Ryan Gordon Criticizes Open-Source Drivers Again

[Wyatt]

That's a very long conversation, but suffice it to say that it flies in the face of traditional Linux software development. If you are in favor of static bundling, you have a whole lot of people that you're going to need to convince. I don't know enough about the topic to make a bulletproof argument either way, but my gut tells me that static bundling is a bad path to start walking down.

There's this weird BSD distro...it's kind of rare, so you might not have heard of it. I think it's called "OSX" or something?

That sounds like a silly way to handle the problem. Now a commercial software vendor is tied to a package maintainer for each and every distribution they want to support.

No, a vendor is tied to the tarball they release. Distro users are tied to the package maintainer, and only if they have a valid license file. Autodesk is a decent example (though their licensing software is an arse pain at times). This applies all the way down your post. Distros refusing to package is not the vendor's fault. Dependencies are not the vendor's problem. That's why we have distros.

We need a unified packaging system.

No. What we need are companies who have been educated to understand the role of distributions in distributing software and distros that aren't getting in their users' way by not packaging software the users want. It's not that complicated.

...the current 'look at me, this new gimmick in my environment/distro that asks for all applications to be rewritten is totally awesome!' attitude.

Erm...the only distro I see doing that is Ubuntu because they're obsessed with iterating on the wheel so they can run their users over more efficiently. (This may be unfair, but having tried Unity, I am shocked at how poor it is.)

[pingufunkybeat]

Which is... where?

Dozens and dozens of computer architectures, supercomputers, universities, webservers and, yes, many desktops.

You can get a complete, multi-user, multi-purpose OS with top performance for free.

Please compare the speed at which the Linux kernel gets patches to the speed at which Adobe Flash gets patches, and you might understand why openness is important.

There are many places where Linux excels. There are also some where it does not -- like games. I wouldn't trade the advantages for a latest DRM-infested Blizzard borefest, which is where many new Linux users want to take it. Openness is important, but it carries certain costs. I can live with them.

1% of the desktop is fine. It's a really usable system. I can run scientific calculation on Linux running Octave, and it doesn't cost a thing. A single MATLAB license on a legal MS windows environment with all the needed extensions would be more expensive than our entire lab, including the robots.

I'm very happy that some zealots insisted on giving me an open alternative.

[FreeBooteR69]

I think they should just distribute the software using tarballs and let the the individual distributions build systems and package managers manage the tarballs. With Archlinux we would use the Arch build system then install with pacman. Works great and is well documented.

[mirv]

Which is... where? 1% of the desktop? Not much to brag about on the linux side, frankly. And it's been 20 years. Look what Microsoft did with 20 years. Ok we have a successful platform with Android, but I don't know how "open" Android can really claim to be. And ok, linux is successful in the server realm, but most of that success can be attributed to what amounts to a super-cheap variant of UNIX, detached from expensive hardware.

Linux brings very few competitive advantages in the operating system world which is why, in general, it hasn't been embraced.

It's pretty much been the religious-like zealotry that has kept linux back and stuck in the "hobby OS" territory... which, if you think about it, is really where a lot of those religious types wish it to be. Which is fine. But then you can't complain about proprietary video drivers and lack of software and no games for linux. If they want to keep their clutches on it and not open it up to a general user base, then don't expect developers and publishers to support the platform.

You know, that "hobby OS" is used to design AMD chips. Something like 90% (or more) of the design tools used by AMD in their...hmm...K6 line I think, but it might actually be the Opterons, or both, run under linux. Their validation systems and testing environments run under linux. Yes, it started out because it was cheaper than paying for other (buggier) UNIX software tools, but that's actually an advantage. Linux also runs in....actually any embedded system with enough power to run it. Routers, fridges, it can even run on Xilinx soft core processors, even stock markets use it.
It's always been more appealing to engineers than to home desktop users, because at its heart it's still by programmers, for programmers. So I don't think calling it "hobby" is quite accurate. It's never really been made for home user desktops, although work is being done there by some. However, that doesn't mean it's hobby by any means, or not embraced anywhere. It is far more reaching that many seem to realise, far more widely embraced, and actually does fill a desktop niche for those programmers, engineers, or computer oriented who use it.

[ean5533]

It's pretty much been the religious-like zealotry that has kept linux back and stuck in the "hobby OS" territory... which, if you think about it, is really where a lot of those religious types wish it to be. Which is fine. But then you can't complain about proprietary video drivers and lack of software and no games for linux. If they want to keep their clutches on it and not open it up to a general user base, then don't expect developers and publishers to support the platform.

There is a lot of interesting things to talk about in your post, but this really isn't the right thread for it. It's a different topic, and the flamewar that will spawn from it will crush the discussion at hand.

[johnc]

There is a lot of interesting things to talk about in your post, but this really isn't the right thread for it. It's a different topic, and the flamewar that will spawn from it will crush the discussion at hand.

That's true. I don't want to take the thread off point.

I like linux. I just get disappointed at times because I think there's a lot of potential being left on the table.

[ean5533]

That's a very long conversation, but suffice it to say that it flies in the face of traditional Linux software development. If you are in favor of static bundling, you have a whole lot of people that you're going to need to convince. I don't know enough about the topic to make a bulletproof argument either way, but my gut tells me that static bundling is a bad path to start walking down.

There's this weird BSD distro...it's kind of rare, so you might not have heard of it. I think it's called "OSX" or something?

Since my post was talking about "traditional Linux development", I'm not sure why you think a closed-source BSD-based distro is relevant to the conversation. Traditional Linux development is open-source and not based on BSD. Apple's development model is completely different than the traditional Linux development model.

Also, if you have a point to make, can you please make it without the snarky attitude? All it does is deteriorate the quality of the conversation and make you sound like an asshole.

No, a vendor is tied to the tarball they release. Distro users are tied to the package maintainer, and only if they have a valid license file. Autodesk is a decent example (though their licensing software is an arse pain at times). This applies all the way down your post. Distros refusing to package is not the vendor's fault. Dependencies are not the vendor's problem. That's why we have distros.

"Fault" is completely irrelevant. It makes no difference who is philosophically at fault in this situation. From the point of view of the vendor, all that matters is that users won't have the opportunity to use the vendor's software, and as a for-profit company that situation needs to be solved.

No. What we need are companies who have been educated to understand the role of distributions in distributing software and distros that aren't getting in their users' way by not packaging software the users want. It's not that complicated.

If it's not that complicated, why have these problems persisted for so long? If your answer is "because XYZ is stupid", then your argument isn't well thought out.

[kraftman]

From what I can tell, Ryan is a person that deeply loves linux and open source, but is also incredibly pragmatic. His hybrid binary format got shot down by the zealots, which is unfortunate. I for one would love to see more people like him in positions of power, instead of people that argue whether the man-pages are free http://www.emacswiki.org/emacs/2006-03-15 . Here is a guy that can actually improve an experience for users.

While he makes a lot of sense now, his idea about the hybrid binary format was dumb stupid and thankfully it died.

[ean5533]

That's true. I don't want to take the thread off point.

I like linux. I just get disappointed at times because I think there's a lot of potential being left on the table.

I've started a new thread so we can discuss this further.

http://phoronix.com/forums/showthrea...Linux&p=221980

[Wyatt]

Since my post was talking about "traditional Linux development", I'm not sure why you think a closed-source BSD-based distro is relevant to the conversation. Traditional Linux development is open-source and not based on BSD. Apple's development model is completely different than the traditional Linux development model.

Also, if you have a point to make, can you please make it without the snarky attitude? All it does is deteriorate the quality of the conversation and make you sound like an asshole.

Damn, man, what's got your panties in a bunch today? It would be hard for you to have had a worse day than me; why not lighten up a little? Anyway, since you were more focused on the "BSD is not Linux" argument than trying to understand why I said that: what you're suggesting is a poor idea is, in fact, just how OSX packaging works. It's wasteful as all get-out and I'll never dare say I like it, but it exists in significant deployment and it unequivocally works. All humour aside, this isn't something you can just dismiss with "how it feels" arguments.

"Fault" is completely irrelevant. It makes no difference who is philosophically at fault in this situation. From the point of view of the vendor, all that matters is that users won't have the opportunity to use the vendor's software, and as a for-profit company that situation needs to be solved.

Which is why it's a tarball that anyone can install without the package manager or maintainer's blessing. This isn't to say limiting the userbase has ever stopped companies before. Note the platform limitations of games and it should be pretty clear. No, I don't mean in the short-sighted sense that Phoronix tends toward with Windows games, but in the broader ecosystem of iOS/Xbox/PS3 exclusivity that currently plagues the landscape. Vendors have shown they don't care already. That they perceive they need to care for Linux is as baffling as it is wrong.

(Please don't next argue that your distro can't easily package a tarball easily or something.)

If it's not that complicated, why have these problems persisted for so long? If your answer is "because XYZ is stupid", then your argument isn't well thought out.

There is no sin in ignorance but you should realise my "argument" is simply something distros have been doing for well over a decade: packaging software. This isn't about idealism or flights of fancy. This is about removing subjective corner cases in reality through better education.