Wine Devs Have Mixed Feelings Over Direct3D In Gallium3D

[dalingrin]

But Direct3D will not make people migrate to Linux.

We've implemented the following so far:

- FAT support
- NTFS support
- SMB support
- Win API support (WINE)
- MS Office documents support (OOo)
- "Microsoft HTML" support (Gecko+Webkit)
- Mono / .Net support
- Silverlight / Moonlight support
- WMA/WMV/AVI support

...and where are all the people? Still using Windows. You think that they'll switch becasue of Direct3D? Or that THIS is what all the major companies are waiting for? Nah, they don't give a crap, and just like they don't bother programming cross-platform stuff right now, they won't bother programming cross-platform stuff even if Direct3D is available.

Okay...take aways all those interoperability features and see how many desktop users Linux would loose.
I think you'd be surprised how many people use/need those features. I'm not completely disagreeing with you. I don't think Direct3d support is going to bring flocks of users and developers to Linux but I do think acting like the aforementioned features haven't had an impact on Linux usage is a mistake.

[Jimmy]

Do you know the story behind OpenOffice?

The story is well known. My point is that, because of its underdog status, it has to implement support for MS formats to go about its business of killing Microsoft. It doesn't make them promoters of Microsoft technologies. If they succeed, (I'm still holding out ) the fact that they had support for MS formats becomes irrelevant.

And you're right, in the end it's the open standards that will draw people and businesses to it (and other open software). While people aren't switching to OpenOffice because of its import/export capabilities, not having them would be a significant barrier for many new adopters.

But Direct3D will not make people migrate to Linux.

We've implemented the following so far:

- FAT support
- NTFS support
- SMB support
- Win API support (WINE)
- MS Office documents support (OOo)
- "Microsoft HTML" support (Gecko+Webkit)
- Mono / .Net support
- Silverlight / Moonlight support
- WMA/WMV/AVI support

- Flash

No, nobody switches because of those specific things. However they might NOT switch if they weren't there (well probably not Moonlight, no one cares about that Silverlight crap, not even Windows users).

I didn't switch because I was attracted to the FAT support in Linux. However, I might not have considered it seriously if when I plugged in my USB thumbdrive I couldn't access the file system on it. I needed NTFS support to grab photos, mp3s, and personal documents. SMB is nice to have when the parents come over with their laptop. People still send me MS Office documents. Wine is nice for playing games.

These are all great things to have, and really are just temporary in the long run. I don't use NTFS anymore although it would have been a barrier initially. Cumulatively they added up when I switched. Over time they mattered less.

I used to dual boot. What was the final straw? One day Windows didn't boot. I didn't have time to mess with it, so I just did my work on Linux instead. I procrastinated about fixing Windows, just because it was a monotonous and time consuming process that I had endured too many times before. I was able to get my work and play done because of half the things on the list above.

To reiterate, those things mattered less over time but would have been a barrier issue initially. I probably would have gritted my way through the Windows repair if they didn't exist.

In the end I love Linux because I've never had to reinstall because it stopped working, got infected, or got slow over time. I don't need to defragment my file system, clean my registry, manually update programs, and so on.

So yes, those evil technologies are useful and relevant for those looking to dump Windows or take a test drive. By themselves they are nothing and are do not represent the best features of Linux. If Linux crashed like Windows, I wouldn't have switched. It's the whole package deal that makes a Linux OS attractive.

They'll just keep lockign people into their NEW formats and we'll be chasing them instead of making a decent system.

There's some truth here, but the thing that's commonly missed is that developers often only chase balls they are interested in (or paid for).

You may think that a new file system is redundant, not needed, and a waste of time. Time that could be used to improve OpenGL support. However, the developer working on that file system might not be any good at OpenGL stuff or even interested in it. Even if they stop working on something you consider a waste of time, it doesn't mean that they'll pick up something you care about.

IMO you're just as well letting them chase the balls they like and just look for other developers that are genuinely interested in working on the things you deem important. Find ways to encourage, pay, and support the projects and developers you like. It's a lot less futile than dogging programmers who choose to work on something else.

I personally don't think D3D on Linux will see much use outside of Wine (if it even gets used there). It's likely not enough anyway. Chances are if you drink the D3D Koolaid you also drink the rest of DirectX and other Windows APIs that don't port well. My big deal is that it is not worth all the fuss. None of this is gonna kill Linux.

[dalingrin]

Jimmy, I wholeheartedly agree with every sentence, phrase, word and letter you've said.

[BlackStar]

Jimmy, I wholeheartedly agree with every sentence, phrase, word and letter you've said.

++

Also, "[let developers] chase the balls they like". I love this mental image, if only because it is so accurate!

[BlackStar]

I agree that you can consider OpenGL to be broken on Linux at the moment. It works for me, but a game company will have to deal with crappy drivers. But don't lose hope: companies like VMware, Red Hat and AMD are fixing it. D3D10 on Linux is even more broken, because you can actually rely on the fact that it *doesn't* work.

I don't lose hope, I use Linux exclusively and love it despite the issues. These driver problems are certainly not endemic on Linux - indeed, Linux might actually have better OpenGL support on the whole compared to Windows or Mac OS X.

However, I am pragmatist enough to know that if I release a cross-platform OpenGL 2.1 game I will have a smaller potential user-base than if I release a D3D9 Windows-only game. The numbers only get worse if you move to GL3.0 (vs D3D10) and GL4.0 (vs D3D11).

I am also enough of an idealist to not really care: I'd rather support Linux even if that meant a hit in revenue and a harder development process.

[V!NCENT]

The numbers only get worse if you move to GL3.0 (vs D3D10) and GL4.0 (vs D3D11).

That depends. It's not like that as if you started making a game right now; you'd be finnished next week

In two years we'd probably be at OpenGL 3.0 (FLOSS)

[pingufunkybeat]

Well, of course interoperability is important -- I like using USB sticks, for example, and they pretty much all use FAT.

But interoperability is a stopgap measure to help people migrate and help them break out of monopolist lock-in "standards". You can read your old Word files and convert them to ODT. You can access your old data, or interoperate with a Windows network using a Linux computer. Perhaps you can run a very important program which is not available on Linux.

But we should not make FAT the default filesystem, or MS Office the default office format, or SMB the default network share technology. We should not build our desktop on Microsoft technology. Similarly, we should not rely on Direct3d for 3d graphics. I don't see what a native D3D implementation will do, given that all D3D apps are windows-only. It weakens international standards so people can run Windows software instead of asking for native stuff.

THAT is a problem.

[Svartalf]

However, I am pragmatist enough to know that if I release a cross-platform OpenGL 2.1 game I will have a smaller potential user-base than if I release a D3D9 Windows-only game. The numbers only get worse if you move to GL3.0 (vs D3D10) and GL4.0 (vs D3D11).

Really? This depends on several factors.

If you're coding OpenGL 2.1 the way many do, you should be able to target:

Windows (YES...you don't HAVE to code to DX9... )
MacOS
Linux

And, in the large, you can do most of the "fancy" stuff within 2.1. The only notable exception is if you're reaching for tesselation type tricks and the like- then you'll be coding to 3.0. For that, again, the available platforms would be:

Windows
MacOS
Linux

With the caveat on Linux that you have to verify the things you're doing against both AMD and NVidia hardware and state any gotchas up-front.

Now, if you're coding 2.1 in a clean and tight manner so that you can target OpenGL ES 2.0, you also have as potential targets:

PS3 (Though you're probably better off re-coding for their API...)
Any Android with a modern Cortex-A8 class CPU (Read most of the latest)
Any iPhone from the G3 forward.
Any iPad model.
Any ARM Cortex A9 system about to be shipping in the near future.
Many others quite too many to count...

In short, the thinking you're using misses that there's actually quite a bit more out there (Including as many mobile devices as there is potential market share numbers for the consoles...that won't ever be using D3D...).

This isn't saying that you should avoid D3D (You'd miss out on the X-Box stuff then...) but to say that you'd limit your numbers is quite missing the fact that there's quite a few more markets than Microsoft's generated one.

[mdias]

...Similarly, we should not rely on Direct3d for 3d graphics.

Yep, I agree, the thing is there's nothing else comparable we can rely on. OpenGL just isn't on par with D3D for developers.

I don't see what a native D3D implementation will do, given that all D3D apps are windows-only. It weakens international standards so people can run Windows software instead of asking for native stuff.

THAT is a problem.

How did you expect a D3D app/game to not be windows-only if until now D3D only existed on windows?
You seem to believe the only reason we agree D3D1X state tracker is good is to port windows games to linux. That's incorrect, we just want to have a nice (nicer than OGL) API to develop on Linux FOR linux. The added easyness of porting a D3D game is just a nice secondary effect.

[mirv]

Yep, I agree, the thing is there's nothing else comparable we can rely on. OpenGL just isn't on par with D3D for developers.


How did you expect a D3D app/game to not be windows-only if until now D3D only existed on windows?
You seem to believe the only reason we agree D3D1X state tracker is good is to port windows games to linux. That's incorrect, we just want to have a nice (nicer than OGL) API to develop on Linux FOR linux. The added easyness of porting a D3D game is just a nice secondary effect.

Just because people are now devolving into "did not", "did too" school yard level arguments....why do you believe d3d is nicer than OpenGL? And when answering this, please remember that some people still use C quite heavily.