The consequences of Open Source Arrogance

Ten years ago, what Free Software has today did not exist.

Free Software has advanced significantly and is expanding into every software application area. This is no surprising since Free Software started out from vitually nothing in 1984.

Proprietary software has a lead, but the gap is getting narrower all the time.

Microsoft does have a weapon--Bill Gates saw the need of it about ten years ago and set up a Microsoft Research, and hired the best minds in many fields from universities to work there. Some of Microsoft's new technologies (and problems for Free Software) today come from there.

So Free Software needs to have involvement/contributions from the research community to advance to new technologies if Free Software is to take the lead. Free Software needs People with PhDs.

But for existing technologies, Free Software is progressing right on schedule. GUI-type applications are being developed rapidly. I don't see any "consequences" of "arrogances" that you are talking about.

the gap is not getting narrower in the slightest little bit.

there is ZERO funding of DCE/RPC development in the open source arena. DCE/RPC is a key component of Windows NT services, and there is approximately fifty to one hundred man-years of development gone into DCE/RPC-based Windows NT services.

playing catch-up is easier than doing the work in the first place, so the catch-up factor is about one fifth to one tenth of the original development time, with the right people and the right expertise.

guis are useless unless you have the core infrastructure and the right attitude.

you are falling into the same mistake that got netscape, aol and the u.s. dept of justice into trouble: GUIs and browsers and pretty little widgets do not make an OS.

advise you to read article "brief history of windows", just posted.

I don't see any evidence of either ignorance or ego (wrt technical superiority and Microsoft) in the free software developer community. OTOH, I've encountered plenty of arrogant users...

I really have no idea what you're talking about. Like, my impression is most free software developers worship Microsoft's technical output. Compare user interfaces, .Net, etc. Imitation is the sincerest form of flattery!

Have you got any good examples of ignorance and ego? Sure, there are some decisions Microsoft make based on marketing rather than technical principles, and their products contain lots of flaws as a result. But that's not to say they don't have lots of smart people doing smart stuff that we can learn from, nor that our own software is as good as theirs. I didn't think anyone in the developer community was suggesting to the contrary.

yes, i do.

1)

i was asked [paraphrased] "why, when there are perfectly good open standards available and being developed, do we, as open source developers, need to reinvent the microsoft wheel?"

the response to the question was a standing ovation to the questionee by the other attendees of the bristol conference.

2)

i offered, eighteen months ago, to donate the samba tng code, 100,000 lines of code, to the Apache Software Foundation. the reason that it was rejected was given as "microsoft do not follow open standards. we do open standards, therefore we cannot accept your code."

the fact that microsoft use open standards and embrace and extend them, in particular DCE/RPC and the NT domain suite which is what samba tng is all about, seems to have missed them completely.

to be fair, the other reason was that they were afraid of lawsuits from microsoft, even though those lawsuits would not succeed. that lawsuits could have been avoided by isolating the security components in a similar fashion to both openssl and also pgp seems to have escaped the apache software foundation committee's attention.

yes, we do know that microsoft has lots of smart people.

however, the amount of ignorance regarding what microsoft is really doing, and in particular how their network services work, is quite staggering.

maybe the ego bit is because i tend to really drive people up the wall and they get all egotistical and defensive because of me, rather than generally being big-headed.

there's quite a lot of history and stories that i really can't tell you about, it's not the right time to outline them yet.

maybe when the time is right, it will be too late anyway: the opportunity will have been lost due to patents and due to information-restrictive laws such as the DMCA being passed world-wide.

btw you are absolutely right about the marketing-led decision making, and also some of the technical flaws that i know of are historical and some are ignorance regarding development of network applications: they all make for some pretty funny stories.

I think I have asked this before from you (with no answer), but what does "Microsoft's incredible technological lead" consist of? And please do not answer in acronyms.

To me, Microsoft is marketing. Microsoft's invovation is figuring how to get the same custumers to buy the same software year after year with no obvious improvement. Microsoft's software is not the tool of the user, but the tool of Microsoft to expand its profit margin, and expand its market. The needs of users are purely secondary.

The best strategy for free software is no strategy at all. As long free software expands to fulfill the needs of its users, the most basic function of sofware, I see no problems. The main lack of focus I see is precisely where free sofware blatantly copies the software of Microsoft, Apple, et al. When Microsoft has defined your goals you have already lost the battle. To take it a step further, when Microsft has defined your goals you are most likely working in favor of Microsoft.

atai, The history of free software does not begin and end with GNU, although I'm sure RMS would like everyone to believe that.

The notion of a techinal lead is meaningless without a context. In some contexts free sofware from the research community is state of art. In any event, how much research does Microsoft actually do? Can it really compete with all the public universities of the world?

Also, to what exent do the day to day tasks acomplished by users require bleading edge state of the art research? Especially for the tasks acomplished with Microsoft's software. The idea of this article is ridiculous from start, that Microsoft poseses technologies esential to the success of free software. It is an argument expected from Redmond Washington, not from the from page of Advogato.

Someone sees an open source zealot, and says "Open source will never get anywhere, if it only has zealots to offer!" So in the same way, lkcl sees a specific project which wants to focus only on open standards, and generalizes it to the whole open source scene.

There are a number of projects out there which deal specifically with Micro$oft-0wn3d file formats and protocols. For instance, AbiWord. Apache has decided to focus on open protocols, but that is their choice alone.

And they have the right to make that choice too. If lkcl starts a game application project, and another person comes by and demands that the project change its aim to that of developing a web browser, is it not reasonable that the second person gets rejected?

In a world where Apple can come out with mostly-1987 technology, break a few really good features (wasn't NeXTStep freakin' resolution independent?), give it a cute look and call it "modern", "innovative" and God knows what else, who needs the real innovation?

Why are most widget sets, on whatever OS, still flickery? Avoiding flicker has been done by videogames for a long time, so what's the hold up on? Are they waiting for my eyes to BLEED?

Why do I own a computer that can run "Quake III: Arena" well enough for me to beat the crap out of people on the Internet, but that is apparently too slow for some web browsers?

How is that possible?

I'd agree with the fact that, while they might not do as much research as all of the public universities of the world together, they are right up there with any one of these public universities. They have some pretty good ideas, they just mess them up pretty bad most of the time when it goes from "research" to "product" (examples that I'm familiar with: COM, DCOM, COM+, all excellent ideas and concepts, and all mostly a huge pain in the ass to use, particularly the first ones).

For example, I could tell myself that in-process only was the best thing for components with reasonable performance (I know, that's what I did in XPLC), and sure thing, COM+ (used as the component technology in .NET) is in-process only. Same with so-called "XCOPY deployment".

NeXT workstation screens are postscript devices :) :)

COM on win95 is very badly implemented (see brief history of windows article).

COM on NT implementation is much better, is based on DCOM, which in turn is based on DCE/RPC. COM on win95 was a back-port of DCOM to NT with the DCOM bits ripped out.

tk, your attitude is a classic example of what makes it possible for open source people to excuse themselves from paying attention to the critical technology areas that they've missed out on.

much appreciated if you could help by listing some projects that do and don't stick to standards, proprietary or open, we'll go from there, yes? debates are about all sides: i happen to have some specific knowledge of a specific critical technology area that the open source community has COMPLETELY IGNORED.

mr slicker, you got an answer: you may have missed it, i am sorry. i will try without the acronyms.

answer: microsoft gives customers _just_ enough to get work done, and raises the bar of customers expectations with every year that passes.

i.e. most customers can actually, with enough effort and care, get work done. if it crashes 20 times a day, they pay someone to sit there and nurse the machine until the job is done.

what do they care about reliability if the end-results' standards of the work done are high enough to fool the users into thinking that they have achieved something?

and microsoft is constantly raising the bar, and the open source community HAS to reach and exceed that bar because otherwise it ain't gonna get rid of the "linux ain't ready for the desktop" stigma.

additionally, one of those customers was, in the past, the worldwide intelligence community. with the assistance of that intelligence community, they added in a decent security model (the likes of which is impossible to put into open source operating systems such as linux and *bsd because they have posix bound in at the base level: GNU/Hurd, containing a MACH kernel, is the only exception that i know of).

that is a short glimpse into part of the possible answers to your question: i hope that this helps and also that it answers your question satisfactorily, and also that you see and read this answer.

It's not attitude, it's facts. Please read my reply really carefully once more.

And speaking of history: How much do you know about the history of the GNU Project and

It's not attitude, it's facts. Please read my reply really carefully once more.

i happen to have some specific knowledge of a specific critical technology area that the open source community has COMPLETELY IGNORED. The onus is on whoever said this to prove with hard evidence that this knowledge is indeed "critical". And the Linux community has functioned very well without Microsoft's so-called "critical" services. (Again, this is not an opinion.)

And speaking of history: How much do you know about the history of the GNU Project, the Linux kernel project, and the Linux development model as outlined in Eric Raymond's "The Cathedral and the Bazaar"? Have you actually read the works of the luminaries behind them? It's surprising that `outsiders' of the open source scene are clamouring to get rid of the very factors that make open source tick.

What kind of OSes are mainly used on servers?

What does mean « technological lead »? Does it really means leading by technology of leading by commercialization?

mslicker said "The history of free software does not begin and end with GNU, although I'm sure RMS would like everyone to believe that".

GNU's not dead, so we cannot tell if RMS would like everyone to believe that free software end with GNU. But can you give links that confirm that Richard want everyone to believe that Free Software started with GNU?

I can give you a link that infirm that:

" When I started working at the MIT Artificial Intelligence Lab in 1971, I became part of a software-sharing community that had existed for many years. Sharing of software was not limited to our particular community; it is as old as computers, just as sharing of recipes is as old as cooking. But we did it more than most.

The AI Lab used a timesharing operating system called ITS (the Incompatible Timesharing System) that the lab's staff hackers (1) had designed and written in assembler language for the Digital PDP-10, one of the large computers of the era. As a member of this community, an AI lab staff system hacker, my job was to improve this system.

We did not call our software "free software", because that term did not yet exist; but that is what it was. Whenever people from another university or a company wanted to port and use a program, we gladly let them. If you saw someone using an unfamiliar and interesting program, you could always ask to see the source code, so that you could read it, change it, or cannibalize parts of it to make a new program. "

http://www.gnu.org/gnu/thegnuproject.html

You seem to be very angry and upset. You come here firing from the hip at a huge community with random allegations.

Sorry, but I doubt very many of us will feel hit by your criticism.

I'd advice you to turn around the discussion and instead provide "us" (the big ignorant crowd) with suggestions on what we should do to improve.

look at GNUstep Distributed Object ... http://gnustep.it/nicola/Tutorials/DistributedObjects/index.html

In a world where Apple can come out with mostly-1987 technology, break a few really good features (wasn't NeXTStep freakin' resolution independent?), give it a cute look and call it "modern", "innovative" and God knows what else, who needs the real innovation?

Well, Mac OS X is based on OPENSTEP, not NeXTSTEP, so it's not exactly a "1987" technology ... and IT IS modern and innovative, believe me. NeXTSTEP/OPENSTEP technology was years ahead anything else, and still beats current systems on many many points, and particularly in ease of programming. The only problem is that Apple is proprietary and, they seems to throw out the OPENSTEP way of building small cooperating apps (even the services menus is more or less hidden in a submenu now ! :-/ )

So look at GNUstep :) http://www.gnustep.org for a free OpenStep implementation :)

Luke, if you want to persuade people of your point you really need to be able to explain what the Free software community is actually missing by not working on DCE/RPC. Yes, we can see it's obviously a scarily huge task and that it would potentially be needed to talk to MS, and so on and so forth, but fundamentally: ok, so it's really big. What do we get for all that code? What problems does it solve? How much of the complexity is essential to solving those problems, and how much is accidental?

Playing catch-up with Microsoft for the sake of interoperability is about the last thing I want people to spend their time doing, because Microsoft can easily change direction before we're finished, and leave us nothing useful on their end to interoperate with. If the DCE/RPC stuff offers compelling technical advantages over other solutions or potential solutions to the same problems its aimed at (whatever they are anyway...) then I agree we should be taking it more seriously. Otherwise, ditch it, do something simpler, and let MS continue to pour gigabucks into maintaining their 3.5 million lines of code.

So, elevator presentation: What's the actual problem that DCE/RPC solves? You have 30 seconds. If you can do all the exaggerated sighs, cries of "you just don't get it" and eyeball rolling in only five seconds, you still have 25 seconds to talk in.

(Aside: it really does seem weird to be arguing that Worse is Better with somebody promoting Microsoft technologies...)

(Aside 2: tk, bagder - if you take a look at luke's advogato certs, that should provide at least circumstantial evidence that he might just possibly not be a complete newbie to free software)

lkcl, Perhaps your previous answer was as vacuous as your present answer. You still fail to communicate the critical techinical lead Microsoft possess. Why can't this security model be added to open source software? They have the source code, what is stoping them? What bar is Microsoft raising? It is expected that they add something every year, they can't sell exactly the same software. Why must the free software community duplicate the efforts of Microsoft? You have failed to communicate this. When you write article acusing the free community of arrogance for ignoring technologies of Microsoft, you should be expected to explain these technologies and why they are so crucially important.

rio writes:

Well, Mac OS X is based on OPENSTEP, not NeXTSTEP, so it's not exactly a "1987" technology ... and IT IS modern and innovative, believe me.

[1] http://www.forth.com/Content/History/History3.htm#3.1

mslicker said : This is just as weak as lkcl's reponse. All I believe to this point is that you must be enamored with all the shiny buttons.

Well, The OpenStep specification was released in 1994. so it's pretty clear for me that's not a "1987" technology -- Thoses specs are not compatible with NeXTSTEP's APIs by the way (even if they share some concepts and same OO-language, and some OPENSTEP implementations from NeXT were compatible with olders NeXT binaries).

When I speak of innovation and modernism, this is because OpenStep-compliant systems (like Cocoa or GNUstep) are fully object oriented, use a dynamic OO language (ObjC), that the frameworks are very cleanly designed, and there is the availability of Gorm (GNUstep) or Interface Builder (Cocoa) for designing GUI in a clean and fast way. It's possible to create distributed objects and sending messages to them as simply as standard objects, in few lines of code (see this tutorial ). When you localize an application, you not only could translates strings (in an extern file) but you could also create a GUI specific for the language if you want... On GNUstep, you could have an app wrapper (apps are distributed in form of "wrappers", aka a directory containing binaries, ressources, etc.) containing *multiple* binaries -- it's possible to distribute a GNUstep app which would work on GNUstep systems running on differents processors, because containing the differents binaries... etc. And I don't detail the great pastboard/services systems...

For me, *that's* innovative and modern, even if OPENSTEP systems existed back in 1994. And frankly, I don't care about the shiny buttons of OSX, but I really care about the programming framework / environment.

OpenStep and NeXTStep are so close to each others, they're like Windows NT 4.0 and Windows 2000, comparatively speaking.

NeXT had the futuristic OO language and component system that nobody else has yet (.NET is slowly getting there, almost 20 years later), they had InterfaceBuilder in 1987, you had multimedia email (like todays MIME/HTML messages) from the beginning, used Unix technologies (that were already old at the time, so Apple coming up with 30 years old technology and claiming its technical lead is rather funny), Tim Berners-Lee made the first web server and browser on NeXT in 1990, they had a distributed component architecture before it was cool, Doom and Quake were developed on NeXT machines (because they had the best development tools, mostly) and finally, had a highly componentized web browser, OmniWeb, at a time where a graphical web browser at all was ultimately cool.

Mac OS X added what to the mix, apart from a few more applications? This is not revolution at any rate, this is just normal evolution. They hid the Services menu, they made the display resolution dependent and threw the ultimately good usability out of the window (ever noticed that the "close window" button was away from the other button on NeXT, to avoid accidents, or the fact that they came up with the idea of putting the two arrows of scrollbars together, to minimize mouse movement?). They kept some things like the scrollbar designs, but the dangerous button is grouped close to the often-clicked safe buttons, and if I set my screen at the highest resolution that has a decent refresh rate, what I get is that everything is too small for me to see.

When someone does something that seems sucky, but that it's the best ever done, or that I can't think myself of a better way to do it, I cut back on the criticizing. If it's never been done, but I can think of a better way, I'll whine a bit, and try to hack it myself (like XPLC) or tell my idea to the person doing it. But sorry, if it's been done better before, you have no excuse and will be flamed accordingly.

To some I extent I understand his frustation. The media has taken to the word Linux, and has sought to promote Linus Torvalds as the center of the free software universe. Linus Torvalds does not seem to care about freedom or the principles of free software. In interviews he is simply incable reflecting the broader movement that is free software. He does not discuss the history, the international character of free software, the number of people involved to make it actually work, or the broader implications to society. RMS on the other hand, can go into depth and discuss all these issues. Perhaps the establishment sees the ideas promoted by RMS as a threat, and instead focuses on the relatively harmless "just for fun" Torvalds.

Still the GPL makes no provision for how you name software. To distribution makers use of "Linux" is no more than brand recognition. I don't see that getting distribution makers to change their distribution name, or to change the words people use, is either a winable battle, or even a battle worth fighting. Honestly though, I don't know much time he devotes to this issue. Perhaps it is others that make a bigger deal of the issue than RMS does himself.

rio, I might come back to your OpenStep post, I don't want to write off the concept. On the surface it apears to be the synthesis of Unix and Smalltalk. Both originate from 70's. The truely innovative features do not immediatley stand out to me, perhaps I need to dig deeper.

You said that while "He claims in one free software distribution 28% was GNU and 3% was Linux" he take that as an argument to call free software distribution GNU. But he doesn't. He take that as an argument to not to call free software distribution Linux.

"Is GNU or Linux really an appropriate name for this collection of software? Does GNU deserve credit for this 69%" : As you said after, this name is like a brand name. And so, it does not technically means that any software are GNU softwares.

The important point is the factor of cohesion. And personally, I think that the strong link between software in a free software distribution is following the GNU project goal (make a free unix), not the Linux goal, since as you said, Linux is philosophically meaningless.

You analyse well why Richard would like people to talk about GNU/Linux instead of Linux. So you perfectly understand Richard motivation on this subject and it still does not proves that Richard try to "impose himself in unreasonably ways".

A contrario, the linux and gnu text is a reasonable statement.

Note: there's at list one distribution called GNU/Linux.

What the heck is author talking about?

Ten years ago, in the age of Windows 3.0 and MSDOS 5, if a business user was presented with what Open Source has to offer today, then apart from it running like a dog due to not enough memory or disk space, they would be overjoyed and would abandon Microsoft products like they never existed.

Ten years ago people used SCO and not open sources stuff. In fact that's why some finish guy started hacking his own toy/hobby OS that later was picked up by IBM and some other large companies. Do you know what was it called? (if you do, you must be a master, otherwise back to journeyers)

there is ZERO funding of DCE/RPC development in the open source arena. DCE/RPC is a key component of Windows NT services, and there is approximately fifty to one hundred man-years of development gone into DCE/RPC-based Windows NT services.

Did Mr. Gates show you his payroll sheets? How did you conclude that there is 50 to 100 and not 500 to 10,000 years invested in com?

I was actually curious about that FreeDCE effort thing for some time, but seeing what kind of developers project has, I'll stay away until IBM will put some reasonable people on it.

It has been said many times already that advogato's credibility does not mean much, some individuals just enjoy proving it all over.

Stallman's argument for integration is weak. Distributions commonly fill up entire CDs, sometimes multiple CDs. This collection is either entirely sloppy or contains a great deal of redundancy. It is hard not to be aware of the redundancy in these collections, it exists in every catagory of software. This incredible redundancy is not a pattern of cohesion and unity but the pattern of great a number of independant developments. The Unix design facilates this independancy, the level involment required by GNU is surely negligable for this 69% . That RMS tries to fit all this under the banner of GNU demonstrates my argument.

Stallman and the FSF have not had the last word on the philosophy of free software. Their idea of free software combines principles of freedom with principles of control. Others are far less restrictive in their conception of free software. My software of late is in the public domain. It can be used by the good and evil alike. Stallman's attempt to control the use of words and names represents a contradiction to his own principles of freedom. In addition it plainly doesn't make sense. If you use a particular collection, Debian for example, for a long enough period of time, it is impossible not to encounter the terms "GNU", "FSF". GNU software in particular is very good at advertising. Try loading up Emacs to see what I mean.

To note, I am one who uses GNU to describe the system, Linux to refer to a particular kernel, and Debian to refer to a particular collection of software.

Further note, after I wrote and revised this I discovered the GNU/Linux FAQ, which further shows the rediculous ammount of attention Stallman has devoted to this issue. I hope not to have duplicated this in addressing this issue here, after all, there are far more important and interesting things to discuss.

This is another example of the kind of article Advogato would be better without. Short on facts, devoid of curiosity, it bangs the empty drum loudly but has nothing to offer the attention the noise draws. It's just noise.

Does Free Software need a DCE clone? Is MS really better off having to maintain a knockoff? DCE is widely reviled as an overengineered mess, best forgotten. If the main reason for Free Software to have one is that MS has one, that's no reason at all. MS specializes in making and maintaining (or failing to maintain) messes. Arguably it is their distraction with all their own messes that has made it possible for Free Software to catch up so quickly.

"Ordinary business users" don't see any of it. They will adopt Free Software when their belwethers do, and none of the details make the slightest difference to them.

The lack of intrinsic interest in the article has naturally encouraged the discussion to wander off into flame wars. Don't expect anything of interest to show up on this thread.

Luke, which part of "arrogance, ignorance, and ego" does not precisely describe yourself? (Hint: this is the place where you stop to think, and consider carefully, instead of posting more rants -- and maybe even learn something.)

hiya people, well the article was intended to provoke a reaction, and it's certainly done that. i had high hopes for advogato when it first started in november 1999, and those hopes included that the quality of comments would remain high.

so, leaving the personal attacks aside, for which you should all be very ashamed of yourselves, let's forget them and move on.

please bear in mind that i tend to make several jumps and steps, thinking very _very_ fast and in compact form. what that means is that if i was to describe _all_ the details, people would quickly get bored, sidetracked, overwhelmed etc. etc.

i would also like this article to not just be about DCE/RPC, and for it to be about open source moving rapidly forward and then starting to encourage microsoft to cooperate with open source, by demonstrating to them that open source has something to offer.

until open source has caught up with ALL areas in which microsoft has a technological lead, that simply ain't gonna happen.

answer 1: yes it does, and i answer dan and badger's question below to which i refer you for the justification, and it's do do with your reference to "ordinary business users".

answer 2: no, because there already exists FreeDCE, which is available under an OSF compatible license!

answer: please read the other article, "brief history of windows". in it you will find references to the fact that microsoft based MSRPC on the 250,000 lines of DCE 1.1 reference code.

there is direct evidence that can be obtained by running rpctorture against a Windows NT 4.0 system that demonstrates that the code in Windows NT 4.0 has exactly the same vulnerabilities as the DCE 1.1 reference code.

the question is interesting - however i have to say that if answered incorrectly, it distracts from the issues i'm attempting to focus people's attention on.

the OSes that are used are ones that people believe will do the job.

now, whether those OSes _actually_ do the job they think it will do is another matter.

i feel that it is best to leave it at that, let you think that over and ask me some more questions, neh?

it means that there is software that microsoft has that other people do not. and unfortunately, the gap between the code that microsoft has and the code that everyone else has is widening all the time.

the means by which this lead is widening is by fair means and foul, and it's measured in decades, not just some pretty little hack.

several people have hinted at it here, and my brother as an intelligent business user summed it up best.

because microsoft have the dominant share of the desktop, all business users use microsoft. therefore, the desktop and the back office systems of any new business MUST be compatible with microsoft.

any business that decides not to be compatible with microsoft has to have a REAL good business case for doing so, and, in general, such cases are few and far between.

therefore, for any business to succeed, they must use microsoft, or at least to the outside world present themselves, to all intents and purposes, as being microsoft compatible.

take samba as a classic case in hand. samba gets installed and replaces windows as a file, print and, most recently due to my work a login, server. users make COMPLIMENTS to their administrators like this: "hey, that windows server has got SO much more reliable".

now, because the main file server has been converted to samba / unix, that means that the web server can be replaced, and in one company that was using perl scripts, users started logging in to the server and running the perl scripts directly rather than running perl on windows on their local machine (because it was faster!). soon after that, the users started converting their computers to linux!

and how was it achieved?? because samba is FULLY microsoft compatible for all their needs at the time.

if there had been ANYTHING that was not FULLY microsoft compatible at the time, they would NEVER have gone down the road of converting the entire company from entirely microsoft to entirely linux.

:) there is a _lot_ of history behind that, i could - quite literally - write an answer to that question for several weeks, maybe even months.

be careful what you ask me: you might get an answer :)

dce/rpc is a simple means to turn ordinary c-based applications into distributed applications.

the only theoretical limits on the problems that DCE/RPC solves, therefore, are of your own imagining.

at the risk of distracting from the purpose of this article, let's take some examples:

- large, distributed and SECURE file servers, and i mean REALLY distributed.

ibm ran their web server farm off of a distributed file system for the 1996 olympics.

they had a team round-the-clock doing updates world-wide and the mirror sites all just... worked automatically.

- automated resource management and load-balancing.

a large accountancy firm has created a 24x7 call support centre as a DCE/RPC application. it does round-the-clock company-wide technical support by moving the ENTIRE call centre transparently between the UK, US and Japanese offices.

transparently - and securely.

- distributed databases

the UK National Insurance Database is a Terabyte Database system with thousands and thousands of queries per day via the call centre.

it's also a DCE/RPC application.

- size and scale

by doing several small distributed services, you end up with an easily maintainable codebase, even though it looks massive on the outside.

- Windows NT Domains, Exchange, Services and DCOM

All of Microsoft's console / management tools for NT are DCE/RPC client/server applications.

in this respect, the problem that DCE/RPC solves is that user authentication is totally transparent: you log in ONCE and your authentication credentials are passed transparently by DCE/RPC to the applications.

there are no such systems in existence in open source with the same application level of support and transparency for decent distributed security, transport independence and all of the other features in DCE/RPC: _really_ there aren't. if you think there are, you are only kidding yourself, or you haven't examined DCE/RPC in enough depth.

additionally, if you think DCE/RPC is complicated and a lot of code, you are _also_ kidding yourself.

KDE now rolls in at over 10 million lines.

Perl??

GNOME is a 120 mb binary download last time i checked.

StarOffice?

DCE/RPC _used_ to be relatively complex because it was far, far ahead of its time.

by comparison to a lot of projects these days, it is tiny, VERY well designed, compartmentalised etc. into manageable projects that happen to be bundled into one humungous 80mb set of tarballs.

no they damn well can't!!! microsoft has underlying protocols that were designed FIFTEEN years ago that they can't get rid of, for backwards-compatibility reasons.

heck they don't even tell their OWN engineers what the APIs are!!!

i heard that there was some idiot on one of the internal mailing lists discussing a security report i sent in going "SAMR_USER_INFO_21 structure??? what the hell's this guy talking about??? there's no such _thing_ as a NET_USER_INFO_21 structure in my internal copy of the MSDN documentation!!!/" and one of the more experienced and senior employees had to quietly take him aside and show him a subdirectory with the header files and private (private internal!!!) documentation...

the reason for not telling their own people what the internal docs are is two-fold: one is because of BXPA (munitions) licensing; the other is because they don't want their internal developers accidentally writing code to APIs that have "redirection" (i.e. public) layers on top of them, for fear of them producing legacy and compatibility nightmare scenarios.

it's going to be _another_ ten to fifteen years before NetBIOS really goes away.

incidentally and ironically: NetBIOS and the Network Neighbourhood provide EXACTLY the feature-set required to provide COMPLETE and comprehensive Mobile IP services.

the mobile IP IETF committee would learn a LOT from NetBIOS and the Network Neighbourhood, it's a pity they are so arrogant as to think that it's simple enough for them to reinvent such a complex system.

mr torvalds is finnish not finish, it's not just the large companies, it's the intelligence agencies as well, at third hand funding and control, thinking long-term just in case linux succeeds in gaining significant desktop market share.

i have sent mr gates several requests for information. none of them have been for payroll sheets.

i happen to be an expert on dce/rpc and microsoft's core technology in this area, and also happen to be reasonably good at assessing how long it takes ordinary programmers to write code of this order of magnitude.

also, it's not COM - COM is to DCE/RPC what CORBA is to CORBA's underlying RPC mechanism. COM is a set of RPC services and IDL files on top of a very good, extensible RPC mechanism.

regarding your comment about FreeDCE, mr ishmulker, please contact IBM and ask them. why stay away from a good project which you genuinely believe in?

help yourself: go ask the big companies to invest in DCE/RPC

finally in this series of responses, it saddens me to answer this rather important - and personal - question from mr ncm.

the answer is in two parts.

regarding the first - the ignorance: like most people, mr ncm, i have specialist knowledge, and am just as open-minded and sometimes as scared as everyone else in the world is about new things.

however, unlike a lot of people, i have all the classic symptoms of asperger's syndrome, a form of mild autism that makes me hyper-focussed, hyper-sensitive, hyper-aware and easily overloaded. it makes for interesting times and an ability to absorb new knowledge and make new connections and intuitive leaps at a frightening rate.

frightening for those people with fragile egos and a lot of ignorance to hide.

so i am often accused of arrogance when in fact the confidence in my specialist subjects that appears to be ego-driven comes instead from genuinely knowing what i am talking about.

and you will find that on issues that i know nothing about: if i deem them to be useful to me i will be curious, inquisitive and invite people to teach me, and i become child-like because children learn fastest and are most often underestimated.

regarding the second - the ego: if you knew me at all, mr ncm, you would know that i am one of the most altruistic and kind people you will ever meet, and that kindness has been taken advantage of to the extent that i f i ever meet the people responsible for taking advantage of me and they do not apologise for the consequences of their actions within a few seconds of meeting me again, i will either physically attack them without hesitation, or more likely crumple in a heap and require immediate medical assistance, due to standing alpha wave patterns causing a massive neural overload.

do you understand? if not, please ask.

please bear in mind that i tend to make several jumps and steps, thinking very _very_ fast and in compact form.

Invariably, I find that people who try to think very fast (myself included) tend to make lots of mistakes, which they (I) don't realize until the mistakes get out of hand. Nowadays, when I find myself thinking `fast', I try to stop myself from feeling like a genius.

so i am often accused of arrogance when in fact the confidence in my specialist subjects that appears to be ego-driven comes instead from genuinely knowing what i am talking about.

But being an expert in the innards of DCE doesn't immediately mean that one's also an expert in the factors that make open source projects succeed or fail.

I'm an expert in mathematics and logic, so perhaps I should go around preaching to company bosses about how they should be running their businesses (based of course on mathematics and logic).

because microsoft have the dominant share of the desktop, all business users use microsoft. therefore, the desktop and the back office systems of any new business MUST be compatible with microsoft.

This much I agree with. What I disagree with is on the degree of compatibility needed.

Among my friends, almost all use Windows. I have little trouble communicating with them: I use SMTP/POP3/HTML, they use SMTP/POP3/HTML. Sometimes I get Word documents, in which case mswordview comes to the rescue.

I imagine that for a new business, this is about the degree of compatibility that's required.

take samba as a classic case in hand. samba gets installed and replaces windows as a file, print and, most recently due to my work a login, server. users make COMPLIMENTS to their administrators like this: "hey, that windows server has got SO much more reliable".

This isn't a good example, as it's a case of in-house conversion of an existing business, rather than a new business communicating with the outside world.

Even in a situation of in-house conversion, such happy endings aren't guaranteed: some bosses force their system administrators to switch from Samba/Linux back to Windows, for purely political reasons. In contrast, these same bosses wouldn't hesitate to convert their entire organizations from Win3.1 to Win9x. In the end, it all depends on whether they are willing to convert: it's a social problem, not a technical problem.

But anyway, when I talked about cohesion, I talked about philosophical and political cohesion, not technical cohesion. And apart from distros as SuSE, all GNU/Linux distros are based on Free Software, accordingly to the FSF's definition of Free Software. And that's the "pattern of cohesion and unity" you are missing. So, technically speaking, your answer is off-topic on this point.

You also said that "FSF have not had the last word on the philosophy of free software" because "their idea of free software combines principles of freedom with principles of control" and "others are far less restrictive in their conception of free software". First, having the "last word" is not the point. But, more important, you seems to forget that "principles of control" serves only one purpose : promoting "principles of freedom". Some people choose to accept thoses "principles of freedom" (mBSD softwares are free as GPLed softwares) but seems not considering important to protect this "freedom". Some others do, and that's why "principles of control" exist. "principles of control" and not an end, it's a mean.

Saying that while "Stallman's attempt to control the use of words and names" it "represents a contradiction to his own principles of freedom" denotes a misunderstunding of the basis of liberty. Liberty exits only by control and restrictions. For example, abolishing slavery really means forbidding to enslave people. While the purpose is freeing people, the method remains control.

Finally, it seems to me insulting to consider "rediculous" things just because we have no interest in.

lkcl, do you know that describing you as victim of a "form of mild autism that makes [you] hyper-focussed" does not confirm that you are "open-minded". It fact, this description leads to the opposite conclusion.

Anyway, claiming that you "absorb new knowledge" at a "frightening rate", so "frightening for those people with fragile egos and a lot of ignorance to hide", reveal only a high confidence on yourself, nothing else. And you apparently ignore the fact that it does not suffice to get people trusting you. You need to learn to demonstrate, with arguments - not speeches about how you are open-minded, cool and altruistic.

On to your argument[1], political parties don't claim thier members, how can the FSF, as you say a political organziation, claim membership based on the publication of software. If this is the basis for the name chosen, does the OSF lay equal claim to naming rights, since a collection such as Debain meets their defintion of free software?

I know very well the FSF line on protecting freedom. I simply disagree, why can't you accept that? I wasn't even debating this issue, only to point out the FSF is not the only philosphical position. This is a entirely different debate, and so far from original article in topic.

The contradition I point out is obvious, he wants to control the name people use, yet puts no methods of control in his software licenses.

[1] I want to make this distincion clear. I find your argument even more rediculous than Stallman's, I don't want to be misacussed again of improper atribution.

Well even if both originate from the 70's , you don't think that a good synthesis of them would be innovative ? :)

I agree that GNUstep/Cocoa could be seen as not so "innovative" today in the sense that they are implementations (with some additions) of the OpenStep specs, released in 1994. So they aren't so innovative compared to OpenStep.

But my opinion is that they are innovative when you compare them to the other actual solutions :)

The main subject was about RPC, so take a look of how easy is it to create distributed object with GNUstep. And as you could link Objective C code with C (or C++ on apple), it's a fast way of taking some existing code and get it "distributed". If you want some tutorials about GNUstep, looks at this page.

You ask me to "reread his argument, it is a technical argument". Since you are not willing to quote this argument, please, give me a pointer.

But anyway, re-read the thread : you we're contradicting Richard arguments by saying that GNU/Linux have no cohesion. I answered you that a cohesion exists : a political cohesion. It's not part of the Richard topic.

You said "I know very well the FSF line on protecting freedom and I simply disagree, why can't you accept that?". I perfectly can accept the fact that you are not agree with the FSF nor me. Does it means that I should stop thinking that liberties need to be protected and argumenting about it? You said that you are not talking about that. But indeed you were since you bring to this thread the debate about the fact that liberties should be protected or not.

And I can prove what I said, by copying/pasting your own writings:

The first part is pointing out "the FSF is not the only philosphical position".

" Stallman and the FSF have not had the last word on the philosophy of free software. Their idea of free software combines principles of freedom with principles of control. Others are far less restrictive in their conception of free software. My software of late is in the public domain. It can be used by the good and evil alike."

But the following is clearly a step in the debate, saying which opinion is correct and which is not.

" Stallman's attempt to control the use of words and names represents a contradiction to his own principles of freedom. In addition it plainly doesn't make sense."

So when you say "I wasn't even debating this issue, only to point out the FSF is not the only philosphical position", you are in contradiction with you own writings. Too sad.

Also, you say that Richard "wants to control the name people use" but that's wrong. He want to convince people that GNU/Linux fits more than Linux. But it's not controlling : it does not means that he wants people using one expression or another, whatever which, just because of his own choice. Convincing does not mean controlling

mr yeupou and mr tk,

after a lot of thought over the last two years to reflect on how disastrously badly things went at linuxcare, i want you to know that i do appreciate the distinction between technical excellence and leadership.

i led by example on the samba nt domains lists. the difficulty came when endeavouring to communicate three years worth of work under stressful circumstances, and i failed to be able to stand up for myself and my work.

... this is, however, a distraction from the point of the article.

... the reason it is an integrated system--and not just a collection of useful programs--is because the GNU Project set out to make it one.

I thought I already refuted this political cohesion argument, first by showing other perpectives exist within the community, also that publishing software says nothing about your political orientation.

There is no contradiction here, except perhaps an imagined one. That Stallman's actions are a contradiction to his own position is independent of which position is better. One person's actions do not invalidate a philosophical position.

Obviously there is no way he can control the terms people use, but he is doing everything in his power to persuade the usage of "GNU/Linux". By the way, I think he has a valid argument when restricted purely to the system. But people use Linux not to describe the system, but an entire collection of software. That is the focus of his writing, that is where I have argued he is wrong.

This article was posted on OSnews and generated a lot of replies.