Note that Java1.4.2 does something very slick with floating point stuff on PIII and P4: it offloads float arithmetic to the SSE ALU, and double arithmetic on the SSE2 unit (if present), which do speed up floating point a lot -I think more than ot the box FP generation from the MS C++ compilers (though the intel vtune and codeplay vectorizing C++ compilers have an edge). So they are not neglecting the old FP performance. This increases its value for scientific work, like grid based stuff.
That doesnt affect the validity of any of your other arguments, however.
Steve
(who as an Axis developer works on one of those bloated web service implementations)
Funny you should mention date and time. XML schema's dateTime type requires you to spec timezone in the payload, what is it "2003-05-23 15:42:00 Z" style, but.NET framework ignores the zone. Apache Axis doesnt. Result, times sent between Axis and.NET end up 8 or 9 hours out in Pacific time. Hence my use of time_t marshalled as longs.
Could be. I saw one presentation on Linux by a marketing person who said that linux is getting a hold in the US office desktop at the high end -the CAD user, the scientist, the software developer -and was competing with Unix workstations, and these people can afford VMware and Crossover Office to host winapps.
But the third world opportunity was where the linux desktop could go widespread, becaue these people can make do with the current suite of Office.org apps and existing tools. Slick.
NB, I don't get the bit in the article that implies that HP cant handle the demand so they were talking to Dell. They all come from the same factories in Taiwan, so what is the issue.
Big hardware costs in a laptop are display, HDD and CD (say $25)/floppy ($5), plus CPU and memory.
a pure linux laptop can get away with a lower power CPU, less HDD and by the CNET report, the CD. That leaves display.
The nice thing is that not only can they get rid of the MS tax, but they can include a full suite of tools: gimp, OpenOffice.org, evolution, and at a price point that you cannot do legally with a windows box (of course, you can get all the windows stuff illegally at discount rates in Asia: Redhat CDs sell for the same as Windows Server CDs, $5 or so).
The interesting thing is the network effect: if lots of people start to use linux distros, then it makes sense for even people with $$ to use it, offices to use it, etc, as it is what everyone will understand. And if one developing country follows the Open Source path, it can set an example to others.
we already have Shell owning Nigeria, and some Peru/Ecuador war in the sixties or seventies was effectively a proxy war for competing oil companies, each with a different government contract. And of course the US is subject to the current Oil and Gas Administration.
I dont see the computing industry playing such strategic games, though I think Disney may already own Southern California.
I remember sharing a unix cluster when someone logged in as root accidentally did a rm -rf/bin instead of/tmp/bin
Suddenly everyone in our corridor starts seeing messages like "ls: not found". We felt that was pretty major.
You are right -we do need better safety checks not just for dumb newbies, but for experience people who get things wrong every so often. Just like I like to drive mountain roads fast, but I still prefer those with crash barriers:)
you know, I wonder how many of these big F50 companies have to move to Linux on the desktop before we get a 'critical mass' of Linux desktops. Once one 80K seat company moves to linux+openoffice and gets the cost savings (no licensing, no SQL slammer), they probably follow up by abandoning NT server (no exchange, no SQL server, no IIS) and the savings ramp up.
Then it becomes an example for other companies to follow -or suffer a cost penalty for not doing so.
I agree. our IT dept is fussy as anything over windows licenses, even borrowing the CDs for an install is a sign-out sign in effort.
But Linux? they have a stack of the disks they burned themselves, preconfigured with the apps you need. I just go down and ask for one, install it where I like. No audits, no license tracking, nothing.
If you can assert statements (a!=b) at the start of the code, and the compiler uses these assertions then it can act on them. but nobody does, no compiler supports it, etc, etc. Visual C++ does support this trick, but I have never encountered its use in a real project.
I'd be happy with cache prefetch hints in both Java and C#, and portable hinting in C++ (you can do it with ugly macros)
file stores are legal, but indexing it wont be. Maybe that will extend to inode tables.
Or they could just ban ID3 tags and the.mp3 file extension. Right after they turn google off.
(as an aside, sharing files using windows file sharing? blech. Better to use HTTP on a local apache server of some sort) If they are CS students, the RIAA should put an injunction against them for bad application design.
our philosophy on server-apps was 'a box that works so well, you forget where it lives'; I think novell were the first people to pull that off in products.
yes, I've heard the same. DBAs are their own little power group in many organisations, somehow they are more important than the mere programmers, though IMO any product that needs full time admin & tuning is a failure in TCO terms.
XML databases may be the answer. Also, JDO will bind to a OODBMS under the cover -your app doesnt care, but the whole thing just works better. Compare that with the entire Enterprise Java Bean persistence story
good point. do decent object-repositories and half the data binding problems in the book go away. Plus EJB and IBM Global Services consulting revenue getting websphere to work go away as well.
So why have all OODBMS attempts died. Was it because the world wasnt ready (and are they now), or is it because data in RDBMS have better flexibility of re-use?
I didnt notice the typos, but I did find it hard to read in places. All those cricket examples - I hate cricket.
Also, although I write a lot of server side code, I dont have so many issues with data-binding as he has. This is very much the 'design patterns for enterprise apps that talk to databases', not 'design patterns for high availability server-side applications' in general.
Sun dont want to let jboss near the tests, but apache forced them in to it last year, when they threatened to stop working with Sun unless open source projects got access to the test kits (not the reference impls) under various terms and conditions. Axis is the first Apache project to do this, and it has been beneficial. Except when their tests are different from the spec, which happens sometimes.
A certified JBoss (it'll happen eventually) will hurt BEA and IBM, and Sun via licensing fees. But it is good for Java.
What is the problem? Winmodem (as usual?) or sound card.
When i worked for a PC vendor one problem we had with the linux support was that if you get some bit of kit on NDA (like say the CD controller that works while the main PC is off by injecting ATA commands on the powered up CD-ROM drive), you cant publish the docs. And you dont have the time to do the ports yourself. Really we want full disclosure from the ODM vendors in taiwan that design the notebooks.
depends what you call desktops. I just managed to get the org chart to fund an HP 3.06GHz 'workstation' desktop box with 1GB ram and a 72MHz SCSI, with the redhat option. Not that I'll necessarily keep redhat on, but if you are going to pay a tax for an OS that you'll scrub, I'd rather donate it to a linux charity than, say, our friends in Redmond. Though I may dual boot it to Win2K just to run GTA III on occasions.
Still, consumer boxes are different. If Lindows takes off, then you might see it.
We all share a common enemy: the RIAA/MPAA consortium.
Every country is at risk from these pressure groups. What is worse, the government's urges make copyright laws consistent across continents threatens us, because they always move them upwards, dont they. EU is looking at the DMCA, the UK raised its copyright duration from 50 to 70 years to be consistent with europe,....etc etc.
My fear about this tax is the precedent: why not tax ISPs next. Everyone knows 73% of all network traffic is pirate music:)
I had a big argument with a survey company about this in the UK -they (one of the bug survey companies) called me and I immediately asked why they were ignoring my entry on the don't call list.
First they came in with a bogus 'the computer dials random numbers, so it cant use the list' excuse, as if their app cant check the list after generating random numbers, then they came with a more realistic 'statistical validity of the sample' argument.
So I said 'but doesnt my opting out affect your sample's validity' and hung up...
no, but there is no point. They'd be better of buying Sun, or Apple. Sun for Java, Apple for power PC systems people still want.
SCO are like microsoft: a company trying to make money out of middleware. It seemed a good idea when everyone was prepared to pay for middleware, but now that's just, well, silly. Linux, Apache and to an extent the OSS Java projects (apache, jboss and sforge hosted) have helped reduce the value of middleware. Which leaves cash for the hardware vendors and the consultants. Which is where IBM fits in.
Also they've missed out the fact that bicycles are lower cost and often more practical than a luxury car, better suited to european cities and third world countries, and perhaps the future of transport for humanity:)
I think I have access to the SysV code. I also have access to VMS somewhere, and NT. Do I look at them? Why bother? All you can learn is what not to do.
IBM probably have a metric arseload of patents that matter to linux, but like the other players (HP, sun), they are too smart to want to piss off the OSS community. SCO have just declared themselves a legitimate target.
The hard part is picking on something SCO have done that is not a general linux fsf or other OSS contrib. SCO have it easy if they attack the 'enterprise' stuff IBM added.
Nokia did some work on context aware cellphones -added IR and light detection to see if could tell when it was in a briefcase and needed to ring louder. Of course, a dark cinema looks almost the same as a dark briefcase to a fairly stupid cellphone, which is why you dont see this in any products...
Slashdot Mirror
Note that Java1.4.2 does something very slick with floating point stuff on PIII and P4: it offloads float arithmetic to the SSE ALU, and double arithmetic on the SSE2 unit (if present), which do speed up floating point a lot -I think more than ot the box FP generation from the MS C++ compilers (though the intel vtune and codeplay vectorizing C++ compilers have an edge). So they are not neglecting the old FP performance. This increases its value for scientific work, like grid based stuff.
That doesnt affect the validity of any of your other arguments, however.
Steve
(who as an Axis developer works on one of those bloated web service implementations)
better yet, use the Ant1.5 wrapper to wsdl.exe,
<wsdltodotnet> so that you can automate all this as part of your ant based build process.
Funny you should mention date and time. XML schema's dateTime type requires you to spec timezone in the payload, what is it "2003-05-23 15:42:00 Z" style, but .NET framework ignores the zone. Apache Axis doesnt. Result, times sent between Axis and .NET end up 8 or 9 hours out in Pacific time. Hence my use of time_t marshalled as longs.
Could be. I saw one presentation on Linux by a marketing person who said that linux is getting a hold in the US office desktop at the high end -the CAD user, the scientist, the software developer -and was competing with Unix workstations, and these people can afford VMware and Crossover Office to host winapps.
But the third world opportunity was where the linux desktop could go widespread, becaue these people can make do with the current suite of Office.org apps and existing tools. Slick.
NB, I don't get the bit in the article that implies that HP cant handle the demand so they were talking to Dell. They all come from the same factories in Taiwan, so what is the issue.
Big hardware costs in a laptop are display, HDD and CD (say $25) /floppy ($5), plus CPU and memory.
a pure linux laptop can get away with a lower power CPU, less HDD and by the CNET report, the CD. That leaves display.
The nice thing is that not only can they get rid of the MS tax, but they can include a full suite of tools: gimp, OpenOffice.org, evolution, and at a price point that you cannot do legally with a windows box (of course, you can get all the windows stuff illegally at discount rates in Asia: Redhat CDs sell for the same as Windows Server CDs, $5 or so).
The interesting thing is the network effect: if lots of people start to use linux distros, then it makes sense for even people with $$ to use it, offices to use it, etc, as it is what everyone will understand. And if one developing country follows the Open Source path, it can set an example to others.
Something to watch, perhaps.
we already have Shell owning Nigeria, and some Peru/Ecuador war in the sixties or seventies was effectively a proxy war for competing oil companies, each with a different government contract. And of course the US is subject to the current Oil and Gas Administration.
I dont see the computing industry playing such strategic games, though I think Disney may already own Southern California.
I remember sharing a unix cluster when someone logged in as root accidentally did a rm -rf /bin instead of /tmp/bin
:)
Suddenly everyone in our corridor starts seeing messages like "ls: not found".
We felt that was pretty major.
You are right -we do need better safety checks not just for dumb newbies, but for experience people who get things wrong every so often. Just like I like to drive mountain roads fast, but I still prefer those with crash barriers
you know, I wonder how many of these big F50 companies have to move to Linux on the desktop before we get a 'critical mass' of Linux desktops. Once one 80K seat company moves to linux+openoffice and gets the cost savings (no licensing, no SQL slammer), they probably follow up by abandoning NT server (no exchange, no SQL server, no IIS) and the savings ramp up.
Then it becomes an example for other companies to follow -or suffer a cost penalty for not doing so.
I agree. our IT dept is fussy as anything over windows licenses, even borrowing the CDs for an install is a sign-out sign in effort.
But Linux? they have a stack of the disks they burned themselves, preconfigured with the apps you need. I just go down and ask for one, install it where I like. No audits, no license tracking, nothing.
If you can assert statements (a!=b) at the start of the code, and the compiler uses these assertions then it can act on them. but nobody does, no compiler supports it, etc, etc. Visual C++ does support this trick, but I have never encountered its use in a real project.
I'd be happy with cache prefetch hints in both Java and C#, and portable hinting in C++ (you can do it with ugly macros)
file stores are legal, but indexing it wont be. Maybe that will extend to inode tables.
.mp3 file extension. Right after they turn google off.
Or they could just ban ID3 tags and the
(as an aside, sharing files using windows file sharing? blech. Better to use HTTP on a local apache server of some sort) If they are CS students, the RIAA should put an injunction against them for bad application design.
I guess they do. As do WinNT cluster owners.
our philosophy on server-apps was 'a box that works so well, you forget where it lives'; I think novell were the first people to pull that off in products.
yes, I've heard the same. DBAs are their own little power group in many organisations, somehow they are more important than the mere programmers, though IMO any product that needs full time admin & tuning is a failure in TCO terms.
XML databases may be the answer. Also, JDO will bind to a OODBMS under the cover -your app doesnt care, but the whole thing just works better. Compare that with the entire Enterprise Java Bean persistence story
good point. do decent object-repositories and half the data binding problems in the book go away. Plus EJB and IBM Global Services consulting revenue getting websphere to work go away as well.
So why have all OODBMS attempts died. Was it because the world wasnt ready (and are they now), or is it because data in RDBMS have better flexibility of re-use?
I didnt notice the typos, but I did find it hard to read in places. All those cricket examples - I hate cricket.
Also, although I write a lot of server side code, I dont have so many issues with data-binding as he has. This is very much the 'design patterns for enterprise apps that talk to databases', not 'design patterns for high availability server-side applications' in general.
Sun dont want to let jboss near the tests, but apache forced them in to it last year, when they threatened to stop working with Sun unless open source projects got access to the test kits (not the reference impls) under various terms and conditions. Axis is the first Apache project to do this, and it has been beneficial. Except when their tests are different from the spec, which happens sometimes.
A certified JBoss (it'll happen eventually) will hurt BEA and IBM, and Sun via licensing fees. But it is good for Java.
What is the problem? Winmodem (as usual?) or sound card.
When i worked for a PC vendor one problem we had with the linux support was that if you get some bit of kit on NDA (like say the CD controller that works while the main PC is off by injecting ATA commands on the powered up CD-ROM drive), you cant publish the docs. And you dont have the time to do the ports yourself. Really we want full disclosure from the ODM vendors in taiwan that design the notebooks.
-steve
depends what you call desktops. I just managed to get the org chart to fund an HP 3.06GHz 'workstation' desktop box with 1GB ram and a 72MHz SCSI, with the redhat option. Not that I'll necessarily keep redhat on, but if you are going to pay a tax for an OS that you'll scrub, I'd rather donate it to a linux charity than, say, our friends in Redmond. Though I may dual boot it to Win2K just to run GTA III on occasions.
Still, consumer boxes are different. If Lindows takes off, then you might see it.
ok, we'll keep our free health care and speed limits which your SUVs and minivans cant handle, even if you were allowed to :)
Actually you should feel threatened by this move: if they get away with it in Germany, they will target other countries. Be prepared to resist.
-steve
(who considers windows an extra tax on PCs)
We all share a common enemy: the RIAA/MPAA consortium.
....etc etc.
:)
Every country is at risk from these pressure groups. What is worse, the government's urges make copyright laws consistent across continents threatens us, because they always move them upwards, dont they. EU is looking at the DMCA, the UK raised its copyright duration from 50 to 70 years to be consistent with europe,
My fear about this tax is the precedent: why not tax ISPs next. Everyone knows 73% of all network traffic is pirate music
I had a big argument with a survey company about this in the UK -they (one of the bug survey companies) called me and I immediately asked why they were ignoring my entry on the don't call list.
First they came in with a bogus 'the computer dials random numbers, so it cant use the list' excuse, as if their app cant check the list after generating random numbers, then they came with a more realistic 'statistical validity of the sample' argument.
So I said 'but doesnt my opting out affect your sample's validity' and hung up...
no, but there is no point. They'd be better of buying Sun, or Apple. Sun for Java, Apple for power PC systems people still want.
SCO are like microsoft: a company trying to make money out of middleware. It seemed a good idea when everyone was prepared to pay for middleware, but now that's just, well, silly. Linux, Apache and to an extent the OSS Java projects (apache, jboss and sforge hosted) have helped reduce the value of middleware. Which leaves cash for the hardware vendors and the consultants. Which is where IBM fits in.
Also they've missed out the fact that bicycles are lower cost and often more practical than a luxury car, better suited to european cities and third world countries, and perhaps the future of transport for humanity :)
I think I have access to the SysV code. I also have access to VMS somewhere, and NT. Do I look at them? Why bother? All you can learn is what not to do.
yeah, you dont mess with IBM: you cross license.
IBM probably have a metric arseload of patents that matter to linux, but like the other players (HP, sun), they are too smart to want to piss off the OSS community. SCO have just declared themselves a legitimate target.
The hard part is picking on something SCO have done that is not a general linux fsf or other OSS contrib. SCO have it easy if they attack the 'enterprise' stuff IBM added.
Nokia did some work on context aware cellphones -added IR and light detection to see if could tell when it was in a briefcase and needed to ring louder. Of course, a dark cinema looks almost the same as a dark briefcase to a fairly stupid cellphone, which is why you dont see this in any products...