I know at least Novell/SuSE out of the box nowadays allow LDAP auth configuration. Expect big things to come of Novell/SuSE in the future, but for now they have a good start.
That was hilarious, looking through some of their 'transcipts': Weather report excertp: I know this is the way this OS Linux agency lot of clout through northern parts and through eastern parts and you can see how this is this just pushing its weight and keeping eastern parts of Britain but there's no such plan through central and eastern parts he is going to be bringing a lot of snow Anether except: the coming into force this week in August tested in the courts they sit unworkable and unenforceable and that is Senator the papers in the Linux Hymon and Atticus of enemy is not found anywhere at eight fifteen what's coming up later this month it he finally asked if more than willing we're looking
Others have mentioned the prices go lower, but additionally, if you have a blade infrastructure or are building a large scale network and can benefit from blade infrastructure: http://www-132.ibm.com/webapp/wcs /stores/servlet/C ategoryDisplay?categoryId=2586156&storeId=1&catalo gId=-840&langId=-1 A Power server that is actually priced competitively with the Intel blade servers. True the bladecenter chassis isn't well suited to some environments, but there is an option.
The big advantage of vector graphics is not that it reduces size, but that it allows abstraction of pixel-wise screen geometry from GUI design. It allows designers to request a size in inches/centimeters (assuming monitor DPI is known by graphics system, which is generaly the case nowadays), and produce a perfectly scaled image for the screen. You go to 1600x1200 and you don't end up with unreadable dialogs, you end up with really crisp, readable dialogs (allowing user to reduce point size and have it remain readable, so you can still acheive a goal of fitting more stuff on the screen).
As a bonus, editing and particularly the 'print preview' features are then much more easily perfectly representative of what the result will be.
One thought: pressing a key mutes the speaker (or emits alternate sound) before making the noise, noise stops before mute breaks. It's not as if you could hear anything over the tone anyway...
http://www.eweek.com/article2/0,1759,1758758,00.as p is the article I read
I don't think 'betting the farm' on on-demand computing is a good idea for sun. The market is there and has money, but I don't think a company like Sun can sustain itself on almost exclusively this.
IBM does this with their servers, but if they didn't have a bunch of customers also buying the systems whole working to help subsidize research and development of the servers, not to mention manufacturing the systems in volume, the revenue from on-demand would not be sustainable.
I found the comment about IBM pensions amusing, but the reality is that while IBM does have armies of people getting paid keeping 1$/cpu hour from sustaining the company, the statement ignores how much broader IBM's scope is compared to Sun's present and plan.
Goes to SuSE, hands down, for a corporate standard.
It really only is a choice between RedHat and SuSE (reasonable support contracts, etc etc), so the points are a comparison between RedHat EL and SuSE: -YaST is well architected, consistent, and simply well implemented. Text/Graphical both well maintained. redhat-config-* seem amatuerish by comparison, rough around the edges, text interfaces not maintained, not consistent throughout, and just not as well architected or thoroughly implemented (i.e. redhat-config-packages seems to get confused on dependency sets after running up2date). Of course, finding the package you want in SuSE is a lot easier than the really crappy redhat tools.
-SuSE seems to simply get it right with respect to patchlevels and packages in a balance between cutting edge and stability. RedHat seems to manage to get the worst of both worlds, picking up some features that are too bleeding edge to keep things perfectly stable (*cough* nptl in 2.4 *cough* gcc2.96 *cough*), while at other times ignoring relatively stable features/software that would really enhance the functionality of the packages.
-autoYaST is many times more flexible than kickstart for network install configurations, with a much better methodology for driver updates/other customizations at install time than kickstart. As a plus, the installed SuSE system remembers information about its install server and YaST will pull from the install server instead of asking for CDs as redhat-config-packages does, a nice plus. You can acheive similar things through a site up2date server and using up2date at least, but it isn't as nicely automatic.
-One point to redhat, rhn web design is easier to deal with than SuSE's support web site. As a corollary, up2date at least seems a little bit better than YOU (YaST Online Update).
-Potential future plus, if Novell works it right, SuSE will have nice directory management features and won't get screwed up, but this is not a current statement (though SuSE 9 seems to be laying the groundwork with some fundamental OpenLDAP stuff..)
What a lot of people on slashdot fail to realize is that the corporate desktop/standard realm is entirely different from home/personal use. I use Gentoo for my personal system, even my personal workstations, but supporting many other servers and clients, you want to be mostly hands off and have the flexibility to delegate day to day administration tasks to less technical administrators as the need arises. The users aren't going to be able to manage their own workstations expertly, and by the same token the company will not be understanding when you end up with an obscure problem after some emerge that leaves services not functional and you are digging around forums.gentoo.org to find answers. If some update does go awry, pointing management at a company allows a cleaner deferral of responsibility, and the vendor has great incentive to help ($$$) whereas your very-expensive downtime is no more valuable than some poor schmuck's home system with misconfigured X as far as forum's/newsgroups/irc/mailing lists are concerned.
Insurance companies will one day acheive it, and not only that, it will be voluntary on behalf of the customers. I don't know how such laws are written, bun unless they prohibit customers voluntarily submitting this information, they will do nothing to stop it.
First, they'll have an opt-in program where you submit evidence of your 'healthy' lifestyle, citing supermarket card tracking permission and such, in exchange, they'll be some sort of healthy living discount/rebate.
After it becomes widespread, they'll hike rates such that the 'discounted' rates will be equal to the normal rates from before. Their market will have volntarily jumped to the desired scheme.
Strictly speaking, no. SDL at best provides fast/convenient methods for getting data to a surface, but no support for non-GL scaling, and no architecture for video codecs (ala gstreamer's intent). So SDL isn't even a good comparison as their scope is limited relative to DirectX.
As far as cross-platform video architectures, don't know of any.
Strictly speaking, that's not true. Direct3D is analogous to OpenGL, DirectX is more analagous to SDL, in other words including multimedia functionality beyond 3D graphics extending to sound, 2d video acceleration, input devices, etc..
Still, the point is valid, SDL/OpenGL is a cross-platform approach that leaves a product's options open.
Actually, that was saying they were planning on doing it, the picture was simply demonstrative (otherwise, it would be called the most piss-poor photoshop job ever.
Yeah, you know it is bad when they are not only ripping their own plotlines off (even going for extra iterations of Wrath of Khan/Nemesis), but when they are ripping off *odd* numbered movies, you know they are desperate...
As much as I think it is not particularly good, why the *hell* would they care about hiring him?
If it was the frequently heard story of some person doing a technological hack to do amazing things with a product I would agree. An engineer that shows remarkable talent and resourcefulness with a company product is valuable to that company.
All this guy has done is leak whatever he could get a hand on to the internet. He didn't dress it up or drive marketing for Apple above and beyond what is normal, simply fed already eager apple fans information. So if you were thinking he was a marketing genius, it doesn't apply. As his hit ratio approaches 100%, in fact, he makes the 'big announcements' anticlimatic. There is nothing to demonstrate his alternative strategy/schedule for announcements is superior to Apple's plan.
I doubt the claims of tortious interference are valid, and the responsibility of the leaks lies with some eager employee(s) using the kid to preserve anonymity, I seriously doubt this 19-year old was proactively coercing anything out of Apple.
Actually, these VDRs are much simpler to use than VCRs have ever been. VCRs have the distinct disadvantage of relying on the user to know exact times, and having to change media frequently.
VDRs offer smooth, easy menus and pull down the listings for the user to browse, and hit record on a program. Even optionally setting up recurring recording based on series title. Browsing listings and directly specifying operations is a *lot* easier than looking up listings, programming a VCR for each and every occurance, and changing the tape frequently as needed.
Fascinating enough, there is a MythMusic module for MythTV for such a purpose.
Other than that, I don't know much (I actually never used MythMusic), but I've found writing such an application in python, using pygame, pyogg (so it can run sans X, unlike mythtv) is actually pretty straightforward.
In general, I agree that spending the little extra on a brand name server is worth it just to have a single point of support. For home computers, it's not too hard to have the time to troubleshoot a handful of systems to determine which component vendor to bitch at for warranty service. If you have a business at stake and other stuff to do, you want to be able to call your vendor and say 'hardware is broke, fix it' as soon as you identify a hardware problem, without taking further time to narrow it down.
Plus, component vendors aren't necessarily set up to very quickly send out things under warranty, sometimes even having to wait for them to receive your broken part, whereas IBM/HP/Dell *have* to provide very expedient service to survive in the marketplace. Component vendors are catering to hobbyists and resellers, of which hobbyists can wait on a component, and resellers are expected to maintain a spare inventory. While generally a good idea for a company to maintain spare inventory for key components, it is not cost effective to maintain such an inventory for everything enterprise-wide.
Additionally, the sheer amount of system testing done on IBM/HP/Dell designs is worth it. Everything is a standard and in an ideal world component testing would be sufficient and everything will fit in perfectly, but it isn't, and component quirks can cause serious problems when interacting with each other.
All that said, the arguments put forth here about functionality are largely moot with respect to the question. As far as redundant PSUs, SCSI configurations, even hot swap fans, you can purchase the components to make that happen homebrew, but the cost difference is so large most people won't justify it (in fact, when they go to price out such a robust solution, they frequently cannot best an IBM/HP/Dell solution, so it is pointless), and end up buying non-redundant, feeding the thought that homebrew robust servers aren't acheivable. For a lot of naive people, redundancy is not viewed as a valuable aspect. They never experience a PSU burn out and therefore the possibility never crosses their mind.
You are pretty much correct about hardware RAID controllers. Essentially, I wouldn't even bother in any homebrew context with hardware RAID unless buying into a major vendor series. Rather I implement software RAID, and then whatever controller that I can physically plug the drives into becomes sufficient, rather than being locked in to a particular vendor/cost. But I always reserve home brew for the home, because the company has a lot more money and doesn't have to care about the cost of beyond-warranty replacement;)
As far as SCSI vs. SATA, it's a strange argument, because both are available to both the homebrew and professional server market. As said previously, you pay extra for any hot swap capability in the homebrew case than you probably do in the brand name case, but the technologies don't really impact the vendor vs. homebrew argument.
That's assuming PCI-32/33Mhz (which, to be fair, is the interface most SATA cards use). PCI64/66Mhz would be 4,224Mbps, and PCI-X 133/64 would be 8,512.
And of course PCI-E: 1x: ~2,500Mbps 4x: ~5,000Mbps 8x: ~10,000Mbps 16x: ~20,000Mbps. (Though typically only used for graphics card, it is a possiblity).
Just showing that expansion card busses can be much faster than they were restricted to in 1994 (when PCI32/33MHz systems first started being realistically available).
To be fair, PPC64 is not pure 64-bit from the ground up, it too is 64bit extensions to a 32-bit architecture (the processors commonly known as the G3 and G4 were only 32-bit, for example, and the exact same binary MacOSX runs on G5). This is made painfully aware by a lot of linux distros on PPC64, where if you fail to explicitly install the 64bit development utilities (or specify to use them at compile time), you'll end up with PPC32 binaries by default, which aside from linking into 64bit code or trying to do it as a kernel module, you'd never know the difference without running file against it. It is very much similar to the x86_64 to x86 relationship, with the nice distinction that it did start life as a 32-bit platform and only has legacy dating back to then, unlike x86_64 which continues legacy from the intel 8-bit computing days, which means a lot more strange quirks that no longer are optimal.
As far as Intel trying to 'bet the farm' to move the world to IA64, I can guarantee it wouldn't have worked no matter how hard Intel tried. Assume hypothetically that Intel had completely ditched x86 and stopped development and production of IA32 chips. At the time Itanium was ready, AMD had already established itself as a pretty viable solution, not as well respected in business, but certainly on the radar. Now when faced with replacement/upgrade of hardware solutions, companies see the poster-child they've grown up to love, Intel, unable to run their existing applications, and therefore a huge cost to migrate in terms of development. Meanwhile, the suboptimal AMD offers fresh, fast x86 processors. Intel's reputation at that point wasn't enough to offset the huge cost of a platform shift. I remember PentiumPro facing harsh criticism and some market problems due to it's slower execution of 16-bit code, and that was when AMD and Cyrix had pretty equal, small, low-budget marketshare.
Besides, Itanium wasn't exactly pure gold. It had strong points (good High Performance Computing mainly), it had weak points (not good at general workstation use, high volume servers, essentially uses that involve widely varying, unpredictable execution paths).
Slashdot Mirror
I know at least Novell/SuSE out of the box nowadays allow LDAP auth configuration. Expect big things to come of Novell/SuSE in the future, but for now they have a good start.
That was hilarious, looking through some of their 'transcipts':
Weather report excertp:
I know this is the way this OS Linux agency lot of clout through northern parts and through eastern parts and you can see how this is this just pushing its weight and keeping eastern parts of Britain but there's no such plan through central and eastern parts he is going to be bringing a lot of snow
Anether except:
the coming into force this week in August tested in the courts they sit unworkable and unenforceable and that is Senator the papers in the Linux Hymon and Atticus of enemy is not found anywhere at eight fifteen what's coming up later this month it he finally asked if more than willing we're looking
I'm glad speech recognition has come so far...
Just get the Top500.org to switch to distributed.net work units as the benchmark, and suddenly things would change...
I would personally like to kick off a bunch of folding at home on a large cluster, too bad time on these clusters is a very precious commodity...
Others have mentioned the prices go lower, but additionally, if you have a blade infrastructure or are building a large scale network and can benefit from blade infrastructure:s /stores/servlet/C ategoryDisplay?categoryId=2586156&storeId=1&catalo gId=-840&langId=-1
http://www-132.ibm.com/webapp/wc
A Power server that is actually priced competitively with the Intel blade servers. True the bladecenter
chassis isn't well suited to some environments, but there is an option.
Oh, and there is always Apple...
Do you want that in real man months, or mythical man months?
The big advantage of vector graphics is not that it reduces size, but that it allows abstraction of pixel-wise screen geometry from GUI design. It allows designers to request a size in inches/centimeters (assuming monitor DPI is known by graphics system, which is generaly the case nowadays), and produce a perfectly scaled image for the screen. You go to 1600x1200 and you don't end up with unreadable dialogs, you end up with really crisp, readable dialogs (allowing user to reduce point size and have it remain readable, so you can still acheive a goal of fitting more stuff on the screen).
As a bonus, editing and particularly the 'print preview' features are then much more easily perfectly representative of what the result will be.
One thought: pressing a key mutes the speaker (or emits alternate sound) before making the noise, noise stops before mute breaks. It's not as if you could hear anything over the tone anyway...
The question *raised*, not begged, begging the question is entirely different..
http://www.eweek.com/article2/0,1759,1758758,00.as p
is the article I read
I don't think 'betting the farm' on on-demand computing is a good idea for sun. The market is there and has money, but I don't think a company like Sun can sustain itself on almost exclusively this.
IBM does this with their servers, but if they didn't have a bunch of customers also buying the systems whole working to help subsidize research and development of the servers, not to mention manufacturing the systems in volume, the revenue from on-demand would not be sustainable.
I found the comment about IBM pensions amusing, but the reality is that while IBM does have armies of people getting paid keeping 1$/cpu hour from sustaining the company, the statement ignores how much broader IBM's scope is compared to Sun's present and plan.
Goes to SuSE, hands down, for a corporate standard.
It really only is a choice between RedHat and SuSE (reasonable support contracts, etc etc), so the points are a comparison between RedHat EL and SuSE:
-YaST is well architected, consistent, and simply well implemented. Text/Graphical both well maintained. redhat-config-* seem amatuerish by comparison, rough around the edges, text interfaces not maintained, not consistent throughout, and just not as well architected or thoroughly implemented (i.e. redhat-config-packages seems to get confused on dependency sets after running up2date). Of course, finding the package you want in SuSE is a lot easier than the really crappy redhat tools.
-SuSE seems to simply get it right with respect to patchlevels and packages in a balance between cutting edge and stability. RedHat seems to manage to get the worst of both worlds, picking up some features that are too bleeding edge to keep things perfectly stable (*cough* nptl in 2.4 *cough* gcc2.96 *cough*), while at other times ignoring relatively stable features/software that would really enhance the functionality of the packages.
-autoYaST is many times more flexible than kickstart for network install configurations, with a much better methodology for driver updates/other customizations at install time than kickstart. As a plus, the installed SuSE system remembers information about its install server and YaST will pull from the install server instead of asking for CDs as redhat-config-packages does, a nice plus. You can acheive similar things through a site up2date server and using up2date at least, but it isn't as nicely automatic.
-One point to redhat, rhn web design is easier to deal with than SuSE's support web site. As a corollary, up2date at least seems a little bit better than YOU (YaST Online Update).
-Potential future plus, if Novell works it right, SuSE will have nice directory management features and won't get screwed up, but this is not a current statement (though SuSE 9 seems to be laying the groundwork with some fundamental OpenLDAP stuff..)
What a lot of people on slashdot fail to realize is that the corporate desktop/standard realm is entirely different from home/personal use. I use Gentoo for my personal system, even my personal workstations, but supporting many other servers and clients, you want to be mostly hands off and have the flexibility to delegate day to day administration tasks to less technical administrators as the need arises. The users aren't going to be able to manage their own workstations expertly, and by the same token the company will not be understanding when you end up with an obscure problem after some emerge that leaves services not functional and you are digging around forums.gentoo.org to find answers. If some update does go awry, pointing management at a company allows a cleaner deferral of responsibility, and the vendor has great incentive to help ($$$) whereas your very-expensive downtime is no more valuable than some poor schmuck's home system with misconfigured X as far as forum's/newsgroups/irc/mailing lists are concerned.
Insurance companies will one day acheive it, and not only that, it will be voluntary on behalf of the customers. I don't know how such laws are written, bun unless they prohibit customers voluntarily submitting this information, they will do nothing to stop it.
First, they'll have an opt-in program where you submit evidence of your 'healthy' lifestyle, citing supermarket card tracking permission and such, in exchange, they'll be some sort of healthy living discount/rebate.
After it becomes widespread, they'll hike rates such that the 'discounted' rates will be equal to the normal rates from before. Their market will have volntarily jumped to the desired scheme.
Yes, yes, and they can call it Virtual Boy! (. . .)
Strictly speaking, no. SDL at best provides fast/convenient methods for getting data to a surface, but no support for non-GL scaling, and no architecture for video codecs (ala gstreamer's intent). So SDL isn't even a good comparison as their scope is limited relative to DirectX.
As far as cross-platform video architectures, don't know of any.
Strictly speaking, that's not true. Direct3D is analogous to OpenGL, DirectX is more analagous to SDL, in other words including multimedia functionality beyond 3D graphics extending to sound, 2d video acceleration, input devices, etc..
Still, the point is valid, SDL/OpenGL is a cross-platform approach that leaves a product's options open.
The did Gentoo for OSX, which certainly has a significant chunk of proprietary components. Think just like fink.
Actually, that was saying they were planning on doing it, the picture was simply demonstrative (otherwise, it would be called the most piss-poor photoshop job ever.
Yeah, you know it is bad when they are not only ripping their own plotlines off (even going for extra iterations of Wrath of Khan/Nemesis), but when they are ripping off *odd* numbered movies, you know they are desperate...
As much as I think it is not particularly good, why the *hell* would they care about hiring him?
If it was the frequently heard story of some person doing a technological hack to do amazing things with a product I would agree. An engineer that shows remarkable talent and resourcefulness with a company product is valuable to that company.
All this guy has done is leak whatever he could get a hand on to the internet. He didn't dress it up or drive marketing for Apple above and beyond what is normal, simply fed already eager apple fans information. So if you were thinking he was a marketing genius, it doesn't apply. As his hit ratio approaches 100%, in fact, he makes the 'big announcements' anticlimatic. There is nothing to demonstrate his alternative strategy/schedule for announcements is superior to Apple's plan.
I doubt the claims of tortious interference are valid, and the responsibility of the leaks lies with some eager employee(s) using the kid to preserve anonymity, I seriously doubt this 19-year old was proactively coercing anything out of Apple.
Actually, these VDRs are much simpler to use than VCRs have ever been. VCRs have the distinct disadvantage of relying on the user to know exact times, and having to change media frequently.
VDRs offer smooth, easy menus and pull down the listings for the user to browse, and hit record on a program. Even optionally setting up recurring recording based on series title. Browsing listings and directly specifying operations is a *lot* easier than looking up listings, programming a VCR for each and every occurance, and changing the tape frequently as needed.
Fascinating enough, there is a MythMusic module for MythTV for such a purpose.
Other than that, I don't know much (I actually never used MythMusic), but I've found writing such an application in python, using pygame, pyogg (so it can run sans X, unlike mythtv) is actually pretty straightforward.
I see two SSIDs at my house aside from my own, one is a venerable 'linksys', never protected...
The other (WEP protected and such), changes ever so often...
It was 'Voltron', then it was 'MyLittlePony'. I sense a theme....
From one of my favorite Nigerian spam counter-scams:
"I Take IT up the arse."
In general, I agree that spending the little extra on a brand name server is worth it just to have a single point of support. For home computers, it's not too hard to have the time to troubleshoot a handful of systems to determine which component vendor to bitch at for warranty service. If you have a business at stake and other stuff to do, you want to be able to call your vendor and say 'hardware is broke, fix it' as soon as you identify a hardware problem, without taking further time to narrow it down.
;)
Plus, component vendors aren't necessarily set up to very quickly send out things under warranty, sometimes even having to wait for them to receive your broken part, whereas IBM/HP/Dell *have* to provide very expedient service to survive in the marketplace. Component vendors are catering to hobbyists and resellers, of which hobbyists can wait on a component, and resellers are expected to maintain a spare inventory. While generally a good idea for a company to maintain spare inventory for key components, it is not cost effective to maintain such an inventory for everything enterprise-wide.
Additionally, the sheer amount of system testing done on IBM/HP/Dell designs is worth it. Everything is a standard and in an ideal world component testing would be sufficient and everything will fit in perfectly, but it isn't, and component quirks can cause serious problems when interacting with each other.
All that said, the arguments put forth here about functionality are largely moot with respect to the question. As far as redundant PSUs, SCSI configurations, even hot swap fans, you can purchase the components to make that happen homebrew, but the cost difference is so large most people won't justify it (in fact, when they go to price out such a robust solution, they frequently cannot best an IBM/HP/Dell solution, so it is pointless), and end up buying non-redundant, feeding the thought that homebrew robust servers aren't acheivable. For a lot of naive people, redundancy is not viewed as a valuable aspect. They never experience a PSU burn out and therefore the possibility never crosses their mind.
You are pretty much correct about hardware RAID controllers. Essentially, I wouldn't even bother in any homebrew context with hardware RAID unless buying into a major vendor series. Rather I implement software RAID, and then whatever controller that I can physically plug the drives into becomes sufficient, rather than being locked in to a particular vendor/cost. But I always reserve home brew for the home, because the company has a lot more money and doesn't have to care about the cost of beyond-warranty replacement
As far as SCSI vs. SATA, it's a strange argument, because both are available to both the homebrew and professional server market. As said previously, you pay extra for any hot swap capability in the homebrew case than you probably do in the brand name case, but the technologies don't really impact the vendor vs. homebrew argument.
That's assuming PCI-32/33Mhz (which, to be fair, is the interface most SATA cards use).
PCI64/66Mhz would be 4,224Mbps, and PCI-X 133/64 would be 8,512.
And of course PCI-E:
1x: ~2,500Mbps
4x: ~5,000Mbps
8x: ~10,000Mbps
16x: ~20,000Mbps. (Though typically only used for graphics card, it is a possiblity).
Just showing that expansion card busses can be much faster than they were restricted to in 1994 (when PCI32/33MHz systems first started being realistically available).
To be fair, PPC64 is not pure 64-bit from the ground up, it too is 64bit extensions to a 32-bit architecture (the processors commonly known as the G3 and G4 were only 32-bit, for example, and the exact same binary MacOSX runs on G5). This is made painfully aware by a lot of linux distros on PPC64, where if you fail to explicitly install the 64bit development utilities (or specify to use them at compile time), you'll end up with PPC32 binaries by default, which aside from linking into 64bit code or trying to do it as a kernel module, you'd never know the difference without running file against it. It is very much similar to the x86_64 to x86 relationship, with the nice distinction that it did start life as a 32-bit platform and only has legacy dating back to then, unlike x86_64 which continues legacy from the intel 8-bit computing days, which means a lot more strange quirks that no longer are optimal.
As far as Intel trying to 'bet the farm' to move the world to IA64, I can guarantee it wouldn't have worked no matter how hard Intel tried. Assume hypothetically that Intel had completely ditched x86 and stopped development and production of IA32 chips. At the time Itanium was ready, AMD had already established itself as a pretty viable solution, not as well respected in business, but certainly on the radar. Now when faced with replacement/upgrade of hardware solutions, companies see the poster-child they've grown up to love, Intel, unable to run their existing applications, and therefore a huge cost to migrate in terms of development. Meanwhile, the suboptimal AMD offers fresh, fast x86 processors. Intel's reputation at that point wasn't enough to offset the huge cost of a platform shift. I remember PentiumPro facing harsh criticism and some market problems due to it's slower execution of 16-bit code, and that was when AMD and Cyrix had pretty equal, small, low-budget marketshare.
Besides, Itanium wasn't exactly pure gold. It had strong points (good High Performance Computing mainly), it had weak points (not good at general workstation use, high volume servers, essentially uses that involve widely varying, unpredictable execution paths).