Sun has started as a workstation company, so even if they have been very successfull in big accounts, they know that that is not enough to survive.
I think, the Opteron boxes are an good move to get more share in the low end server market.
The interesting part is the way they want so sell their software: from the cooperation with AOL on Netscapes server products (Iplanet) to the current Jave Enterprise System, they still seem to believe in selling software as a commercial, closed source product. Even if they they to license it on a yearly base (and give customers real value, different from Microsoft, which software assurance program mostly anoys customers), they still keep the development process in house.
Even their try to sell Linux for desktops, JDS, is something you have to pay per employee or per seat, although it's mostly based on open source software like mozilla, evolution and gnome.
I don't think that this is doomed from the beginning. They may be successfull, if they can convince customers, that it's not just the software they pay for, but also support, service and updates. This could work, both for companies used to a "classic" way of buying software once and paying extra for support and for companies disappointed by using "unsupported" open source.
But this is the software strategy, which is mostly independant from their formerly very successfull hardware business. And software was only a small part of their business up to now. The hardware part is much bigger (and responsible for most of the service revenue). Even if they have cheap x86 (both Intel and AMD) boxes now, UltraSPARC is still their choice for the big servers, and UltraSPARC is lagging behind more and more in terms of performance, so that even much better RAS features (Reliability, Availability, Serviceability) make it hard to sell those boxes and reason a hefty price tag.
So, even after almost three years with losses, Sun still heads interesting times.:-)
Hardware is a commodity, at least you can get reasonable good hardware from many different vendors, from laptops to servers. And you have a broad choice of software, almost regardless of the hardware you choose.
Software is mostly commodity, too, at least standard software like web browsers or office packages. I think free/open source software is a good indicator for this.
So, what's next? IBM and Microsoft have very different strategies to head the future.
IBM's main focus are companies, have been and probably will ever be. (This is IMHO one of the main reasons, they failed to get OS/2 to mainstream, they have almost no experience with marketing mass products) So they concentrate on businesses, bundle service with their products: buy Tivoli or Notes and spend more money for service than for the software. As a nice side effect, you have your staff at your customer. Where else can you find out more about the needs and wishes of your customer?
Microsoft has lived very good by selling software licenses, especially Windows and Office. They want to have a monopoly on single part in an otherwise open industry, and sell/license this part to all other companies competing in this market. They do it with Windows, they do it with Office. That's pretty easy earned money. (It's even easier earned, if they sell it through an OEM, because that OEM is responsible to provide end customer support.)
A Sun manager once told me on the question, why Microsoft doesn't engage in service like IBM does, that service doesn't make as much profit as their license busieness does. They prefer to expand their business modell into other markets.
I don't know, how long they'll be able to keep this strategy working; but it has worked for servers, internet software, games and PDAs. So, right now it's still working, and their strategy to extent it into multimedia or content distribution looks quite promising. Wouldn't it be nice to earn a penny on every song sold? Or maybe even a little more?
Microsofts strategy seems a little bit more risky than IBM's, but it also promises much more profit. And if it fails sometime, there's still a full war chest to realign the company.
You pay 50 USD per employee and get the software to run it on all
your PCs, your employees can even install it at home. You also get one
year support and all updates during this year. If you want support
& updates after that, you have to pay 50 USD per employee
again.
You pay 100 USD for each PC you want to install the software on. (This is the way for the home user to go.) As with the other licensing modell, you can use the software as long as you want, but you have just one year updates and support. After that, you have to pay 100 USD again.
Because Sun used a version of Suse's Enterprise Linux, they have to
pay some money for each copy of JDS they sell to Suse. This makes it difficult for Sun to give JDS away for free.
Maybe they'll switch the base distribution when Novell/Suse/Ximian switches to KDE in the future.;-)
You wrote: In episode 3 [...] Yoda goes to Dagobah (though it might not be revealed in the 3rd movie) [...]
I'm pretty sure, it will be shown. (I don't expect much new stuff in Episode III, BTW.) You remember the old tree on Dagobah, where Luke had his first "date" with the dark side of The Force?:-)
In Episode VI it is explained, that Yoda was able to hide on Dagobah, because he killed a Sith Lord there, exactly where that tree is, and whose left behind aura saved Yoda from being found by the emperor.
So we just need a "victim" for Yoda. And we already know who him.:-)
George Lukas let Yoda already have a quite absurd fight with Count Dooku in Episode II. Count Dooku is nowhere mentioned in Episode IV to VI and somebody explained somewhere, that Sith always show up in packs of two. The Emperor, Darth Vader and Count Dooku (aka Darth Tyranus, thanks to imdb) are one too much. So bye, bye, Dooku, R.I.P.
So not only the Count has to die, but the studio bosses also love lengthy fighting scenes, to attract the not so much Star War addicted. So I would be very, very surprised, if we don't see Count Dooku dying on Dagobah in Episode III.
I think, the Star Wars universe is quite well known, so there is more than enough fabric for an excess length Episode III. There will be some new details (why or how get C3PO and R2D2 the leap to Episode IV?) or dispensable scenes (with Jar Jar, for example *eg*). But I think, that the basic story board is already pretty well known.
From my point of view, the Sun 386i has never been anything else than a toy.
The first SPARC box has been presented 1987 or 1988, the 386i some month later. There have been two models of the 386i, comapred to a complete line of workstations and servers with the new SPARC processors. Including great new technology (RISC was new an kewl), much better performance and a fancy new case design from frog design.
So it was pretty clear, that Sun would bet their future on Sparc. And not on Intel processors.
I'm not sure, which one of the four founders compared workstations and servers to sports cars and trucks. And especially constructing sports cards would be much more fun. Even if the servers earned the money, Sun has been young enough that times to spend that money (and more) on a fun workstation with the 80386, the first Intel chip, compareable to Motorolas 68k line of processors.
Windows 3.x was tailored around the 80286, which was the first x86 processor with the protected mode (but still with 16 bit/24 bit). (Even if it could use some of the features of the 80386)
But the 80386 was the first 32 bit x86 processor from Intel, enhancing the protected mode and introducing the virtual 86 mode.
Even if the 80286 and the 80386 have many advanced features (protected mode with 16/24 Bits and 32 Bits) compared to the 8088, MS-DOS was a 8088 operating system in it's core till the end (emm386 was just an addon).
Windows was unlike GEM just a set of APIs for programms with a GUI, but also added some funktions usually found in the core of an operating system: memory management, (limited) multitasking, etc.
Looking back, this seems to me an important point for the success of windows: with windows, everybody could benefit from the advanced possibilities of their 80x86 processor (and leaving the 640k memory limit behind) and still run the old MS-DOS applications.
I don't see Win95 as a big leap. Okay, the desktop looks completely different from Win 3.1 (the last release before that). But there's not that much more. People used Win 3.x even years after the release of 1995. The old 3.x just worked for them.
The real milestone was Windows 3.0. The versions before (1.0, 2.0 and the interem releases/286 and/386) have been just ugly.
I remember back when I showed Windows 3.0 a friend of mine (as computer addicted as I was). He was more than impressed, he was thrilled. When the installer switched to graphics mode, this first reaction was: wow!
We liked Windows 3.0, it was - in our eyes - even prettier than those times Amiga or ST desktops.
Looking back, Windows 3.0 and 3.1 had already some of the most important features incorporeted, like truetype font support (Linux has still problems in that area even now... *sigh*) or COM to embedd applications.
Another important feature was a programming modell, leaving the 640k barrier and the 8088 behind. They could even use the features of the then state of the art 80386.
Windows 95 had many improvements, like the completely new desktop or the 32 bit API. (which got backported in parts as Win32s, remember?) But I think without the huge success of Windows 3.x, we would have never seen Win95. We would use OS/2 right now, a version that would look very different from the desktop IBM created after the divorce with Microsoft. (Because the main cause for Microsoft for this divorce was the huge success of Windows 3.x. Why bother with a difficult partner, when you can have even more success alone?)
Later, when Windows NT hit the streets, I was used to work with the programm manager (windows 3.x and nt 3.x style), that I needed a lot of time, to get familiar with Windows 95 and NT4.
When X was created, servers with many (ASCII or EBCDIC) terminals have been much more common than today. PCs and graphic workstations have been too expensive (either in acquisition costs and administration costs) to deploy on every desk.
And X Terminals (from NCD, Tektronics, DEC and many others) have been quite popular in Unix environments for a while.
But as we all know, the PC has won. Maybe not the whole war, but at any means the last battles.
(*sigh* Talking 'bout war is not amusing this times...)
Just for the records: NT 4.0 was released in 1996.
It's basically a Windows NT 3.51 with the Windows 95 desktop.
This desktop is far from being capable of many of the nicer features buried in the NT kernel.
A much better solution was promised for Windows NT 5 aka Cairo, but that failed to deliver and changed into Windows 2000, which still has the Windows 95 destop, after all with some changes like in Windows 98 or ME.
I've a book from my parents about live in the year 2000. It's from somewhere in the late fifties or early sixties and has a whole bunch of articles from then renowned people. At least most of them have impressive titles.;-)
They wrote about almost everything: social stuff, controlling the weather, living in space, man like robots making all the domestic work and finding the final solutions for many environmental and energy problems.
They just missed one point: micro electronics, computers, internet and all the stuff that keeps most of the people here busy.:-)
The only thing that comes near that, are robots. They are fascinated about robots. (They would like the first episode from the Animatrix.:-)) They write lengthy about them doing all kinds of work, walking, speaking, grabing things, etc.
I'm just wondering: what did they think, the computing power behind that comes from? A room full of valves in every apartment to clean the floor and wash the dishes?
Remembering the kneeling down RMS demanded from the KDE team after the Qt license change, because they "lost" the right to do so after violating the GPL by linking GPLed code with Qt?
I'm wondering, why he doesn't demand a similar beg for mercy from everybody not saying "guh-new linux".
Could be a little time consuming, but definitely a nice thing for his ego.;-)
I was impressed by this book review, so that I was already thinking about a purchase and checking at the german amazon website (guess where I'm from?) for price and availability.
The book had two quite bad reviews there. (2.5 points out of 5, which seems to be a pretty bad rating for amazon) The main point from both was the missing link between the java build tools, which seem to be discussed considerably, and eXtreme programming.
Solaris 9/x86 really killed or just deferreded?
on
No Solaris 9 for x86
·
· Score: 4, Informative
A Sun engineer told me yesterday, that Solaris 9 for x86 will be deferred some time, but _not_ eol'ed.
There is currently a beta for x86 and a release is still planned and worked on.
I believe this engineer quite trusworthy, especially more than a Linux gazette...
Another interesting piece of information from this source: they are stopping the possibility to download Solaris 8 x86 from their webserver, but you have to buy the media kit.
If you're nowadays a GUI developer, I expect you to use Windows as your primary platform. (exceptions only proove the rule *g*)
C#/.NET is targeted primarily at Web Services, (almost) the same with the J2EE plattform. So I don't think, that GUI programming is the "killer feature" for the future.
From my point of view, Sun takes very much effort to bring good development practices like design patterns to Java. (See the J2EE pattern repository at java.sun.com for example) And Erich Gamma (one of the Gang of Four, which wrote the most famous pattern book) has stated in a conference (JavaOne? don't remember), that Java is the language/library/plattform with the highest pattern density he knows.
I think this can be much more an "killer feature"; design and architecture of distributed and multi tiered applications is much more diffcult (and therefor important) compared with monolitic (Windows) applications.
To lure programmers from (more or less) monolitic programms (like Windows GUI apps) to web services is one of the most important tasks for Microsoft and Sun. It will be interesting to see wether fancy tools (Microsoft VisualStudio) or good software engineering practices are the better strategy.
DEC was renowned for nothing so much as their inability to market good products.
Uh, I thought that company was IBM?;-)
However, another question also arises. The Alpha has been around for, what, a decade or so now? Possibly the architecture is nearing the end of its life cycle, and if so no one is going to want to spend much to acquire it.
If you take a look at the market, Alpha is one of the "youngest" chips around. x86 is succesfull for more than twenty years now. And SPARC ('87) and POWER ('89?) reach their 15 yrs anniversary.
I also remember a paper from DEC with a planned lifetime for the Alpha architecture with 20 or 25 years.
I think that time has proven that you don't need a completely new archicture to keep pace with the technical advance. (It's about evolution or revolution.) AFAIR Tomshardware has a great article about (changing) chip architecture and (pretty static) instruction sets.
Great article. At this point, it's the only article rated higher than 2 and that's fully deserved.
But I have some additional points:
6. The world is not object oriented. Even if oo is a usefull tool, it is no silver bullet.
7. RDBMS are proven technology and rather well standardised, OODBMS aren't. Currently there is a proposal for a standard (java data objects), but even that only addresses one plattform.
This is not the first discussion on/. about the best technology for the generation of dynamic web content. So I read only the conclusion of the article, and that looked okay for me.
My personal favourite is Java. Not because jsp or servlets are superior to php or something else from a technology view. (Even if they have a small advantage in my eyes.) But with Java, independent from the kind of task, you get a much better tool from a software engineering point of view. You get a basket full with the best practices and concepts for building complex applications: OOA/OOD, patterns, components, etc. Java is full of these and if you start understanding them, they'll lead to better programms.
Most of this should be possible with php (as an example), too. But a short search at www.php.org about advanced design topics got me only basic hints like "put your code in libraries, not in the page" or an unanswered questions for a mvc framework.
I started development of web applications with perl and cgi four years ago. This is the code centric approach stated in the posting above. Later I started embedding perl code in html. Both times I ended with a big mess. (This is okay, if you can sell your boss a "technology and design update".)
My first article about jsp was about the model-vier-controller (mvc) design pattern, too. If you separate the data modell, the programming logic behind your user interface (the controller) and the user interface itself (the view), you can quite simple replace the html interface with a (for example) wap interface. This is mostly the same idea as from the author above.
As I said before: good design should also be possible with php, embedded perl or something else. But from what I see, people don't do it.
There are compilers (and interpreters?) for several languages, which create Bytecode for the JVM.
Besides that, most of the non-Microsoft languages for.NET are only of academic interest.
And, more interesting, they usually lack those features, the MSIL doesnt support, multiple inheritance in Eiffel for example.
Native execution.
I don't think, that performance is a point. The HotSpot technology looks at the execution hot spots and optimizes them at runtime agressively. Dynamic optimization can gather some performance improvements, no (static) compiler can do.
Cross-platform
The examples, which come with the.NET Beta are quite windows centric. And with the knowledge, that Windows NT started as cross-plattform, I doubt.NET will ever run satisfying on another plattform that Microsofts own.
In my eyes,.NOT (no typo) nothing else than Microsofts own version of Java. They started five years later with all advantages and disadvantages.
The main problem with this is the long tradition of copying Microsoft has. MS-DOS was inspired by CP/M (and a little Unix), Windows was inspired by Mac, etc. Copying Java will be a lot more difficult, because it's a little bit more than a single program.
Even if the core APIs of operating systems are quite similar, the GUI APIs aren't. This looks like the biggest problem for Java and Swing for me. There are few cross-plattfrom GUI libraries and if you dig a little, you find lots of unhappy people. In my eyes, cross-plattform GUI development is quite tricky and I don't expect redoing it in Java from scratch to fix the common problems.
As far as I know Sun is working hard to improve Swing performance. I dont know details, there are two more releases (1.3.1 and 1.4) planned for this year. Hopefully with faster swing.
BTW: Last week I got a notice that the bug, that prevented Swing apps running through a ssh tunnel, was fixed. Looks like they are really working at Swing.
Of course Java isn't dead. Even if Java isn't very popular at the desktop, because of bad browsers and the lack of demand for platform independance (because 95% of all desktops run Windows), it's very alive at the server, which is (together with others like IBM) Sun's domain. And I mean real servers, not these funny little NT boxes.;-)
If you take a look at the raw numbers, there are more mobile phones than PCs and the numbers. If you can get the mobiles into the internet, there is a *huge* demand for servers. Got it?:-)
Microsoft tries something similar with.NET, but IMHO they have two drawbacks: they are some years behind und their world picture is PC centric.
Here in Germany, the oncoming standard for set-top boxes includes java. So if Java is already there, why bother with two (very similar, but incompatible) standards for very similar markets?
No. A standard is a piece of paper and not a piece of software. Microsoft first makes the software and after that eventually a piece of paper, describing it, which has often been unusable. With a standard, different software companies should be able to compete to provide the best implementation.
Windows is no standard in this sense; Office, too.
And even C# and.NET starting as a piece of paper, the examples are so windows specific, it's mostly doubtfull, that it will get anytime something else than a next generation windows.
In my eyes microkernels have been a nice idea, but didn't made it into mainstream and got obsolete in the meantime. Let me try to explain why.
In the eighties, there was the the RISC vs. CISC discussion with microprocessors. Make the processors lean and fast and do the complex stuff in software. This matches quite fine with the monolitic kernel vs. microkernel debate. Make the kernel small and fast, concentrate on the important things and do the rest in user mode.
Today, CISC and RISC are obsolete idioms. Modern processors still use many of the ideas from the initial RISC concept, even the CISC processor families like the x86. But processor design has evolved in the meantime. A POWER4 or ULTRAsparc III chips is at least as complex as a Pentium 4.
I think, the same has happened with os kernel design. The main idea behind the microkernel was better structuring of the kernel and the systems services. Even if current systems like Linux don't obtain the modulariy of (for example) the hurd, there is much more of it in current operating systems than a (for example) Unix system from the eighties. So current mainstream operating systems inherit some (the good) ideas from microkernels and keep some of older, but proven concepts.
I'm not in the internal architecture of the Linux kernel, but as far as I understand it, there is such a structure. It's not the chaos, which AST claims monolithic kernels to be.
If you take a look at Windows 2000 (there are some slides at www.microsoft.com), which claims to have a microkernel, most system services run in kernel mode. This doesn't look much like a "pure" microkernel for me any more.
Slashdot Mirror
Sun has started as a workstation company, so even if they have been very successfull in big accounts, they know that that is not enough to survive.
I think, the Opteron boxes are an good move to get more share in the low end server market.
The interesting part is the way they want so sell their software: from the cooperation with AOL on Netscapes server products (Iplanet) to the current Jave Enterprise System, they still seem to believe in selling software as a commercial, closed source product. Even if they they to license it on a yearly base (and give customers real value, different from Microsoft, which software assurance program mostly anoys customers), they still keep the development process in house.
Even their try to sell Linux for desktops, JDS, is something you have to pay per employee or per seat, although it's mostly based on open source software like mozilla, evolution and gnome.
I don't think that this is doomed from the beginning. They may be successfull, if they can convince customers, that it's not just the software they pay for, but also support, service and updates. This could work, both for companies used to a "classic" way of buying software once and paying extra for support and for companies disappointed by using "unsupported" open source.
But this is the software strategy, which is mostly independant from their formerly very successfull hardware business. And software was only a small part of their business up to now. The hardware part is much bigger (and responsible for most of the service revenue). Even if they have cheap x86 (both Intel and AMD) boxes now, UltraSPARC is still their choice for the big servers, and UltraSPARC is lagging behind more and more in terms of performance, so that even much better RAS features (Reliability, Availability, Serviceability) make it hard to sell those boxes and reason a hefty price tag.
So, even after almost three years with losses, Sun still heads interesting times. :-)
Hardware is a commodity, at least you can get reasonable good hardware from many different vendors, from laptops to servers. And you have a broad choice of software, almost regardless of the hardware you choose.
Software is mostly commodity, too, at least standard software like web browsers or office packages. I think free/open source software is a good indicator for this.
So, what's next? IBM and Microsoft have very different strategies to head the future.
IBM's main focus are companies, have been and probably will ever be. (This is IMHO one of the main reasons, they failed to get OS/2 to mainstream, they have almost no experience with marketing mass products) So they concentrate on businesses, bundle service with their products: buy Tivoli or Notes and spend more money for service than for the software. As a nice side effect, you have your staff at your customer. Where else can you find out more about the needs and wishes of your customer?
Microsoft has lived very good by selling software licenses, especially Windows and Office. They want to have a monopoly on single part in an otherwise open industry, and sell/license this part to all other companies competing in this market. They do it with Windows, they do it with Office. That's pretty easy earned money. (It's even easier earned, if they sell it through an OEM, because that OEM is responsible to provide end customer support.)
A Sun manager once told me on the question, why Microsoft doesn't engage in service like IBM does, that service doesn't make as much profit as their license busieness does. They prefer to expand their business modell into other markets.
I don't know, how long they'll be able to keep this strategy working; but it has worked for servers, internet software, games and PDAs. So, right now it's still working, and their strategy to extent it into multimedia or content distribution looks quite promising. Wouldn't it be nice to earn a penny on every song sold? Or maybe even a little more?
Microsofts strategy seems a little bit more risky than IBM's, but it also promises much more profit. And if it fails sometime, there's still a full war chest to realign the company.
JDS can also be subscription based.
There are basically two licensing modells:
Because Sun used a version of Suse's Enterprise Linux, they have to pay some money for each copy of JDS they sell to Suse. This makes it difficult for Sun to give JDS away for free.
Maybe they'll switch the base distribution when Novell/Suse/Ximian switches to KDE in the future. ;-)
Hi,
this is completely off topic, but du you know a vt220 emulation with user definable fonts (softfonts?) implemented?
I tried some teminal emulators, e.g. jta, to replay some old ascii art movies, but none had this feature.
t.i.a
Sun breaking away from TI would most likely be very damaging to TI.
Yes, but if you have to choose wether to piss off an old friend, who has been quite unreliable lately, or get pissed yourself, what would you choose?
You wrote:
In episode 3 [...] Yoda goes to Dagobah (though it might not be revealed in the 3rd movie) [...]
I'm pretty sure, it will be shown. (I don't expect much new stuff in Episode III, BTW.) You remember the old tree on Dagobah, where Luke had his first "date" with the dark side of The Force? :-)
In Episode VI it is explained, that Yoda was able to hide on Dagobah, because he killed a Sith Lord there, exactly where that tree is, and whose left behind aura saved Yoda from being found by the emperor.
So we just need a "victim" for Yoda. And we already know who him. :-)
George Lukas let Yoda already have a quite absurd fight with Count Dooku in Episode II. Count Dooku is nowhere mentioned in Episode IV to VI and somebody explained somewhere, that Sith always show up in packs of two. The Emperor, Darth Vader and Count Dooku (aka Darth Tyranus, thanks to imdb) are one too much. So bye, bye, Dooku, R.I.P.
So not only the Count has to die, but the studio bosses also love lengthy fighting scenes, to attract the not so much Star War addicted. So I would be very, very surprised, if we don't see Count Dooku dying on Dagobah in Episode III.
I think, the Star Wars universe is quite well known, so there is more than enough fabric for an excess length Episode III. There will be some new details (why or how get C3PO and R2D2 the leap to Episode IV?) or dispensable scenes (with Jar Jar, for example *eg*). But I think, that the basic story board is already pretty well known.
From my point of view, the Sun 386i has never been anything else than a toy.
The first SPARC box has been presented 1987 or 1988, the 386i some month later. There have been two models of the 386i, comapred to a complete line of workstations and servers with the new SPARC processors. Including great new technology (RISC was new an kewl), much better performance and a fancy new case design from frog design.
So it was pretty clear, that Sun would bet their future on Sparc. And not on Intel processors.
I'm not sure, which one of the four founders compared workstations and servers to sports cars and trucks. And especially constructing sports cards would be much more fun. Even if the servers earned the money, Sun has been young enough that times to spend that money (and more) on a fun workstation with the 80386, the first Intel chip, compareable to Motorolas 68k line of processors.
Windows 3.x was tailored around the 80286, which was the first x86 processor with the protected mode (but still with 16 bit/24 bit). (Even if it could use some of the features of the 80386)
But the 80386 was the first 32 bit x86 processor from Intel, enhancing the protected mode and introducing the virtual 86 mode.
Even if the 80286 and the 80386 have many advanced features (protected mode with 16/24 Bits and 32 Bits) compared to the 8088, MS-DOS was a 8088 operating system in it's core till the end (emm386 was just an addon).
Windows was unlike GEM just a set of APIs for programms with a GUI, but also added some funktions usually found in the core of an operating system: memory management, (limited) multitasking, etc.
Looking back, this seems to me an important point for the success of windows: with windows, everybody could benefit from the advanced possibilities of their 80x86 processor (and leaving the 640k memory limit behind) and still run the old MS-DOS applications.
I don't see Win95 as a big leap. Okay, the desktop looks completely different from Win 3.1 (the last release before that). But there's not that much more. People used Win 3.x even years after the release of 1995. The old 3.x just worked for them.
/286 and /386) have been just ugly.
The real milestone was Windows 3.0. The versions before (1.0, 2.0 and the interem releases
I remember back when I showed Windows 3.0 a friend of mine (as computer addicted as I was). He was more than impressed, he was thrilled. When the installer switched to graphics mode, this first reaction was: wow!
We liked Windows 3.0, it was - in our eyes - even prettier than those times Amiga or ST desktops.
Looking back, Windows 3.0 and 3.1 had already some of the most important features incorporeted, like truetype font support (Linux has still problems in that area even now... *sigh*) or COM to embedd applications.
Another important feature was a programming modell, leaving the 640k barrier and the 8088 behind. They could even use the features of the then state of the art 80386.
Windows 95 had many improvements, like the completely new desktop or the 32 bit API. (which got backported in parts as Win32s, remember?) But I think without the huge success of Windows 3.x, we would have never seen Win95. We would use OS/2 right now, a version that would look very different from the desktop IBM created after the divorce with Microsoft. (Because the main cause for Microsoft for this divorce was the huge success of Windows 3.x. Why bother with a difficult partner, when you can have even more success alone?)
Later, when Windows NT hit the streets, I was used to work with the programm manager (windows 3.x and nt 3.x style), that I needed a lot of time, to get familiar with Windows 95 and NT4.
Short answer: Terminals
When X was created, servers with many (ASCII or EBCDIC) terminals have been much more common than today. PCs and graphic workstations have been too expensive (either in acquisition costs and administration costs) to deploy on every desk.
And X Terminals (from NCD, Tektronics, DEC and many others) have been quite popular in Unix environments for a while.
But as we all know, the PC has won. Maybe not the whole war, but at any means the last battles.
(*sigh* Talking 'bout war is not amusing this times...)
Just for the records: NT 4.0 was released in 1996.
It's basically a Windows NT 3.51 with the Windows 95 desktop.
This desktop is far from being capable of many of the nicer features buried in the NT kernel.
A much better solution was promised for Windows NT 5 aka Cairo, but that failed to deliver and changed into Windows 2000, which still has the Windows 95 destop, after all with some changes like in Windows 98 or ME.
I've a book from my parents about live in the year 2000. It's from somewhere in the late fifties or early sixties and has a whole bunch of articles from then renowned people. At least most of them have impressive titles. ;-)
:-)
:-))
They wrote about almost everything: social stuff, controlling the weather, living in space, man like robots making all the domestic work and finding the final solutions for many environmental and energy problems.
They just missed one point: micro electronics, computers, internet and all the stuff that keeps most of the people here busy.
The only thing that comes near that, are robots. They are fascinated about robots. (They would like the first episode from the Animatrix.
They write lengthy about them doing all kinds of work, walking, speaking, grabing things, etc.
I'm just wondering: what did they think, the computing power behind that comes from? A room full of valves in every apartment to clean the floor and wash the dishes?
Remembering the kneeling down RMS demanded from the KDE team after the Qt license change, because they "lost" the right to do so after violating the GPL by linking GPLed code with Qt?
;-)
I'm wondering, why he doesn't demand a similar beg for mercy from everybody not saying "guh-new linux".
Could be a little time consuming, but definitely a nice thing for his ego.
I was impressed by this book review, so that I was already thinking about a purchase and checking at the german amazon website (guess where I'm from?) for price and availability.
The book had two quite bad reviews there. (2.5 points out of 5, which seems to be a pretty bad rating for amazon) The main point from both was the missing link between the java build tools, which seem to be discussed considerably, and eXtreme programming.
A Sun engineer told me yesterday, that Solaris 9 for x86 will be deferred some time, but _not_ eol'ed.
There is currently a beta for x86 and a release is still planned and worked on.
I believe this engineer quite trusworthy, especially more than a Linux gazette...
Another interesting piece of information from this source: they are stopping the possibility to download Solaris 8 x86 from their webserver, but you have to buy the media kit.
If you're nowadays a GUI developer, I expect you to use Windows as your primary platform. (exceptions only proove the rule *g*)
C#/.NET is targeted primarily at Web Services, (almost) the same with the J2EE plattform. So I don't think, that GUI programming is the "killer feature" for the future.
From my point of view, Sun takes very much effort to bring good development practices like design patterns to Java. (See the J2EE pattern repository at java.sun.com for example) And Erich Gamma (one of the Gang of Four, which wrote the most famous pattern book) has stated in a conference (JavaOne? don't remember), that Java is the language/library/plattform with the highest pattern density he knows.
I think this can be much more an "killer feature"; design and architecture of distributed and multi tiered applications is much more diffcult (and therefor important) compared with monolitic (Windows) applications.
To lure programmers from (more or less) monolitic programms (like Windows GUI apps) to web services is one of the most important tasks for Microsoft and Sun. It will be interesting to see wether fancy tools (Microsoft VisualStudio) or good software engineering practices are the better strategy.
DEC was renowned for nothing so much as their inability to market good products.
Uh, I thought that company was IBM? ;-)
However, another question also arises. The Alpha has been around for, what, a decade or so now? Possibly the architecture is nearing the end of its life cycle, and if so no one is going to want to spend much to acquire it.
If you take a look at the market, Alpha is one of the "youngest" chips around. x86 is succesfull for more than twenty years now. And SPARC ('87) and POWER ('89?) reach their 15 yrs anniversary.
I also remember a paper from DEC with a planned lifetime for the Alpha architecture with 20 or 25 years.
I think that time has proven that you don't need a completely new archicture to keep pace with the technical advance. (It's about evolution or revolution.) AFAIR Tomshardware has a great article about (changing) chip architecture and (pretty static) instruction sets.
Great article. At this point, it's the only article rated higher than 2 and that's fully deserved.
But I have some additional points:
6. The world is not object oriented. Even if oo is a usefull tool, it is no silver bullet.
7. RDBMS are proven technology and rather well standardised, OODBMS aren't. Currently there is a proposal for a standard (java data objects), but even that only addresses one plattform.
This is not the first discussion on /. about the best technology for the generation of dynamic web content. So I read only the conclusion of the article, and that looked okay for me.
My personal favourite is Java. Not because jsp or servlets are superior to php or something else from a technology view. (Even if they have a small advantage in my eyes.) But with Java, independent from the kind of task, you get a much better tool from a software engineering point of view. You get a basket full with the best practices and concepts for building complex applications: OOA/OOD, patterns, components, etc. Java is full of these and if you start understanding them, they'll lead to better programms.
Most of this should be possible with php (as an example), too. But a short search at www.php.org about advanced design topics got me only basic hints like "put your code in libraries, not in the page" or an unanswered questions for a mvc framework.
I started development of web applications with perl and cgi four years ago. This is the code centric approach stated in the posting above. Later I started embedding perl code in html. Both times I ended with a big mess. (This is okay, if you can sell your boss a "technology and design update".)
My first article about jsp was about the model-vier-controller (mvc) design pattern, too. If you separate the data modell, the programming logic behind your user interface (the controller) and the user interface itself (the view), you can quite simple replace the html interface with a (for example) wap interface. This is mostly the same idea as from the author above.
As I said before: good design should also be possible with php, embedded perl or something else. But from what I see, people don't do it.
Any Language.
There are compilers (and interpreters?) for several languages, which create Bytecode for the JVM. .NET are only of academic interest.
Besides that, most of the non-Microsoft languages for
And, more interesting, they usually lack those features, the MSIL doesnt support, multiple inheritance in Eiffel for example.
Native execution.
I don't think, that performance is a point. The HotSpot technology looks at the execution hot spots and optimizes them at runtime agressively. Dynamic optimization can gather some performance improvements, no (static) compiler can do.
Cross-platform
The examples, which come with the .NET Beta are quite windows centric. And with the knowledge, that Windows NT started as cross-plattform, I doubt .NET will ever run satisfying on another plattform that Microsofts own.
In my eyes, .NOT (no typo) nothing else than Microsofts own version of Java. They started five years later with all advantages and disadvantages.
The main problem with this is the long tradition of copying Microsoft has. MS-DOS was inspired by CP/M (and a little Unix), Windows was inspired by Mac, etc. Copying Java will be a lot more difficult, because it's a little bit more than a single program.
Even if the core APIs of operating systems are quite similar, the GUI APIs aren't. This looks like the biggest problem for Java and Swing for me. There are few cross-plattfrom GUI libraries and if you dig a little, you find lots of unhappy people. In my eyes, cross-plattform GUI development is quite tricky and I don't expect redoing it in Java from scratch to fix the common problems.
As far as I know Sun is working hard to improve Swing performance. I dont know details, there are two more releases (1.3.1 and 1.4) planned for this year. Hopefully with faster swing.
BTW: Last week I got a notice that the bug, that prevented Swing apps running through a ssh tunnel, was fixed. Looks like they are really working at Swing.
Of course Java isn't dead. Even if Java isn't very popular at the desktop, because of bad browsers and the lack of demand for platform independance (because 95% of all desktops run Windows), it's very alive at the server, which is (together with others like IBM) Sun's domain. And I mean real servers, not these funny little NT boxes. ;-)
:-)
.NET, but IMHO they have two drawbacks: they are some years behind und their world picture is PC centric.
If you take a look at the raw numbers, there are more mobile phones than PCs and the numbers. If you can get the mobiles into the internet, there is a *huge* demand for servers. Got it?
Microsoft tries something similar with
Here in Germany, the oncoming standard for set-top boxes includes java. So if Java is already there, why bother with two (very similar, but incompatible) standards for very similar markets?
They enforce standards and create opportunities.
No. A standard is a piece of paper and not a piece of software. Microsoft first makes the software and after that eventually a piece of paper, describing it, which has often been unusable. With a standard, different software companies should be able to compete to provide the best implementation.
Windows is no standard in this sense; Office, too.
And even C# and .NET starting as a piece of paper, the examples are so windows specific, it's mostly doubtfull, that it will get anytime something else than a next generation windows.
In my eyes microkernels have been a nice idea, but didn't made it into mainstream and got obsolete in the meantime. Let me try to explain why.
In the eighties, there was the the RISC vs. CISC discussion with microprocessors. Make the processors lean and fast and do the complex stuff in software. This matches quite fine with the monolitic kernel vs. microkernel debate. Make the kernel small and fast, concentrate on the important things and do the rest in user mode.
Today, CISC and RISC are obsolete idioms. Modern processors still use many of the ideas from the initial RISC concept, even the CISC processor families like the x86. But processor design has evolved in the meantime. A POWER4 or ULTRAsparc III chips is at least as complex as a Pentium 4.
I think, the same has happened with os kernel design. The main idea behind the microkernel was better structuring of the kernel and the systems services. Even if current systems like Linux don't obtain the modulariy of (for example) the hurd, there is much more of it in current operating systems than a (for example) Unix system from the eighties. So current mainstream operating systems inherit some (the good) ideas from microkernels and keep some of older, but proven concepts.
I'm not in the internal architecture of the Linux kernel, but as far as I understand it, there is such a structure. It's not the chaos, which AST claims monolithic kernels to be.
If you take a look at Windows 2000 (there are some slides at www.microsoft.com), which claims to have a microkernel, most system services run in kernel mode. This doesn't look much like a "pure" microkernel for me any more.