TR Forums

I have been thinking about FOSS zealots who complain about Linux Distros that bundles proprietary software with their product, and also complain about users who are usingboth FOSS and proprietary software.

I have some issues with this. Yes I prefer to use FOSS instead of proprietary software. However I am also a realist and willing to comprise on this. Here's why:

Linux has to work on the hardware the user already has.

Linux also has to have the same [q]functionality[/q] as Windows. Otherwise what is the point?

The FOSS development model may not work well with some types of software. Games may be a good example of this.

I rather use Linux with proprietary software then not being able to use Linux as all.

So what do you guys think, am I being reasonable here?

It's a perfectly reasonable position, and I share your pragmatic position. However, I do understand why the FOSS zealots push the positions they do (security, better software, and longevity), and try to support companies who are FOSS friendly.

Hardware drivers and firmware are different issues then userland software.

I use proprietary applications on Linux, Komodo IDE to be specific, and I have run proprietary software on Linux. I don't have a problem with it. Really, the risk of the software becoming abandonware is about the same as if it were FOSS. I can code, but I don't really want to support an application, especially if it's something as complicated as Zimbra. I'll just move to something else.

The kernel devs like to break the kernel ABI/API. This is why drivers need to be rebuilt for new kernels, and this is also why Red Hat uses one kernel version for RHEL and backports updates and fixes. This is more complicated, but I'm on the side of the devs on this one. It's a piece of hardware, and not having a FOSS driver is silly. If people can't use the hardware, they aren't going to buy it. They give the driver away to make the hardware useful, so giving away the driver code too isn't that much of a leap. The worst thing that happens is people buy the hardware and improve the driver. Of course, there needs to be documentation, and that's a separate issue and really a sign of a bad company culture, in my opinion.

Firmware is a gray area. As much as I'd like to have binary blob free kernel, it's not practical as this really is where all of the secret sauce is. However, this is also an area that could be exploitable, and having visibility into what is going on would be ideal. Particularly basebands in cellphones.

Of course, there are only so many manhours available, and most companies aren't interested in something like desktop Linux. Compromises just have to be made.

Stallman!

So extreme he doesn't even want to use a computer where the hardware isn't open source. (Basically wants documentation on the hardware level so he knows exactly what it's doing)

Flatland_Spider wrote:

Of course, there are only so many manhours available, and most companies aren't interested in something like desktop Linux. Compromises just have to be made.

Yes they do, and Rome wasn't built in a day.

Flatland_Spider wrote:

It's a perfectly reasonable position, and I share your pragmatic position. However, I do understand why the FOSS zealots push the positions they do (security, better software, and longevity), and try to support companies who are FOSS friendly.

Hardware drivers and firmware are different issues then userland software.
- - -
Of course, there are only so many manhours available, and most companies aren't interested in something like desktop Linux. Compromises just have to be made.

There is just a couple of problem with that first position though... in theory, it's actually great, but in practice, which especially the last year has shown, is that FOSS doesn't in actuality make software more secure, although many people still assume this. But after both openssl, bind and apache having been subjected to some nasty things I think going forward might be better. But the discipline to do proper secure coding is hard, no matter if it's FOSS or closed source.

Now the good thing with FOSS of course is that it's open, but that by itself doesn't make it audited. And even closed source software can be audited in many circumstances under NDA, etc. Now, one other incentive is that with closed source, you can generally have monetary leverage against the developer if something breaks. Which in some cases is the correct for the business if not for either the greater good or the availability of such software.

That said, there are also shining examples of where companies do keep the software open, and actively develops for it, but makes their money from implementation and other expertise in building a system around that. Which IMO is the better way to combine business with FOSS.



When it comes to hardware drivers, especially the cutting edge, part of what makes something special and better then other things might be certain optimization, and I'm pretty sure the companies that produce said hardware aren't going to want to have their newest IP out there for all too see. Now if there was a system to get audited closed source parts for integration in an otherwise open framework, that would be a decent compromise.


As for windows on the desktop, it's kind of hard to take away the fact that the ecosystem Microsoft provides is kind of hard to beat. I'm sure you can do equally well on other plattforms, but instead of money, it'll take time and talent, which in many places, are really hard to get to. And when that talent leaves, you can't just go out and get another one as easy.

I understand where the zealots are coming from, but the "all or nothing" mindset probably hurts their cause more than it helps, by driving potential users away. IMO extreme FOSS evangelism also helped feed the "if you use any FOSS software in your business then you need to disclose all of your own proprietary code to the world" FUD that was being pushed by Microsoft back in the day; it has taken us roughly a decade to dispel that myth.

I prefer to use FOSS alternatives when they are available; but in some cases those alternatives either don't exist, or lack significant functionality. I have no qualms with using proprietary software on Linux if it provides functionality I want that I can't get otherwise. I also maintain a Windows 7 VM on my Linux desktop, for those cases where a piece of proprietary software is Windows-only.

Oftentimes lack of support for a given device is due to hardware manufacturers refusing to release detailed specs to the people who write the FOSS drivers. I research my hardware purchases to ensure that I choose devices from vendors who work with the FOSS community instead of against it; devices from these vendors will typically work with the "out of box" drivers that come with a recent Linux kernel.

I've been developing software for nearly 4 decades, with the past decade or so being primarily on Linux. I'm a little annoyed that I haven't contributed to any FOSS projects myself; fixing that is one of my New Year's resolutions.

Much more pragmatic series of posts than what I'd of expected just 5 years ago. In the lifetime of any movement that ultimately becomes mainstream, the hardliners get pushed to the outskirts, so this is progress. The next step is just admitting most users don't care at all about the legal status of underlying source code. A few years after that sinks in to the linux community, THEN we'll have the Year of the Linux Desktop.

For example, for a basic Xeon server I bought last month, I tried a bunch of different distros. Landed on unRaid. Easiest thing to manage I've ever used. Had to pay for a license, but I saved that much money worth of my personal time, and don't know if any source code is available -- and really don't care. It, along with Docker, works like a dream.

w76 wrote:

Much more pragmatic series of posts than what I'd of expected just 5 years ago. In the lifetime of any movement that ultimately becomes mainstream, the hardliners get pushed to the outskirts, so this is progress. The next step is just admitting most users don't care at all about the legal status of underlying source code. A few years after that sinks in to the linux community, THEN we'll have the Year of the Linux Desktop.

OTOH, Windows 10 really seems to drive home the point that you don't control your PC any more if you run a proprietary OS. Source code (or lack thereof) aside, this will push more people to at least give Linux a try. Yes, many of these people will end up going back to Windows; but some will stay with Linux.

I doubt we will ever truly have "Year of the Linux Desktop" though. At the end of the day, FOSS will probably have to be content with dominating in the data center.

just brew it! wrote:

OTOH, Windows 10 really seems to drive home the point that you don't control your PC any more if you run a proprietary OS. Source code (or lack thereof) aside, this will push more people to at least give Linux a try. Yes, many of these people will end up going back to Windows; but some will stay with Linux.

At least our base will be growing however slowly.
w76 wrote:

Much more pragmatic series of posts than what I'd of expected just 5 years ago.

Some of the zealots have grown up.

whm1974 wrote:

just brew it! wrote:

OTOH, Windows 10 really seems to drive home the point that you don't control your PC any more if you run a proprietary OS. Source code (or lack thereof) aside, this will push more people to at least give Linux a try. Yes, many of these people will end up going back to Windows; but some will stay with Linux.

At least our base will be growing however slowly.
w76 wrote:

Much more pragmatic series of posts than what I'd of expected just 5 years ago.

Some of the zealots have grown up.

Part of "growing up" is ditching the "ours" vs. "theirs" mentality.

At the end of the day everyone wants the same thing -- an OS that lets them use their computing devices to do the stuff they need/want to do.

Yeah functionality is king. People don't buy operating systems, applications, or even computers. What they do buy is functionality

It's a nice idea and concept. Democracy is also a nice idea and concept.

I look forward to when I can ditch glass-panels OS. Turns out glass is bad as keeping rocks out. Never though of Windows in that wording before.

I think that the focus on FOSS applications should be put to the back-burner to first start pushing a FOSS OS, or at least an OS where transitioning it to FOSS is within reasons. Once that is a decent sized market share resume a big push to FOSS. Otherwise to much of a chicken-and-egg problem imho.

whm1974 wrote:

Yeah functionality is king. People don't buy operating systems, applications, or even computers. What they do buy is functionality

Precieved functionality. Just had someone who moved from Brazil 10ish years ago talk to me about learning digital marketing and how he really needed a Mac to do digital design.

Perception is more important than reality on many ways. No matter how much I despise that statement.

I guess my reasons for using Linux are not in any camp, I switched to Linux for the following reasons:

1) Privacy was among the first reasons. I can't bring myself to use Win 10 because of all the telemetry and I stopped using Win 7 and 8.1 at home for the same reasons, mainly that MS keeps sneaking in added telemetry via updates. I know there has been telemetry since Win XP but it's really gotten our of hand now. Along the same vein, the fact that the MS EULA says that you give MS an irrevocable, royalty-free, world-wide license to use your content as they see fit. Along this same line I won't use Ubuntu proper because of that sending your Dash search queries to Canonical, I know it can be turned off but still just the thought is enough to turn me off to using it. When they release the new 16.x version, which they have promised completely removes this functionality, I will consider it.

2) I like the modular nature of Linux distros. With Windows it's not possible to remove the built in browsers, media players, desktop, etc, and now with Win 10 MS tries to push their own apps when you try to use 3rd party apps. I will say there are Linux distros, such as Ubuntu MATE, which while it's a really well done distro and fast as hell, it's way to tightly integrated in terms of the desktop and the apps it ships with by default. I tried to uninstall Transmission torrent client and was informed that in order to do that I would have to uninstall the entire MATE desktop. This is unacceptable and it leads to the reality that it's very easy to break your install to the point where it's easier to just reinstall the OS just by trying to customize certain parameters. Linux Mint is much better in these regards, better than the official Ubuntu variants.

3) This one is a very big seller for me, namely much better ram usage. The one thing that annoys me with Windows is that no matter how much ram you have it still hits the page file hard and while you can disable the page file Windows really starts running like crap if you do. Linux on the other hand makes excellent use of ram, monitoring both ram and swap usage I find that even at 1.5gb ram usage Mint has not started touching the swap file. At about 5gb ram usage it barely hits 500mb swap usage. This goes hand in hand with the excellent thread scheduler, one time I had no swap file enabled and the ram usage was at 7.5gb used out of 8gb total ram and Mint still ram like a champ. On Windows if you are using that much ram out of your total ram the system runs like molasses, to the point you can barely move the mouse.

4) I like the ability to theme most distros to your hearts content and you can install and try out various desktops or you can distro hop to see what's new. Along this same lime I like the fact that you can install Linux on a thumb drive as a "Live" install so you can try most distro without having to touch the hard disks.

I would like to see some changes to Linux, namely the ability to add drives in a fashion similar to Windows, so that hardware vendors could ship Linux drivers with their hardware. You shouldn't have to either hope the kernel your distro ships with supports your hardware, or update you kernel, custom compile a module and edit obscure config files to get hardware working. One thing I like about distros like the Ubuntu based one's is that they ship with everything enabled in the kernel and then blacklist hardware that's not present and updating the kernel is easy as pie. Still I would like to see the ability to simply download a driver, click install and have everything taken care of .

I think the biggest thing holding Linux back is the GPL, namely the newest version and the ideology that surrounds it. The GPL is for lack of a better term a parasitic license, like the Borg, it really seems to scare off many developers and companies and it's understandable.

This is why I don't think we'll ever see a "Year of the Linux Desktop", I think if any alternative OS takes over the desktop it will be a BSD based distro, like the excellent PC-BSD or DragonFly, which is also a very nicely done distro. The BSD software licenses are much more realistic in terms of being friendly to both end users and companies and it's why Apple chose that kernel for use to build up OSX.

NoOne ButMe wrote:

I think that the focus on FOSS applications should be put to the back-burner to first start pushing a FOSS OS, or at least an OS where transitioning it to FOSS is within reasons. Once that is a decent sized market share resume a big push to FOSS. Otherwise to much of a chicken-and-egg problem imho.

Err... what the heck are you talking about? The OS already exists. The problem is that there aren't equivalents/ports for all of the commercial Windows applications and games. This does not matter as much as it used to, with the move of applications to a "cloud" model; e.g. the Web-based version of Office 365 works just fine in Linux, since it is browser-based. But it's still the main issue holding back wider adoption.

sophisticles wrote:

I would like to see some changes to Linux, namely the ability to add drives in a fashion similar to Windows, so that hardware vendors could ship Linux drivers with their hardware. You shouldn't have to either hope the kernel your distro ships with supports your hardware, or update you kernel, custom compile a module and edit obscure config files to get hardware working. One thing I like about distros like the Ubuntu based one's is that they ship with everything enabled in the kernel and then blacklist hardware that's not present and updating the kernel is easy as pie. Still I would like to see the ability to simply download a driver, click install and have everything taken care of .

Yeah, I agree the driver model is problematic; but IMO this is a secondary issue. Lack of Linux ports of most commercial software is (and will likely remain) the big roadblock.

A lot could be done to make the driver situation better even without revamping the driver model. For example, the process of recompiling the driver (or the shim in the case of a proprietary binary driver) could be standardized and automated, allowing the sort of "one click" driver installation people are accustomed to on Windows.

I'm not a linux user but would like to be. You know how you can get that virtual PC software to run WinXP programs in Win7? And then there is VMWare which I understand can host Linux, Windows, etc. on a single hardware platform. Is there currently a virtual PC package for linux that can host my most used windows application (excel)?

just brew it! wrote:

NoOne ButMe wrote:

I think that the focus on FOSS applications should be put to the back-burner to first start pushing a FOSS OS, or at least an OS where transitioning it to FOSS is within reasons. Once that is a decent sized market share resume a big push to FOSS. Otherwise to much of a chicken-and-egg problem imho.

Err... what the heck are you talking about? The OS already exists. The problem is that there aren't equivalents/ports for all of the commercial Windows applications and games. This does not matter as much as it used to, with the move of applications to a "cloud" model; e.g. the Web-based version of Office 365 works just fine in Linux, since it is browser-based. But it's still the main issue holding back wider adoption.

Does it? How many people use Linux that is purely FOSS when talking GUI an what goes into what people require as the baseline for a modern OS. If that is fully FOSS good. I didn't understand that in that case. Thanks.

doesn't stop the push to move to grow the market share by using non-FOSS. Linux market share in the consumer space where it is FOSS at the core, given it is anything, is tiny. i'm not considering Android FOSS given my understanding of how much the core parts of it require to many services by a different company to really be "usable".

@Mr Bill -

VMware and VirtualBox are both available for Linux, and should be able to host any x86 (or x86-64) OS released in the past decade and a half, give or take. You just need a CPU that supports VT-x (Intel) or SVM (AMD), and enough RAM to comfortably run the host and guest OS(es) simultaneously. The main things to be aware of are: 1) Performance won't be as good as on a native PC, but for most office and software development use cases it is more than good enough; 2) 3D acceleration in the guest OS is somewhat limited, and not always 100% stable; and 3) you may occasionally run into glitches with some USB peripherals (but this has improved quite a bit over the past few years).

Running Word/Excel to handle documents that LibreOffice has trouble importing is one of the reasons I keep a Windows 7 VM handy on my Linux desktop.

NoOne ButMe wrote:

Does it? How many people use Linux that is purely FOSS when talking GUI an what goes into what people require as the baseline for a modern OS. If that is fully FOSS good. I didn't understand that in that case. Thanks.

GPU drivers are still a sore spot in terms of having a 100% FOSS desktop OS. But if you have an Intel GPU, or an older nVidia/AMD one, it is certainly possible, as many of these have reasonable FOSS drivers now.

My current home desktop is 100% FOSS at the OS/driver level. System is using an AM3+ motherboard (so not bleeding edge), and all onboard devices (audio, NIC, temp/fan monitoring, USB2/3, etc.) are fully supported in Linux out-of-box. GPU is an older Radeon HD 6xxx series one, also using the FOSS Radeon driver that came with the distro.

Proprietary application software I have installed includes Chrome (which has a FOSS core, but also includes proprietary code in the form of the integrated Flash plugin), VirtualBox (which is also mostly FOSS, but you need to install a proprietary binary plugin to get full USB virtualization support), and Altera's FPGA tools (for some consulting work I was doing).

I also have a copy of Windows 7 installed in VirtualBox, for those occasions where Windows is necessary for whatever reason. I spend very little time in the Windows VM. It probably spends more time downloading and installing patches than I spend actually using it.

Edit: Current laptop is a refurb "off lease" HP ProBook (model is about 3 years old, core i5 processor). Situation is similar to the desktop -- everything "just works" with the drivers that came bundled with the Linux distro. Even including sleep/wake, which has historically been somewhat problematic in Linux (this was a pleasant surprise).

Than JBI. I would still stick with my general point of pushing the install base of the FOSS based Linux variants that exist wider so there is a larger market before aiming to move to 100% FOSS. Get a market large enough where at least one major OEM sees a need to address it. Which is what needs to happen in the end I believe for it to truly succeed.
Thanks for the info again.

NoOne ButMe wrote:

Than JBI. I would still stick with my general point of pushing the install base of the FOSS based Linux variants that exist wider so there is a larger market before aiming to move to 100% FOSS. Get a market large enough where at least one major OEM sees a need to address it. Which is what needs to happen in the end I believe for it to truly succeed.

Major OEMs (HP and Dell) already support Linux, and offer it as a pre-installed option on some of their systems. No, it isn't offered across all product lines; but it is available.

The biggest issue for the majority of users remains availability of common commercial applications.

JBI: Your 6470b (if I remembered correctly from the other thread) is probably running the IGP (an HD3000), right? Not that HP had hybrid graphics at that point anyway, which helps here. As I recall, Ubuntu ran fine on a 6530b - which is just an older (Penryn) version of yours. Save for its own glitches with Unity and the file manager one day, of course.

And yeah, nearly all business machines I've seen support linux in some flavor or another.

As I've mentioned before though, I've still had stability issues with drivers. An HP 8000 elite with a geforce 210 failed miserably in linux with every chance I gave it (different drivers, different distros, never terribly stable), but ran Windows 8.1 totally fine. Though the machine ran CentOS fine as soon as I pulled out the 210 and ripped out the proprietary nvidia driver. Apps are the easy part in my particular case.

just brew It! wrote:

The biggest issue for the majority of users remains availability of common commercial applications.

If these applications were available, would there be a huge increase of Linux users.

whm1974 wrote:

just brew It! wrote:

The biggest issue for the majority of users remains availability of common commercial applications.

If these applications were available, would there be a huge increase of Linux users.

Well, it would certainly remove one of the main impediments to wider adoption. Would it get the Linux market share on the desktop above single digits? Impossible to say; but my gut says probably not. Wider availability of commercial applications (and major game titles) would, however, help shift the perception that desktop Linux is something that only computer geeks use.

just brew it! wrote:

Wider availability of commercial applications (and major game titles) would, however, help shift the perception that desktop Linux is something that only computer geeks use.

That perception hasn't been true for some time now. I've gotten my dad and a friend to use Linux and both them love it.

whm1974 wrote:

just brew it! wrote:

Wider availability of commercial applications (and major game titles) would, however, help shift the perception that desktop Linux is something that only computer geeks use.

That perception hasn't been true for some time now. I've gotten my dad and a friend to use Linux and both them love it.

The perception still exists among the vast majority of the general public. There are isolated pockets of people who have been convinced otherwise.

Speaking of which... among the ads that were shown before the previews at the showing of the Star Wars movie I attended this evening, there was a (very) brief ad for Red Hat. It was just a image with their logo, and text along the lines of "Red Hat - Tested, Reliable, Trusted" (I may be mis-remembering the exact words). I was very surprised, but pleased to see some Linux advertising in a mainstream venue, even if the word "Linux" was not explicitly mentioned. And also more than a little puzzled by the advertising strategy; Red Hat isn't even aimed at the consumer market. Maybe the reasoning is that a lot of IT managers are Star Wars fans?

just brew it! wrote:

NoOne ButMe wrote:

Than JBI. I would still stick with my general point of pushing the install base of the FOSS based Linux variants that exist wider so there is a larger market before aiming to move to 100% FOSS. Get a market large enough where at least one major OEM sees a need to address it. Which is what needs to happen in the end I believe for it to truly succeed.

Major OEMs (HP and Dell) already support Linux, and offer it as a pre-installed option on some of their systems. No, it isn't offered across all product lines; but it is available.

The biggest issue for the majority of users remains availability of common commercial applications.

It isn't offered to consumers as a viable option. And I'm saying offering FOSS Linux. FOSS top to bottom shipping and pre installed. That's the part I'm sceptical about.

Of course, Windows also ends up being a net benefit in cost for many cheap computers and development and marketing funds. And Intel's own development funds are being pushed towards features that are harder to ship working Linux easily.

just brew it! wrote:

You just need a CPU that supports VT-x (Intel) or SVM (AMD)

I see you keep posting this but it is not true at least for the case of VMWare. VT-x It is only required if you wish to run 64-bit guests on intel processors. For AMD it requires segment-limit support in long mode (found in Winchester and newer, well before SVM debuted in Orleans).

NoOne ButMe wrote:

It isn't offered to consumers as a viable option. And I'm saying offering FOSS Linux. FOSS top to bottom shipping and pre installed. That's the part I'm sceptical about.

I'd say it is much more important that they offer it as an option in the first place, even if there's a proprietary driver or two. A "foot in the door" sort of thing. Given my recent smooth experience installing Ubuntu 14.04 on an HP ProBook, I see no reason why they couldn't ship a 100% FOSS system; not sure if they do or not. But you can definitely get laptops and desktops with Linux pre-installed from major OEMs now.

Deanjo wrote:

just brew it! wrote:

You just need a CPU that supports VT-x (Intel) or SVM (AMD)

I see you keep posting this but it is not true at least for the case of VMWare. VT-x It is only required if you wish to run 64-bit guests on intel processors. For AMD it requires segment-limit support in long mode (found in Winchester and newer, well before SVM debuted in Orleans).

OK, it's a bit of an over-generalization. However, the last time I tried to use VirtualBox on a system without VT-x, the performance was so bad it was virtually (heh) unusable. VMware may very well be better; they've been around since before VT-x existed so they had to figure out how to do it efficiently without hardware assist.

Perhaps a more accurate (but wordier) statement would be "a CPU with VT-x or SVM is strongly recommended for performance and compatibility reasons, especially if you plan to use VirtualBox". Happy now?