision making and false advertising in the processing (causing failed promises and ruined expectations).

Anyway, that already happened. The possible positive outcome - they could learn from these mistakes, and avoid them in their future games. But I'm not confident that they would, since they didn't comment on any of their future plans.

I completely agree with you on that. It reminds me of the late 90s and early 00s when various game companies decided to try porting some of their games to Linux to dip their toe in the water to see how viable it was to do so. For the most part I imagine these companies did it in an exploratory fashion rather than a core business decision to maximize profitability. There simply was no available demographics out there as to the viability and so either nobody did it, or someone had to explore it. Epic Megagames was one of the companies, as was id Software and a few others. Additionally though, some of the core decision makers of these companies were themselves die hard Linux enthusiasts and had personal reasons to make this happen that went beyond "good business investment decision" or even "exploratory business decision" and fell more into "Linux is cool!" territory.

In the end though, from a business perspective all if not virtually all of those experiments ended up being nothing more than ideological as there might have been strong enthusiast interest in having games on Linux at the time but their absolutely was no business demographic to make doing so viable in terms of return on investment for resources put into it, or such experiments would have ended up turning out to be more successful than any of them were. After the first several waves of companies doing this died out, and the massive Loki Games scandal/fiasco, big name Linux gaming kind of took a vacation for a while in terms of volume anyway.

It took some time, but eventually things picked up a few years ago and I credit Valve software for being the biggest shot in the arm for Linux gaming to come along yet. Valve provided a lot of enthusiasm and excitement for the platform and brought a lot of interest and information to developers to get them to take Linux seriously, as well as countless contributions Valve made to the Linux/OSS ecosystem to improve it for gaming as well, which in turn helped many projects outside of the world of gaming as an added bonus.

Many people hate Valve and Steam for their own various reasons, and without agreeing nor disagreeing with any of them, I can say that despite how anyone feels about it all, Valve has done some great things for Linux and Linux gaming that are IMHO more than any other commercial company has done specifically for video gaming on Linux and I don't think for example that GOG would even have a Linux games section currently if it weren't for Valve's own excitement and contributions to Linux.

I also think almost everything Valve has done for Linux and involving Linux to date has been entirely a loss-lead type of thing. In other words, they've most likely lost more money investing in Linux to date in the short term than they've made back, and that will likely be true for time to come yet. They're invested in the long haul though and have the money to do that too, so that's ok.

But a big company like Valve, or even one that isn't quite as big like GOG may have the resources to put into Linux to a greater or lesser degree, when it comes to individual game developers - they do not necessarily have the resources to devote to that, or even if they do technically have the resources, it isn't necessarily the best use of their limited resources to pursue it if there are other products, projects, features or things they could invest those same resources into that they perceive would give them a bigger return on investment and/or growth of their company to allocate the resources towards.

So while I'd love to see every game made fully portable from day one and be released on as many platforms as possible including Linux, at the same time I completely understand why some developers may not find it viable as an investment to do so regardless of how much desire many of us Linux enthusiasts would like to see it happen for our own (possibly even selfish) reasons of being free from the confines of a proprietary operating system or for other reasons. :)

Except in case of CDPR they clearly expressed their interest to release Witcher 3 for Linux and they were working on it (but obviously hit some major roadblocks). The only reason I can see for their failed attempt is poor planning. Or that decision came to them very late, and their management wasn't anticipating those difficulties. Problem is, they were warned about them in advanced many times, but they didn't pay attention until it was too late to fix things.

As an anti-DRM fanatic, I entirely agree with your statement.

But look at this issue in the following way:

GOG is available on Linux, and advertise their availability on Linux. However:
- Linux version of certain games are absent from GOG;
- Galaxy for GOG does not exist;
- CDPR itself does not support Linux on its flagship product.

I do not expect that The Witcher 3 on Linux would turn a profit on the short term. However it would show that GOG was a valid alternative to Steam on Linux, that they were taking this new environment seriously. Instead, GOG acts like a paper tiger.

Supporting Linux on day one would be expensive, and would delay the main release platform. Planning to support Linux and supporting it later would be cheaper, and an investment easier to recoup. If GOG would set a strong foothold on Linux (the strong anti-DRM sentiment on Linux would make it easier to accomplish this goal), the investment of supporting Linux could pay off in the medium run.

When? The only thing I've ever seen concerning the game on Linux was some pre-release advertising SNAFU that occurred on Steam which made the game appear accidentally as if it were going to be released on Linux+SteamOS. A day or two later once Steam or CDPR realized the error they corrected it. I haven't seen any official comments or statements or even leaks that they plan to or planned to release it on Linux anywhere. Have they said that, and do you have any links or pointers to where to find the information? I've searched around and not found anything personally but it would be interesting to know.

I like Linux for my own developing, but man. Doing this for Linux community is a nightmare.

Witcher 3 releases on Linux -> People is going to start asking for concrete package versions for their distributions because the most probable thing is that it would be a Ubuntu package, which of course "No one likes once you learn to use Linux" :P

Man... I like Linux, but if it would be no enough with the small size it´s community is in gaming, there are way too many packages/distros. Like, seriously, some of these distros should get their shit together and make one distro somehow if they don´t want Ubuntu.

Sure, I've seen discussion topics come up about Linux game availability here before. It's a game by game thing from what I understand rather than a one answer fits all, but it is my understanding that for some games that have Linux builds, the publisher has simply not sent the Linux builds to GOG. I presume in these cases that GOG has requested them as I can't see any particularly good reason why they wouldn't in general. Another possibility is that they have indeed gotten the Linux build however the quality or supportability of it did not stand up to their scrutiny and testing and they didn't want to put out something that they considered low quality that would be a burden on support. That is pure speculation on my part but I think it is reasonable conjecture. Another one, is that a lot of game devs already find it difficult to send GOG their game and game updates on Windows as it is and often patches lag far behind here compared to Steam. I'd have to assume that a Linux port of such a game would be even more likely to experience this problem here, so it is possible that either the publisher or GOG might forgo a Linux build for this reason as well.

As for Galaxy on Linux, you're right that it does not publicly exist at the present moment. As a long time die hard Linux user and developer (since 1994) in a perfect world I would like to not only see equal treatment for Galaxy on Linux, but also every single game released or updated to support Linux either natively or using some emulation/virtualization or other shims, and for Linux to become a dominant gaming platform that replaces Windows completely and relegates it to a fraction of the market.

Having said that - at the present moment I'm not bothered by Galaxy not existing on Linux and would rather see GOG allocate their extremely finite developer resources to implementing all of the actual core features in the client that it is missing and worry about platform portability later. On Steam Linux makes up something like 1% of the market or less if memory serves correct and while we have no public statistics like that for GOG I will assume that the statistics here are no better than that if not even smaller. If GOG had to allocate and dedicate even 5% of their development resources to getting Galaxy to run on Linux currently to do so for 1% of their target market that would be an unbalanced allocation of resources. If they were to allocate 1% of their resources for 1% of that market then that would make sense, but unless they have 100 developers working on Galaxy and can allocate one of them dedicated to Linux, it would be more difficult to allocate the appropriate amount of resources for the market it reaches while the client has large generic features missing that affect and are anticipated by the much larger audience out there.

I don't think GOG is ignoring Linux, I think they are focusing on what the most important features are for the majority of users right now and while Linux is important it is out-prioritized by more important development at the moment. That not only does not bother me as a die hard Linux nut, but it makes perfect business resource allocation sense to me.

Those updates were posted in this very thread (I thought it's already general knowledge for those who follow this topic, but apparently not yet. I'll put those updates in the first post as historic artifacts. Let me just collect them.

I updated the first post with more info in chronological order. Since those are e-mail answers, take that as is.

You can see how situation gradually changed from more enthusiastic, to more pessimistic, which I suppose reflects various roadblocks and issues they encountered during attempted porting (or its planning).

You can also review this thread for various other details.

Ah, thanks for the info. I've no reason to believe the comments are not genuine so I'll accept them. Having said that, there's not a lot to go on but perhaps they were working on it and it was turning out to require more dedication of development resources than they were willing to continuously pour into it and eventually needed to reallocate the developer(s) to CP2077. Either way it's kind of sad to see. It does underscore a point I often make though which is that I firmly believe game developers are much better off by not pre-announcing things like this even unofficially, and that they're better off to remain dead silent, not even saying "no comment", because anything they do actually say, whether it is a casual non-committal comment or an official company statement - can and will come back in the future and be used against them if they do not come through with people's expectations.

Still, it's the first real information I've seen on the topic even if it is indirect and not easily independently confirmed. Perhaps the Linux gaming market share will grow over time and cause various companies to re-evaluate releasing their games for Linux though. I have to admit, a few years ago looking at the growth of Linux in this space, I honestly felt that Linux gaming usage was going to dwarf Mac gaming within 12-24 months, but several years have passed now and that hasn't happened. Linux gaming has experienced a growth of players and of games, but Mac growth seems to have occurred at the same time and at a faster rate also, keeping Linux down in 3rd place (excluding consoles and other platforms).

Would be nice to see TW3 released for Linux, but on the other hand I'd rather see CP2077 come out sooner. Perhaps they will release CP2077 with Linux support from day 1, and have designed it from day 1 as highly portable, then can take that expertise back and work on TW3 for Linux. :)

Complete silence can prevent failed expectations, but it also distances them from the community, which puts them squarely in the legacy media mindset. Another way to avoid this is to be more open, and communicate changes of plans to the community in the timely manner (with explaining why they changed). Someone will be upset either way, but being open can prevent a lot of frustrations from cancelled plans and the like, plus community would feel more direct connection with developers. Usually crowdfunded projects follow the later approach.

Problems start with approach in between. I.e. when they produce something like those SteamOS ads and confirm Linux plans, and then for a long time don't update anyone about any cancellations or encountered issues. That for sure causes failed expectations.