Linux: Fighting the FUD of Forking

sebFlyte writes "Fighting the MS FUD machine is a full time job for some open source developers, especially now Microsoft have thrown in the issue of the possibility of Linux forking (as Unix did)... it would also seem that Gates has moved on from telling people to 'get the facts' and creating FUD around patents and IP to criticising the open source communty's ability to create interoperable software."

Comments

Microsoft and Interoperability ?

[Mauvaisours]

From the guy that brought you exchange server and MS office closed format.

Re:Microsoft and Interoperability ?

[altp]

And MS Works, which isn't compatible with MS Office out of the box.

Re:Microsoft and Interoperability ?

[Daengbo]

From the article, you'll see that "interoperability" here means with itself. Meaning: Windows works best with windows and other OSs don't work as well with it. This is somewhat true, but OSS's strength is operating with every OS and every arch under the sun...

This all goes hand-in-hand with Samba's impending complaint over MS's licensing agreement in the EU dispute.

Re:Microsoft and Interoperability ?

[tomhudson]

Their strange view on interoperability:

For example, interoperability is sometimes viewed merely as adherence to a published specification of some kind, either from one or more vendors or a standards organization. But simply publishing a specification may not be enough, because it overlooks much of the hard work it takes to successfully develop interoperable products – namely, ensuring that the "contract" defined by a specification is successfully implemented in software and tested in a production environment.

No wonder they're always breaking specs.

The whole article is a puff piece. Even the above-quoted sentence really doesn't say anything.

But I do have to admit Microsoft is way ahead on interoperability - many more viruses and trojans "just work" with their systems.

Anyone who believes this mindless pap deserves what they get.

Re:Microsoft and Interoperability ?

[beh]

I don't quite see the problem here.

I am not afraid of forks, if they are executed well.

Look at some examples we've had in the past:

gcc fork - when the gcc development started to slow down, a new group forked it and the primary thing it did was to speed up development.

emacs fork - emacs had had a notice for ages saying that "X11 support was coming RSN", but nothing happened for quite a while. The Lucid-Emacs (later became XEmacs) happened and within a very short amount of time there was quite a hustle and bustle of activity between the two - Yes, there are some interoperability issues here in that both designed their respective GUI concepts a bit differently. But both evolved at a much quicker pace then if we only had one. (Especially good in this case, was that the lucid/xemacs team decided that sticking to old packages like the age old c-mode wasn't a good thing and that there were better alternatives to be used, and they didn't shy away from using them - much to the advantage of the entire community.

If there should be a linux fork, I am not really afraid of it, since those who will fork it, will know that they will also NEED interoperability (an issue that emacs/xemacs didn't really have in that sense, as the files you edit with them ARE interoperable -- and I don't think a linux fork that will make the formats of binaries / shared libs different, will find much acceptance, unless they also manage to continue supporting the old formats as well (pretty much like you can still use a.out binaries, if you still have the kernel support for it compiled in).

I don't think we should just have a kernel-fork just for the sake of it - but if there are good reasons for a fork, I am not afraid of it - in fact, I'd rather welcome it.

Benedikt

Re:Microsoft and Interoperability ?

[Lodragandraoidh]

Gates is speaking to CIOs here. The writing is on the wall, and the change of tac, I believe is a last ditch effort to stave off the latest Apple attack (think about what most business desktop systems do, and look at the new Apple Mac Mini - for as cheap or cheaper price you can upgrade your aging windows ME and 2000 machines to something that just works - without all the problems with viruses and trojans; the added benefit is you can gather up all those old PCs and build a beowulf cluster for number crunching/modeling - a win-win situation, if there ever was one).

Of course, Gates will be happy to put the fear of god (or in this case, interoperability) in the minds of the people who make the decisions to buy or not to buy. If the CIO is not a computer guy - then he might just buy this latest broadside....

Re:Microsoft and Interoperability ?

[beh]

I would go further and say even those examples show the most important characteristic:

The push ahead innovation - in this case, on the OSS side. Of course, in a sense it wastes development time, but on the other hand - when are you more motivated to code? When you're working on something nobody else in the OSS world is working on - or if you're working on something that has competition and you want to show off that your piece of software is better...?

Re:Microsoft and Interoperability ?

[Master of Transhuman]

Nobody said that EVERYTHING is going to break on Longhorn.

But enough of it is going to break to make switching a pain in the butt, you can be sure about that. But not so much is going to break that NOBODY is going to switch.

The stupid large corporations are screwed anyway, because they have vendor lock-in due to their unwillingness to train anybody to use another OS, so they'll buy Longhorn regardless of the expense and conversion problems.

Small businesses, OTOH, have somewhat more flexibility to switch to another OS or keep using the old one. This varies by business since some businesses don't want to train or convert either.

It took three years for most people to upgrade from Windows 2000 and 98 to XP because there wasn't enough reason to do so (from 2000 anyway). Microsoft doesn't want to repeat that mistake. ALso they want to differentiate from Linux more strongly. So this time the OS will be VERY different - which will break things.

Microsoft doesn't care because they have forced the corporations into a licensing scheme that pretty much forces corps to upgrade every three years or lose money on the deal (even though they've already lost money since Longhorn is late - a major corp complaint.)

However, if the hardware upgrade requirements are as reported, Microsoft could find itself in deep crap. Which is probably why they dumped WinFS (which, BTW, is a feature they've been promising for about the last ten years - and haven't delivered on yet). I expect to see Avalon reduced in functionality over the next year as well - with the result that Longhorn will end up being just a different version of XP with some new eye-candy - and Microsoft will be back where it started with no one bothering to upgrade.

The bottom line: Windows is now so bloated and so screwed up that even Microsoft can't change it effectively.

Re:Microsoft and Interoperability ?

[Stephen Samuel]

Longhorn isn't going to 'break' compatibility with current software. Your old programs are still going to work on it...

Well, kinda... But if you want it to work properly, well sell you a new version of your (otherwise perfectly working) software for a mere $500 a copy.

Microsoft thrives on non-interoperability. You remembe the debacle of word'97? It couldn't save properly in word5 format. Once you bought one copy of word '97 you had to upgrade every copy of word in your company or deal with unusable copies of various documents interrupting the work flow all over the place.

(yeah.. they fixed that problem a year later but by that time, most companies had paid Microsoft the billions of dollars in upgrade fees, which was the entire intention.
(it might have been word '95 that did this, but you get my point)

In any case, Longhorn is going to be different enough from current windows that it's probably going to be just about as nasty (and expensive) to 'upgrade' to the arbitrary restrictions of Longhorn as it will be to upgrade to Linux and Open Software.

Linux forked a long time ago

[castlec]

Each distribution has typically has its own fork. The glory of the GPL rings true here. No one can be hurt from a fork. The better code, how ever one wishes to evaluate better, will live on. As others have already noted, "Nothing to see here. Move along."

Re:Linux forked a long time ago

[PornMaster]

Each distribution may come from a kernel with a set of customization patches, but aren't they all applied to mainline kernels? That's not quite the same thing as forking.

Re:Linux forked a long time ago

[MarkRose]

No one can be hurt from a fork.

Wrong.

I thought only BSD has forks...

[Anonymous Coward]

.. well at least that demon mascot does.

Re:I thought only BSD has forks...

[greed]

.. well at least that demon mascot does.

That's a daemon, damnit!

Ummm... I don't see how or why

[agraupe]

I think he's trying to make the point that someone could make a fork that is incompatible with all linux programs, or something like that. It doesn't make sense to me, because such a product would only be the result of its creators having a strong urge to shoot themselves in the foot.

Linux distros *are* forking

[defile]

I've ignored Red Hat and SuSE for about 5 years now, focusing mainly on Debian, Slackware, Gentoo, etc.

Now that I've used a Red Hat system again, I was completely dazzled by how drastically different the experiences are. I expect the GUI to be more polished, naturally, but so many underlying things are different as well. All in all, they're things I can learn, and binary and source compatibility are still there, but it's the trend that's disturbing.

All of the traditional UNIX vendors forked in order to raise the barrier of exit for people who wanted to switch platforms. Sun's platform is still alive today because Solaris is such a unique beast that you have administrators trained solely in the art of this platform. All the UNIX part does is allow for some kind of source compatibility. Maybe.

Cisco took TCP/IP, which was practically invented (and perfected?) on a BSD box and threw it away to build a new proprietary OS to run specifically on their routers.

It's hard to find a major distribution shipping the vanilla kernel these days. When does, for example, SuSE decide that binary compatibility with other distros is keeping them from "enhancing" the user experience? Can they resist?

I'd like to be wrong about all of this.

Re:Linux distros *are* forking

[John Fulmer]

It's posts liket his that make me wish there were a '-1 Bullshit' moderation tag.....

Patches added to kernels != 'forking'.
Different software subsystems != 'forking'.
Different methods of hardware detection and setup != 'forking'.

If that's true, then Linux forked in the early/mid '90s, because Redhat used a more SysV-like bootup system, and SLackware used a more BSD-like bootup system.

Distros have always had sometimes significant differences between them. I've never, however, had problems getting things to run between distros, except for maybe library differences and versioning hell.

And the first distro that removes 'binary compatibility' will simply go away, because at that point it will cease to be 'Linux'.

The "linux won't split" article said it best

[mrjb]

The whole nature of open source is based on interoperability. It is this very nature that made the Internet possible. Where standards are nonexistent, they are being created; for instance, look at the Jack Audio Connection Kit that allows all Linux audio applications (that support it) to interconnect. As a result, developers do not keep reinventing the wheel all the time; instead, they learn how to work with the provided interface, and just build what does not exist yet.

Regarding "fighting the FUD machine"...

[PornMaster]

Does wider adoption benefit the developers of OSS, or would they be better spending their time working on the software than fighting FUD?

(I mean this as a serious question, not trolling)

Re:That's rich...

[MoonFog]

And it's also a convenient coincidence that they talk about this after recently releasing Office 2003 which supports saving the file as XML, and releasing the DTD giving others an option to be compatible with this XML format (albeit the license is somewhat incompatible with the GPL).

Fud Fighters

[breakbeatninja]

I think it's important to recognize that Microsoft, SCO and other like minded companies will do whatever it takes to distribute harmful and baseless propaganda in order to further their cause (monopolization of desktop and server markets, proprietization of media and so on). The fact is, while each distribution has minor differences in the way userland and package management is iemplemented, the fundamental Linux kernel is the same and works across all of the distributions.

As we've seen in previous anti-Linux efforts on Microsoft's part, this is another effort to steer current Microsoft users away from Linux that may be considering it to lower licensing fees and hardware overhead. We all know it takes a *lot* more sysadmin time and monetary investment in hardware and software to reach the same results with a Microsoft-based workstation or server vs. a Linux or Unix equivelent. While Microsoft's sales are strong, their propaganda efforts show some desperation and fear.

While open source developers may spend a lot of time battling Microsoft's rhetoric, I think it's more important to concentrate on creating a solid operating system for everyone, from the hobbiest to the corporate user. The best way to beat Microsoft at its own game is not to play it. That is, Microsoft seems to value marketing and scare tactics over actual development and innovation. Let's not let Linux fall in Microsoft's trap of smoke and mirrors.

Re:Fud Fighters

[Anonymous Coward]

While open source developers may spend a lot of time battling Microsoft's rhetoric, I think it's more important to concentrate on creating a solid operating system for everyone, from the hobbiest to the corporate user.

It's easy to agree with the principle behind this, but one reason that Linux can pull in the support it needs to make it a great OS for all is that it is seen as having the potental for making a significant impact on corporate use. Without this potential you wouldn't find companies like IBM investing so much in improving linux, or companies like Oracle porting their software to it. If Microsoft et al. win their FUD war then corporations aren't going to pick up linux, and its not going to attract the kind of investment that benefits everyone that uses it.

Luckily, however, the linux community is diverse and it can provide people to battle the FUD as well as kernel hackers to improve the platform. We should support them all.

Re:Fud Fighters

[Tim C]

I think it's important to recognize that Microsoft, SCO and other like minded companies will do whatever it takes to distribute harmful and baseless propaganda in order to further their cause

While that's true, there are two other things to keep in mind as well:

1) just because MS, SCO or whoever says something, does not automatically mean that it's FUD, and dismissing everything they say as such is foolish and dangerous

2) there's a fair amount of FUD generated and repeated here about MS and other such companies and their products; perhaps that's fair, perhaps it lowers us to their level. Personally, I lean towards the latter.

We all know it takes a *lot* more sysadmin time and monetary investment in hardware and software to reach the same results with a Microsoft-based workstation or server vs. a Linux or Unix equivelent.

See, here's an example. I've used Linux (Mandrake mostly, but also RedHat and Slackware) and Windows (9x, NT, 2K, XP) as my main desktop at various times over the last few years. In the hands of someone who knows what they're doing, it absolutely does not require "a *lot* more sysadmin time" and money to get the same results. If anything, for the tasks I perform daily (general computing use and programming, etc), Windows just beats Linux, but only because of software. I am required to use my company's Exchange server for calendaring, and so Outlook is a must. I can run Windows under VMWare and still use Linux for everything else, but that ups the hardware and admin requirements. If not for that, and the need to edit Word docs, I could use Windows and Linux interchangeably. (Note that OO is not an option. If Word messes up a client doc, that's one thing, but if I choose to use OO and it messes it up, it's my neck on the line)

MS fails to grasp a simple idea once again

[Weaselmancer]

The reason why forking isn't a problem is because the open source community knows how to read the friggin' RFCs before we code something. Unlike a certain software giant who lives in Redmond.

Doesn't matter if there is one branch of a big project or 1000 forks. If they stick to specs, they are all interchangeable. Like your window manager. As long as they do what they're supposed to do, stick to specs and play fair - it doesn't matter which one you use.

This gives the user choice, which is why MS finds it to be such an alien concept.

There are 2 types of forking

[Xpilot]

The first kind is where each version of the software is slightly different, yet ABI compatible with one another. That's what the Linux kernel is, and Linus and co. have tried hard to maintain this. In essence, every time a developer sets up his own tree, it's a "fork" of the Linux kernel, but that's ok because binary compatibility is still maintained, and those changes will probably be merged back upstream anyway. Good news all round.

The second kind is where a substantial group of developers get into a messy political argument and take the codebase in a wildly different direction and becomes a new project in itself. This isn't necessarily a bad thing either, as you'll see cross-pollination between projects (like in the BSD's). However this may be what the FUD-mongerers are hinting at. I have yet to see any signs that this will happen though - it's downright impractical to fork the Linux kernel in a wildly incompatible fashion with the rest of the developer community - for one thing, there's a whole shitload of drivers you now have to maintain yourself. Not an easy job.

As for distros being different...well it's always been this way. Yet Linux's growth has been phenomenal, and with efforts like the LSB in place you won't find that distros diverge too far from one another.

Things look bright for Linux, any way you go. Don't listen to the FUD mongerers.

Re:Just like politics

[grozzie2]

Win 95, Win 98, WinNT, Win2K, WinXp. Looks like a hell of a lot of incompatible forking to me....

Forget forking!

[bigtallmofo]

Linux developers need more spooning.

Life in the ecosystem forks

[ch-chuck]

Funny how someone who talks a lot about the software 'ecosystem' wants customers to invest in this one dinosaur - instead of being amazed at the natural process of species differentiation and survival of the fittest.

Forking, interoperability and FUD

[bernywork]

I think this is brillant, couldn't have come from a more knowledgable person at a better time. Especially given that in the past day or two a nice little article got published up on Groklaw about the SMB / CIFS protocol and what legnths they have to go to, to reverse engineer / pull it apart on the wire. It's essentially a slightly intelligent brute force method.

Take a look. I couldn't have made the timing for this article any better if I tried.

http://www.groklaw.net/article.php?story=200502050 10415933

I second Tridge's motion that when Microsoft really wants to come to the party on interoperability, let me know. I want to be there.

Personally, I think the major reason why they are going through what they are doing for interoperability now, it's all because of market pressure with the rise of open source, and the open standards which it follows. See what's happening with all the governments demanding open standards for documents etc?

*sigh* when will they learn?

Distribution forks are killing Linux

[Anonymous Coward]

The problem isn't the forks. It is the install procedure that each fork will require to run the same piece of software.

With windows, you download a program and double click the install button. It doesn't matter if you are running Windows 98 / ME, NT, 2K, or XP. The thing installs and (sometimes) suns correctly. Try downloading a package (NOT SOURCE) built for some old version of RedHat and installing it on a new Slackware distribution. It just plain does not work by default.

This is what Bill G was talking about and I agree with him. There is no "standard" in the distribution. Where should the program go when it gets installed and does the distribution that it is getting run on grant the required privileges to install it there by default?

The Linux community needs to agree on a few things, one of the basics would be userland programs and config scripts. Until then it will continue to be it's own worst enemy as people won't be able to call themselves Linux admins, they will only be able to call themselves Distribution admins.

Re:Distribution forks are killing Linux

[Zphbeeblbrox]

I'm sorry but I have to disagree with you here. There is already a "standard" of sorts. /etc will work when nothing else will. /usr/bin also will work when nothing else will. The real problem here is that the developers of applications pretty much leave this kind of thing up to the distributions. If developers took the time to come up with one click installs for their apps then people might not be so dependent on the distro's The distribution should not be responsible for solving the install issues for their apps. That is the developers problem. When Gnome or KDE offer a one click install for their product then standardization will come. Until then each distribution will continue to offer their own "unique" way of doing things.

Re:Distribution forks are killing Linux

[ultranova]

Try downloading a package (NOT SOURCE) built for some old version of RedHat and installing it on a new Slackware distribution. It just plain does not work by default.

Of course it doesn't, any more than simply copying an install dir of a Windows app from one machine to the other will work. If you refuse to follow the normal installation routine of software for your system, the program doesn't work.

The normal installation routine for a Linux system is as follows:

1. Download the source package.
2. Untar it.
3. Give the following command:"./configure ; make ; su ; make install ; exit".

This has worked for almost every program I've ever installed. No programming skills needed, no need to have the slightest idea of what the commands at the last step do. Just memorize and type that litany. Shouldn't be too hard for anyone (and shouldn't be too hard to make a program that does it for you with a single mouseclick).

Sure, you need to download libraries sometimes. I've often hunted DLL's for Windows around the net to get some program or another to work. You want to avoid this, use your distributions package manager. You don't want to use automation, you need to install any missing pieces manually (which is usually no more difficult than reading the error message from configure, typing it to Google, and downloading and installing whatever comes up).

Simply because something is distributed in source format doesn't mean that you'd need to know anything about programming to get it to work. The source is for friendly neighbourdir compiler to read, Joe User doesn't need to care about it. After all, Joe doesn't need to know anything about compression technology to unzip the ZIP files that Windows programs are distributed in either, so why should he know or care what that "gcc" is that gets run when the install command is executed ?

Re:Just like politics

[conteXXt]

here an idea:

pick one (use any criteria you desire), learn it.

Pick another one, learn the differences.

Try a third, more difficult distro (pick gentoo)

learn it inside and out.

Give yourself a long break from Windows (a month should do it) now go back to Windows and hate it properly.

Then never ever ever again ask What is the best distro..... as the answer will always be Gentoo anyway (ok ignore that last bit :-)

Forking is aweful.

[Phat_Tony]

I think he knows what he's talking about here, forking is really awful. Over and over again, I've seen people with NT or 95 buying games that only run under 2000 or XP, or the newer forks like 98, ME, XP, or 2000 failing to run software from the older forks like NT, 95, or 3.1. One of my nephew's favorite games under 95 wouldn't even run under 98. It's really confusing for customers too, especially now that there are things that still say "Windows" like CE, but that run entirely different and mostly incompatible software. My Mom ran NT, and several times bought software that wouldn't work on that fork, which was so different from the concurrent "95" fork.

Anyway, I'm pretty sure that Bill would know what a pain in the ass it is for an operating system to have a bunch of divergent and not always compatible offerings available.

Re:Just like politics

[RicJohnson]

This is my BIGGEST problem with slashdot
Why is it every time I admit I am a Microsoft guy, I get modded down?
I really WANT to learn Linux. Can anyone help me understand Forking, or is everyone on /. a 1337 that has no time for a gnuB.
There are some helpful comments here I know can help my MS peers get off the ground - if only SOMBODY could mod the parent up!

Re:Just like politics

[jander]

I am not sure if you are trolling or if you really are missing the point, so I will give you the benefit of the doubt.

Lets take two scenarios:
Scenario 1: You write a killer app. You keep it a closed source model, and some greedy big company decides it likes your software and wants to make money from it. Whats to stop them from doing a little reverse engineering, decompiling, and then "adapting" it to make it look like they wrote it? It's not very hard to do, really. And, since your source is closed, the only person that can really prove that SomeBigCo stole your software is you.
What is your recourse? You sue SomeBigCo for copyright infringement. However, you have to prove that you wrote it in the first place, which may or may not be an easy thing to do. SomeBigCo can also throw lots of lawyers at it and it will plod through the legal system. If your lucky and you win, you may see some money in a decade or so. In the mean time, you will need to spend your hard earned money on legal fees.

Scenario 2: You write a killer app. You open source it and use the GPL for the license. You are making money not on sale of the software (although that could be done), but on service, support, and perhaps customizing the software for others. Others have seen your software, liked it and have sent helpful patches and features to you, thereby enhancing your product. SomeBigCo sees it and wants it for their own. They have a choice then - they can either follow the GPL and release any changes as open source (thereby leveraging their developers to your effort), or they adapt it and try to make it their own.
If they release their enhancements as opensource, that is fine because you can incorporate it into your source tree. Since you are making money off service and support, you still have a revenue stream off of your software and that will not diminish by SomeBigCo helping your development.
If they try to make it their own, then you still have copyright law to enforce the copyright of your software. Only this time, since it is GPL, you stand a good chance of getting the Free Software Foundation's lawyers involved, rather than paying for your own. Additionally, a company may have no qualms about stealing software from an individual or even a small company, but only the foolhardy would risk a GPL lawsuit.

How does this protect you from "Open Source Vultures"? By using a service and support model, you will have a steady stream of income regardless of whether you are doing the development. No matter what, if SomeBigCo decides to try and steal your software, you still will need to litigate the issue. However, with a closed source model, your legal expenses will be your burden. Also, don't forget you will need to pay a lawyer to write your license for you, and if not written correctly it could allow SomeBigCo to take your software through a legal loophole. The GPL is a very stong license, and although it is not fully tested in court it is strong enough to considered formidable by the Nazgul of Armonk (IBM's lawyers).

Finally, think of it this way - writing software is very much like the pursuit of scientific research. The very act or reading this is a culmination of work and ideas created by others, all built upon one another to create this gestalt called slashdot ;-)