What Should Valve Do For Linux & Open-Source?

[DMJC]

The Loki Back Catalogue

I think Valve should pick up the Loki Back Catalogue and release it on Steam. It's about 20 games. The sourcecode to most of which is still kicking around some of the developer's drives. Myth 2, Railroad Tycoon 2, Sim City 3000, Descent 3, Heroes of Might and Magic 3, and a bunch of other classic games, perfect to kick off the platform's launch. Instantly boosts the profile of the platform and being old titles probably much cheaper to accomplish than a lot of other efforts. I'd love to say, finish porting Deus Ex 1 to Linux, but I think that's a bit unrealistic now. Still, getting the Loki Catalogue onto Steam would be a really nice thing.

The strength of Linux Steam over Mac Steam is going to be in legacy games. Aspyr Media Inc own half the Mac ports of games people actually want to own, and they want their own online service to succeed on the Mac. Valve doesn't have this problem on Linux, I expect we'll see Doom 1/2/3 Quake 1/2/3/4 and Wolfenstein games all on Steam in packs. Along with UT. The major driver for Steam's success is game sales, with people handing over tons of $5/20/50 transactions buying up old titles as well as newer ones. I think if Valve can get a similar large range of new/old games appearing on Steam, they'll be able to drive similar marketplace forces on their linux selections which could really bust the market wide open for Linux games. Valve should also look at all the Dosbox compatible games on Steam like Commander Keen and launch them with Dosbox for Linux.

[Adarion]

Phoronix: What Should Valve Do For Linux & Open-Source?

My random and unsorted list, just brainstormed after work:


* support of generic Linux - say: support/use standard libs like Qt or GTK, glibc and stuff. Not to have any distribution specific stuff.

* good Multi-CPU/Core Support

* try to support the free GPU driver stack, do not limit to detonators (are they still called like this?) or fglrx

* free as in freedom client?
DRM causes always troubles, and Linux users are especially sensitive when it comes to DRM and binary blobs

* Will my license of HL1 and HL2 also be usable on Linux once the thing is ported?



In the 2nd view, more indirectly:

* have better drivers (free/blobs) for GPUs and generally hardware.
(That then includes OpenGL, OpenVG, video acceleration (UVD anyone?), power management and multi GPU switching/rendering and S3TC and more)

* have HW vendors/creators recognize Linux

* have HW vendors/creators not just recognize Linux but generally think of releasing drivers under a free license and offering specifications to talk with the hardware free of NDA

* have more people think about using LIBSDL and things like it right from the start, creating games already as platform independent as possible.
So as long as the HW fits it can then easily be release on the 3 "main platforms" at the same time. (Linux on x86/amd64, W32/64 and MacOSX)
It might also be easier then to go a step further to different hardware (ppc, arm) or an OS that is close by like the BSDs.

* have then more games and other software platform independent, for everybody to use / to buy a license.

Cheers and good night from Germany 15 minutes before midnight *yawn*

[MaxToTheMax]

I'd suggest implementing the APT protocol as part of their digital distribution system, so we don't actually need to have the Steam client installed. That would help a lot with giving it a "native" feel.

I'm not sure what good a new debugger would do, since GDB is already pretty good. In fact, development tools in general are very good on Linux already (being the hackers' platform.) However, new performance analysis tools specifically for games would be nice, as we are lacking in that area somewhat. Improvements to the speed of Callgrind would make it more usable for gaming (since its current slowness on the CPU tends to under-emphasize what parts of your code are taking too long on the GPU.)

I'm going to chime in with the others and say that doing something about the S3TC patents would be a godsend. I'm working on some stuff for Alien Arena that uses up a lot of texture memory, but I don't want to introduce a dependency on patented tech. The fact is, fixing this problem would be in Valve's best interest, making Linux a more robust gaming platform. Challenging the patents and getting them declared invalid would be ideal (at least one court has declared them invalid, so you'd have some credibility there.) Buying the patents and burying them would be second-best. Licensing the patents and then distributing a software library to everyone free of charge, regardless of distro, would be third-best (if that's legal, I'm not sure if it is.)

[Kernelcoffee]

- Develop / Port their games/tools on Linux
- Identify what's wrong / missing in the open-source drivers and propose patches / work with the community to do the job
- Promote open documentation for drivers and Linux game / cross-platform devel.
- Create a true united community around Linux game / cross-platform devel devel. (Linux Gaming Foundation ?)
- Continue to make awesome games.

Games are a goddamn driving force when it comes to popularity contest among the common people (AKA those who don't know what a terminal is and don't care), which ultimately hold the mighty coin that everybody needs to put food in their mouth.

If Valve can be this company that's going to make Linux game devel relevant enough so you can say "yes I have native popular games on Linux that make rainbow puking unicorns look like dead ponies", and so that you don't have to dual-boot to play the latest game with your friends.
Then by all means, I'm ready to stand up and at least show my appreciation in any form I can (short of giving my organs), well the most likely way to do so for me will be to actually use the steam client, buy games, and get the Linux statistic up to show that there is a gaming Linux market. Well if they have a debug mode that you can enable to report statistic on game performance or crash report, so they can identify issues, I would be more than happy to write detailed reports.

I don't care if Valve doesn't release a single line of code of their games as open-source, what matters to me is that the foundations on which the games resides get stronger, simpler and more coherent, so that other people that are willing to do create open-source games or not can do so more easily and more efficiently.

Valve through Steam has a real gaming catalog, a strong and well known tool to provide people with games and enough influence to make the market move toward a more Linux-friendly one.

[M1kkko]

* Will my license of HL1 and HL2 also be usable on Linux once the thing is ported?

Half-life 2 yes, probably just the same way as it's on Mac. Your game saves will also likely be synchronized through Steam Cloud and you'll be able to play multiplayer games against Windows and Mac users etc.

I don't know when/if half-life 1 is going to be ported, but at least it fully works under Wine. At least it's not natively available on the Mac right now.

I'd suggest implementing the APT protocol as part of their digital distribution system, so we don't actually need to have the Steam client installed. That would help a lot with giving it a "native" feel.

I'm afraid that may not be possible. Steam is essentially a package manager, and it has its own way of downloading/managing games. Also the games are launched by the Steam client, which requires authentication with your Steam Account and also provides essential services used by the games (community overlay, achievements, etc) not to mention digital restrictions management.

Sorry, but you're in no way going to buy/download/run Steam games without a Steam client and a user account.

[Serafean]

Leverage gallium

While their work anywhere along the OSS graphics stack would be very appreciated, I think that they'd gain a lot simply by leveraging gallium, namely the D3D tracker. They could facilitate the porting effort by implementing DirectX (APIs are not copyrightable, remember?). I'm not talking about binary compatibility, but about source compatibility, which would mean that the porting effort would mostly be about recompiling (and since they hve the source code...).
My reasoning behind this :
- face it, there are already many games using DX, if the same API were available on linux, the likes of EA would have one argument less.
- An open implementation of DX would be very interesting for future development : MS being at the helm, mutiny! (to complete the metaphore) Meaning DX could be pushed forward by those really using it.
- We'd get an open multi-api platform. No API would be favored before another. (As we have now with GTK and Qt)

I'll admit that seeing them helping around the graphic stack (anywhere) is a far better future than a DX implementation, but IMO the thought is compelling...

Serafean

[MaxToTheMax]

I'm afraid that may not be possible. Steam is essentially a package manager, and it has its own way of downloading/managing games. Also the games are launched by the Steam client, which requires authentication with your Steam Account and also provides essential services used by the games (community overlay, achievements, etc) not to mention digital restrictions management.

Sorry, but you're in no way going to buy/download/run Steam games without a Steam client and a user account.

Well, at least don't make me look at a Steam client. They should think of it like developing for a smartphone: it should fit in with the existing look-and-feel, and individual Steam apps shouldn't look too different (from an end-user POV) from apps installed from the repos.

[smitty3268]

I agree with Pickle in that open sourcing their back catalogue, such as Half Life, would be awesome.

This will be an unpopular opinion on this board, but I don't think that open-sourcing their back catalog is the best use of their time.

I agree that it would be awesome, and open-sourcing anything is always good. But I'd much rather see a lot of other things come out of valve first.


There are 2 reasons:

1. Valve really supports their old games well to begin with. I can buy a copy of Half-Life (from 1998) right now off Steam, and it is fully supported. I can also get Half-Life Source, which is the same game ported to their more current engine. Based on past results, I fully expect to still be able to buy a supported version off of Steam 10 years from now. Or rather, I'm sure i'll get it for free, because they bundle that with a bunch of their other games for free.

2. Old Valve games were heavily MS dependent. It wouldn't be like Q3 or some ID code drop, where the code already basically works on linux. You'd have to essentially do a complete rewrite of the game - the benefit, of course, would be that you could always see the reference to look at exactly how the original game was done, but it would be a ton of effort for any project to get things working. At least past the WINE stage. To avoid that, you'd need to start with games that had already been ported to the Source engine, and that's new enough that i don't think Valve would be very receptive towards open sourcing that.

[smitty3268]

Well, at least don't make me look at a Steam client. They should think of it like developing for a smartphone: it should fit in with the existing look-and-feel, and individual Steam apps shouldn't look too different (from an end-user POV) from apps installed from the repos.

In windows, installed apps usually give you the option of appearing in the windows start menu or not. I assume the same will be true on linux.

Then you just have the steam client as an icon in your system tray.

[RussianNeuroMancer]

I think Valve should pick up the Loki Back Catalogue and release it on Steam. It's about 20 games. The sourcecode to most of which is still kicking around some of the developer's drives. Myth 2, Railroad Tycoon 2, Sim City 3000, Descent 3, Heroes of Might and Magic 3, and a bunch of other classic games, perfect to kick off the platform's launch. Instantly boosts the profile of the platform and being old titles probably much cheaper to accomplish than a lot of other efforts. I'd love to say, finish porting Deus Ex 1 to Linux, but I think that's a bit unrealistic now. Still, getting the Loki Catalogue onto Steam would be a really nice thing.

The strength of Linux Steam over Mac Steam is going to be in legacy games. Aspyr Media Inc own half the Mac ports of games people actually want to own, and they want their own online service to succeed on the Mac. Valve doesn't have this problem on Linux, I expect we'll see Doom 1/2/3 Quake 1/2/3/4 and Wolfenstein games all on Steam in packs. Along with UT. The major driver for Steam's success is game sales, with people handing over tons of $5/20/50 transactions buying up old titles as well as newer ones. I think if Valve can get a similar large range of new/old games appearing on Steam, they'll be able to drive similar marketplace forces on their linux selections which could really bust the market wide open for Linux games. Valve should also look at all the Dosbox compatible games on Steam like Commander Keen and launch them with Dosbox for Linux.

This is job for GOG, not for Valve.