Slashdot Mirror
Dave McAllister (SGI) on Linux and Chilli
Mintslice writes "Dave McAllister, SGI's Directory of Technical Strategy has been touring Australia recently. The Age is running this story about comments he made at at local LUG (LUV). It runs over SGI's intentions for Linux, what they're doing to help development, what this means for marketing at SGI, and a treasure trove of bits and pieces including Chilli Recipes. Something for everyone. "
How can anyone believe SGI has good intentions for Linux after what they have been doing to customers? To put it most simply they are price gougers. Before this gets called flamebait or whatever let me explain...
SGI forces it's customers to buy *$1200* software to run gcc on Irix. I'm not shitting you about this. You can see people having this problem all over USENET and when asked for the product people are given the cold shoulder by SGI. Now what kind of business would still be running Irix 5.3 or lower? None that I can think of so that basically means that only hobbyists are still using the old Indigos and 4D series machines. $1200 to make gcc work is simply ridiculous. It should work with Irix without any additional packages but yet SGI chooses to GOUGE THE CONSUMER. I have asked them to put the needed files for 5.3 (since they also run a freeware site for 5.3) on thier freeware page, but they just blow me off.
Due to their tendency to price gouge and generally just piss people off I refuse to believe that they have only good intentions for Linux. I hope you understand where I'm coming from on this. Price gouging is wrong.
To quote the article: "McAllister estimates about 20 million machines are running Linux ("seats"). He expects this to rise, so that within three years there will be as many Linux PCs as those with Windows. If this is true, what should be worrying Microsoft - which is still wrestling with a US judge's finding that it is a monopoly - is McAllister's prediction that Linux will grow into a fully fledged desktop competitor, with a host of applications." That sounds to me like predicting a Windows-Linux desktop war in the not-to-distant future in a galaxy not-to-far away... I'd be interested to know how many desktop users there are of Linux (any flavour). And how many actually use Linux as their primary desktop machine. Off the topic totally - what experiences have you Slashdotters had with office suites such as ApplixOffice, Star Office, etc?? -Spud. [The Matrix - powered by Linux]
-- "e-idiot: stupidity for the next Millenium."
I wonder what he means when he says the open source community is not innovative. It's certainly true that Linux developed out of Linus imitating (? or learning from history?) UNIX and all that, but surely Linux has come a long way from there? I'm not that well-versed with the kernel myself, so I can't really say with much assurance in what way Linux has "innovated".
But how about ESR's argument that open source is driven by "tickling the developer's itch"? From that standpoint, doesn't open source drive itself to be innovative, eg. when someone says, "hey, such and such a feature would be a neat thing to have!", goes and implement it, and develops a "bazaar" (to use ESR's terminology) and ultimately resulting in a high quality product that surely has innovated? What qualifies as an "innovation" anyway? Surely ESR's fetchmail is an innovative product of open source development? There certainly hasn't been anything quite like it before, although it does take its ideas from several places. After all, every innovation must start from something; something "totally creative" that you dream up without getting any ideas from anywhere else is probably not that great anyway.
I would argue that Open Source does encourage innovation. Yes, a lot of open source software in existence today are written for the sake of having a free alternative to a commercial solution, but that doesn't mean that all open source software is essentially copying existing ideas without innovation. I'd surely like to know what he meant when he said that open source people are not innovative.
mikre he sophia he tou Mikrosophou.
Have you seen this page? http://oss.sgi.com/projects/xfs/ Haven't really looked at it but it looks like it's got plenty of info
"The open-source community is a good imitator but not a good innovator."
Correction, the Linux community is a good imitator.
Open-source (Apache et al) doesn't IMHO fit into that category.
Linux is open-source but open-source is more than just Linux!
The statement that the open-source community is not a good innovator is, like most complex issues, both true and untrue.
On the one hand, only the most blind of observers would suggest that novel products are not emerging from that quarter. The sheer volume of announcements on Freshmeat is just flabbergasting, and scattered like jewels in among the 95% of fairly ordinary stuff are some really excellent software products and many priceless ideas.
But on the other hand, and maybe this is where SGI is coming from, innovation in the Linux kernel is comparatively minimal. I don't think anyone would go so far as to say that it is stifled, but the fact remains that the choice of which ideas are accepted into the official release and which are not is in the hands of a very few people (maybe three or four, or possibly just the one). That must have an effect on innovation, however much we respect the people in question.
As a little example of the above, the DIPC project implemented a gem of an idea (I have absolutely nothing to do with it, by the way): allowing processes that communicate through System V IPC mechanisms on a single host to do so even if they are on different machines, while maintaining 100% compile-time application compatibility because the only difference at the API is a single bit in the IPC headers which you'd flick off or on for single or multiple machine operation. That's innovation, usefulness and elegance rolled into one. But no, Linus didn't want to put it into the standard kernel, and to say that the developers were greatly dispirited is the understatement of the year.
It's worth reflecting that if a kernel facility isn't part of the standard distribution, or worse, if it's available only as a patch, then for all intents and purposes it doesn't exist. We musn't get ourselves into a situation where innovation in the Linux kernel suffers as a result of this possibility.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
You don't know what your talking about and it shows. IRIX scales vastly better than Linux and as good or better than the other proprietary Unices. If it didn't it wouldn't be routinely running on 16-256 CPU machines. Linux is jus tnow getting a threaded TCP/IP stack for example. Linux lost the Mindcraft benchmarks for a good reason, 2.2.x doesn't scale well.
You must have really botched something if your getting 2 hours uptime, or more likely you just made that number up to add drama to your lame post.
As for Linux blowing SGI away on the desktop, I don't see it and neither did Linux in his Comdex speach where he expressed disappointment with the Linux desktop. Linux has no serious commercial applications in key desktop areas like CAD and Animation. I hope this will be changing soon but right now Linux just isn't flying on the desktop outside of Netscape, Star Office, Gimp and a few other free or open source apps like Blender which just don't stack up against commercial apps.
@de_machina
Also, bear in mind that McAllister himself said that a culture change was needed. Corporations, like people, can and do change over time. I mean, look at IBM. If you had told me five years ago that Big Blue would be pouring money into Linux and Apache and Java and XML, I would have ask for some of whatever you were smoking. I probably would have said that it was impossible for that bunch of white-shirted-blue-tied-COBOL-writing-batch-jobber s to learn new tricks.
Give SGI some time to change. They might surprise you yet.
"Genius may have its limitations, but stupidity is not thus handicapped." --Elbert Hubbard (1856-1915)
Microsoft has been uttering the word in every other sentence, but it usually means buying or stealing someone's ideas or technology, seizing the market by force, and taking credit for it.
There are orders of magnitude more innovative ideas at my company than there are developers to implement them, and most of them would just make our software more complex and unwieldy.
Probably the most difficult part of designing GOOD software is knowing what to leave out. Conservative judgement is called for in a tool that many people depend on--otherwise you end up with Windows! Sure, let's throw a graphics API into the kernel! Let's download COM objects into our browser that can do as they like! Let make every piece and every layer interconnect with every other piece and layer so the whole thing becomes a huge Gordian's knot that even the sword of the U.S. Government can't slice through!
A conservative approach means some worthy features will be left out, but it's a lot easier to add a good feature later on than to take out a bad one. One of the more innovative languages I know is Oberon, a language that can be clearly and unambiguously defined in fifty-odd pages, as compared to the hundreds of pages needed to define C++, which is still ambiguous and subject to the implementor's interpretation. Technologies like CORBA existed long before HTTP, but the web wasn't built on them. HTTP was innovative in its simplicity!
Whatever one's take on the platform, one of the big reasons for Java's popularity is that the language has been simplified considerably from C++ and the developers tried to use only tried and true features. Of course, now I hear they're reconsidering operator overloading--don't get me started...
I submit that the best way to encourage innovation is to resist the urge to fold every new feature into the core, instead adding only those features that are absolutely essential to allow multiple, competing innovations to be created on top of them.
The open source community, especially the Linux community, has been accused of "chasing tail lights." I think that this is not only valid, I think it's not the least bit shameful! After all, in a race, on must catch up before pulling ahead, and in some circumstances it is best to let others go first, especially where dead-ends and rickety bridges are concerned.
The strength of the open source community, as I see it, is that on the whole it seems to distinguish between innovation and feature-itis. Perhaps the most "innovative" thing one can do sometimes is make the damn thing work before adding more features! Now I would call that down-right revolutionary!
Brent Rowland
100,000 lemmings can't all be wrong.
from the dipc-2.0.lsm
Copying-policy: GPL. Copyright (C) Kamran Karimi
I've just had a browse of their mailing list archives to see if there is a licensing issue. It turns out that DIPC has been GPL'd since August. (No idea what it was before that.)
Trying to read between the lines in some of their posts, it doesn't seem that licensing was the issue though, because apparently DIPC was getting good support from Alan Cox in the way of header patches but no encouragement from Linus. (Despite being a fan, Alan is said not to have had the time to press the issue further). I get the impression that there was a lot left unsaid.
Each case has its own particulars, but leaving that aside, it's the general case that may be of some concern in the future. Linux is a standard-bearer in the free software community, so we need innovation to be encouraged in the kernel as elsewhere if we're not to be seen as lacking in that department. If the coding is of a good standard and is well integrated, the default answer to inclusion in the development branch needs to be "yes". (Hopefully it is, already -- maybe someone from the kernel list will comment). We can't afford SGI to be right about lack of innovation.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
Brent makes a very good point. Innovation is important, but not at the expense of any other good properties. Good additions are those that are not only good in themselves but also good in their integration and interaction with other subsystems around them. The decades-old buzzword of "modular" is as important today as ever, and the even older one of "coupling" still rules the roost. If a new subsystem spreads static tendrils throughout older well-proven code or interacts dynamically with many other parts in complex ways, it's a disaster waiting to happen.
;-)
Talking about relevant terms, perhaps we should find a little more use for one that academia values a lot: elegance. It encompasses all of the above in one word, and certainly C++ would never have passed its harsh judgement.
[I don't want to imply that DIPC wasn't up to scratch in that area. The user-level interface is wonderfully elegant, but I haven't looked at the implementation at all, yet.]
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
I'm not so sure about that. Most people who work with open standards and university/science/peer review systems say that open source is pretty much exactly like they have been operating all along.
And, more importantly, open source/free software is not positioning itself as an alternative to scientific and university work: but to closed, propriatory corporation development. When we compare the innovation of FS, we have to look to the innovation of private software developers, which has been (esp under the last ten years) very limited.
-
We cannot reason ourselves out of our basic irrationality. All we can do is learn the art of being irrational in a reasonable way.
A while back (like in the 70's) people used to say exactly the same thing about the Japanese, "they can copy inventions but they can't invent things of their own". This may have been true whilst they were playing catch-up, but once they had caught up, they started innovating. I believe that Linux will also follow this pattern - catch up first, then overtake. That's when we'll see move innovation happening in the open-source community.
HH
Yellow tigers crouched in jungles in her dark eyes.
She's just dressing, goodbye windows, tired starlings.
In a sane computing environment, any terminal provides complete access to any computer. I can and often do have situations in which I'm using a keyboard/monitor combination from computer #1, have got some software CD physically loaded in computer #2's CD drive, and am running the program from that CD that's NFS-mounted on computer #3--all the while with I/O going back to machine #1.
What I'm trying to tell you is that where I happen to be sitting makes absolutely no difference in determining what I'm doing. It's completely location-independent. The idea of actually physical collocation is some throwback to the Stone Age of computers. No matter where I am, everything is transparently accessible. This renders ridiculous the question of what kernel is being run by the computer physically closest to me. I don't have to move my body around to access any file or program. Given a network, all it takes is for these programs and files to exist somewhere on some machine on which I have a valid account.
As I look at the programs currently displaying on this terminal, I see that many of them are actually hosted on computers are running a Linux-based operating system, mainly RedHat with some SuSE. But several others are running running on an OpenBSD box, and one is running under FreeBSD. I currently have only one active program running on Solaris, and not under SunOS. The window manager keeping track of all this for me, tvtwm, happens to be running on OpenBSD. But really, this doesn't matter a bit.
I've found that Prisoners of Bill have a very hard time with this concept. The confusion is inherent in certain types of questions, such as the one I've answered here. Free yourself from the cognitive restrictions that Microsoft has insinuated into your worldview.