Evolution really needs a way of modifying keymappings easily (if it already has one, I'd love to hear about it - haven't found it). What pissed me off when I upgraded was the move from "q" to "`" for turning the preview pane on and off. I do that frequently, and I still haven't found a way of generating backtick in a way that Evolution recognizes, so currently I'm stuck using the menus, slowing me down considerably.
The vendors are heavily represented in the W3C, and the discussions are open. Anyone can start implementing the specs way before they're final. It's not like people are waiting until the W3C releases the specs before starting to work on implementations. Much of the discussions in the W3C are based on what issues pop up as people try to implement various aspects of working drafts.
Uhm. Dam projects are generally considered extremely destructive to the local ecology, so I really wonder what side effects whoever claims that no negative side effects have been seen have been looking for.
Did you really read the article? The article clearly point out that the technology used is vastly different from the French and Canadian plants, are more environmentally friendly, and doesn't depend on huge tidal differences.
Apparently the fjord in question is considered good for this pilot project especially because it doesn't have much high velocity currents. High speed currents tends to cause strong eddies etc., which they apparently would prefer to not have to deal with.
They're not being built in Oslo. Contrary to what some people seem to believe, Oslo is not that far north;). The plant is being built near Hammerfest, far north of the arctic circle.
Read the article more closely. It isn't Oslo that is getting the generator, but Hammerfest. Just like most news sources, the place listed in the start of the article is not the place the article is about, but the origin of the story.
I didn't say I'd spend less than $40 on a SCSI controller. I certainly won't, and I'll spend quite a bit more on a good drive. But that money will be a hell of a lot better spent than spending $40 on more memory for a box that already has 128MB and practically never swaps.
As for what I've used to run on a 16MB box, you're right: No Mozilla, Open Office or anything related to Java. That wasn't the point. The point is that even if I just start X on my box at home, with 128MB of RAM, and still have more than 64MB left any heavy disk activity will still make the box slow to a crawl, even for basic things like moving windows. When doing the same on an old, classic Pentium box I used to run, with 16MB RAM and a good SCSI controller and disk, the same thing was never a problem.
You demonstrate very clearly that you don't know what the hell you're talking about. The issue isn't IO throughput - IO throughput is good in my current system, but at the expense of good interactivity.
And most home users DO notice this, whenever they try to start an application. Try starting OpenOffice on a system with a crappy IDE controller, regardless of the amount of RAM. Try the same with Mozilla. The only benefit extra RAM will buy you in a system that hardly ever swaps in the first place is caching of more apps in the buffercache, but that will only help you if you keep starting the same apps. In my case I tend to load Mozilla and Evolution and keep them running more or less permanently (and my system still rarely swaps, in fact to verify that swapping didn't affect the performance much, I turned off my swap partition and had no problems but the performance issues were still as bad).
If you truly believe that anyone would see significant improvements from adding more memory when their system rarely swaps, and you rarely start any large apps more than once after booting the machine, then you're deluded.
Yes, for your example, if someone is running a setup that waste tons of memory at startup, they might benefit. But that's not the case here, and not even that common - I've never come across a manufacturer that don't adapt their setup to perform reasonably with the amount of memory it ships with, including keeping the amount of crap that gets loaded at startup down in "low memory" machines.
And if you think I'm complaining about not wanting to spend $40, you obviously can't even read properly, considering that what I've been saying is that I'd prefer to spend my money on getting a proper SCSI setup, and that will cost significantly more than $40. But contrary to buying more meory, it won't be money straight down the toilet.
Fact is, we don't NEED more memory, unless we keep on running more and more bloated software. I run my share of bloated software, but certainly not enough to need any more than 128MB, and more most of what I do I'd have no problems working with much less without noticing. Same goes for CPU. My current box has a 1GHz Athlon, but I'm practically never close to 100% CPU utilization - even passing 20% would be uncommon. That's not what slow my system down. Disk is. Not primarily transfer rates, though they suck too (though the gap in transfer rates between IDE and SCSI is closing). Whats much more important is latency, and for cheap IDE controllers the way they slow the rest of the system to a crawl whenever I do IO.
This is a growing problem with low cost PC's: All the emphasis is on raising CPU speeds etc., and the way they finance that is by cutting corners all over the place, including junk like Winmodems (which has the same issues - ever tried a Conexant based one under Linux? It makes the box extremely unresponsive while dialling, for instance) and cheap IDE. Those things are quickly becoming much more noticable on low cost PC's than memory limitations.
No, I won't be amazed at the performance increase, because my system rarely touch swap, and regularly have 40-50 MB allocated to buffers. I've happily run Linux with X in 16MB with good performance on systems with a decent SCSI disk.
Your attitude is a good demonstration of what's wrong with advances in computing these days. The main chokepoints for an ordinary home desktop is shitty IO subsystems and bloated software, not RAM or CPU.
When I DO get around to upgrading my machine, a good SCSI controller and a decent disk will be far higher on my list than more memory.
Amiga Basic was a Microsoft product, so I'm not surprised. It was also slower at many tasks on my Amiga 500 than my Commodore 64 were...
As for stability, the programs I used were stable, and even when I was writing my own assembly programs I rarely crashed the machine. But the very nature of a machine without MMU is that it is trivial to crash and programs needs to be written to a much higher quality to be usable...
But I still like the Amiga. It still has innovative features that I aren't in widespread use, such as Datatypes, a sane, well working DLL system, a standard scripting language for automating applications that made it easy to expose functionality from your app (Arexx), Assigns (aliases for paths that shows up in file selector windows etc.), Screens (thought some version of Enlightenment apparently introduced screens that works approximately like the Amiga, except for the ability to have different resolutions on different screens) and more.
I like Linux, and it's certainly more solid, but from a design point of view the Amiga was much more well structured and clean than any Linux distribution I've used (and don't get me started on Windows;)
Ah, but that combination means that you can do quantum cryptography that is unbreakable (as in there is no way of attacking the algorithm itself - you can still attempt to bruteforce it, as you can with any encoded data you get hold of):
First you transmit a random stream of bits, and you verify which bits weren't eavesdropped on. Then you use the bits you know isn't in the hands of an adversary as a onetime pad for the real message.
The whole point of RAID is to combine unreliable parts into something that's more reliable than each of the components. But you have to make a judgement as to how reliable you want something, and cost.
Often you can choose between a free standing drive that's high quality and expensive, or several cheap drives combined in a RAID at the same price. If the chance of the single drive failing any given day is 1% (which is of course ridiculously high), and the alternative is two cheaper drives mirrored with 5% likelyhood of either of them dying, then provided that the RAID controller is as reliable as the controller for the other drive, the mirror solution is 5%*5% = 0.25% likely to lose data every day, and is indeed the better choice.
RAID doesn't automatically give you more reliability to a single drive, but for mirroring and RAID-5 it gives you more reliability than a single drive of the same type that you build your RAID array with. Sometimes you use RAID in order to get high reliability, sometimes you use it to be able to cut cost but retain reasonable reliability.
Sounds like in the projects you mentioned, they had equated RAID with reliability, and then gone with a cheap solution believing that RAID is
RAID regardless of the quality of the individual components, not because
of an inherent problem with IDE RAID.
I HATE IDE. Both my desktops (work, home) are crippled by shitty IDE controllers and drivers. Moving the data fast enough isn't all of it - the thing that gets to me is the way the machine almost freeze up whenever disk activity is high. That's why I prefer SCSI wherever budgets allow (and I'll certainly be adding a SCSI drive to my home machine soon).
It's depressing to have a 1GHz Athlon system with 128MB RAM and see system responsiveness being worse that classic Pentium servers I used 7-8 years ago whenever there's disk activity...
Despite that, whatever harddrive you are running is quite likely to make use of one or more patents licensed from IBM, or is likely to have been manufacturing using a process licensed from IBM. They may not be the most reliable drive manufacturer, but they are on the leading edge of developing harddrive technology.
In the past they've even licensed their technology to their partners before they started using in their own manufacturing in some cases, in order to ensure acceptance in the market ("you buy this tech, and you'll have an advantage over everyone for a while, us included").
The Amiga OS in itself was rock stable. The problems (and the guru meditations:) came whenever you used an application that contained any bugs - which meant quite frequently. So yes, the OS was stable, the system as a whole wasn't thanks to the lack of memory protection. You don't have to go back that far though, to find Amiga users arguing that memory protection was a waste of resources and shouldn't be added to the OS.
Thats bullshit, and shows that you have no idea what you're talking about. No Amiga model was even capable of a text mode, and that was one of the things that really annoyed me when I got my first Amiga, as updating a screenful of text was sometimes slower than on my C64 (got better once people started learning how to exploit the platform, though).
The did walk away from it, but felt obliged to let people know how Microsoft had tried to use them for their PR ploy. Given the way Microsoft treated them I think that's entirely within their rights.
How do you think a company that have 40 billion dollars should act? Surely they wouldn't be giving away money left and right. If anything, the more devious, cynical and underhanded a company is, the more likely the company would be to have 40 billion, provided they also have the marketing skill to appear like great benevolent innovators.
Try asking the American public what they think about Microsoft and Bill Gates, and see the value of good marketing.
Microsoft aren't required to show anyone respect, but neither are anyone else.However if you don't show people respect, you have to expect people not to like it and to not show you any respect either.
As for your question, read the open letter from Schoolnet to Microsoft. Schoolnet got an offer of cheap Acer machines, but Acer's distributor was contractually bound by Microsoft to deliver the machines with Windows preloaded (presumably the same type of deal that Microsoft are no longer allowed to use in the US). When approached, Microsoft would not offer them to donate the licenses, costing about 9000 USD, but instead offered Office licenses worth about 2000 USD instead.
Schoolnet then entered into negotiations with Microsoft as they "would be happy to provide Microsoft with an opportunity to develop a potential alternative to our viable Open Source LTSP refurbished LAN and stand-alone Linux-PC solutions" (from Schoolnets letter). In other words: They'll incure 9000 USD worth of Windows license fees if they buy the Acer PC's, so they wanted to see whether they could actually get any use out of it.
However Microsoft tried stiffing them by offering the software for "free" except for some unspecified R&D costs.
Schoolnet further wrote: " I have, from the very beginning made it VERY clear that SchoolNet has NO desire to REPLACE Linux with Microsoft, but would be happy to accommodate an AFFORDABLE Microsoft diskless refurbished thin-client LAN alternative for potential use in areas where Microsoft distributors would be able to provide technical support to such proprietary Microsoft LAN alternatives."
It should be pretty clear that they're ready to use what they have and try to get just the hardware donated, or even to buy the hardware, but Microsoft tried to trick them into a deal they could gain a PR advantage from but that would end up being costly to Schoolnet without giving them much to show for it.
Netproject is a membership organization for companies that wish to cooperate on strategies for use of Open Source and open standards - it's not a normal consulting company. As such they are unlikely to be looking at more than recouping costs - they're more interesting in providing a workable strategy that will allow their members to participate in the deployment of open source based systems later. Among the people cooperating with Netproject you'll find Alan Cox and Eric Raymond for instance.
So "only" 250,000 euros might be more than enough to produce a report that provide workable strategies.
Cambridge most certainly does pay Microsoft for it under Microsofts Select 4 licensing. See this page for some superficial information about (no prices etc).
Look into where software engineers are employed. Only an extremely tiny minority is working on general purpose programs for a company that publish software commercially. The vast majority of software development is proprietary development done to the specifications of a company. They won't go away because there's suddenly open source projects available.
Sure, many of these projects will be smaller because they can draw on open source components. But similarly many more projects will suddenly make economic sense exactly because the company can cut the projected cost by improving an existing open source project instead of building something from scratch.
For software projects that aren't directly revenue generating, that is a critical factor. Open source has already been the enabler for several projects I've worked on: Thanks to Linux and other open source we've been able to cut the cost of the projects enough that we've been able to spend money elsewhere instead - including on further software development of features we would've gone without if we didn't have access to open source.
We're not creating a lack of jobs - we're widening the market by creating a platform of commodity software that can be customized cheaply enough to enable projects that would otherwise be dead long before getting even to a requirements specification because of cost issues.
Modifying well tested commodity software is also a smaller risk, and less complex, and as such should hopefully in the long run reduce the failure rates of IT projects, which would further increase the chances of getting projects approved.
And of course his donations of large quantities of shares he'd be unable to sell at reasonable prices has nothing to do with any wish of reducing his tax bill... None, whatsoever - he couldn't possibly have other motivations for his donations...
I'm not saying the end result isn't good, but don't think that he's doing this purely out of good will considering the amount of money he can save by donating shares this way.
Slashdot Mirror
Evolution really needs a way of modifying keymappings easily (if it already has one, I'd love to hear about it - haven't found it). What pissed me off when I upgraded was the move from "q" to "`" for turning the preview pane on and off. I do that frequently, and I still haven't found a way of generating backtick in a way that Evolution recognizes, so currently I'm stuck using the menus, slowing me down considerably.
The vendors are heavily represented in the W3C, and the discussions are open. Anyone can start implementing the specs way before they're final. It's not like people are waiting until the W3C releases the specs before starting to work on implementations. Much of the discussions in the W3C are based on what issues pop up as people try to implement various aspects of working drafts.
Uhm. Dam projects are generally considered extremely destructive to the local ecology, so I really wonder what side effects whoever claims that no negative side effects have been seen have been looking for.
Did you really read the article? The article clearly point out that the technology used is vastly different from the French and Canadian plants, are more environmentally friendly, and doesn't depend on huge tidal differences.
Apparently the fjord in question is considered good for this pilot project especially because it doesn't have much high velocity currents. High speed currents tends to cause strong eddies etc., which they apparently would prefer to not have to deal with.
That's 100 million Norwegian Kroner (NOK), not $100 million... About $13 million.
They're not being built in Oslo. Contrary to what some people seem to believe, Oslo is not that far north ;). The plant is being built near Hammerfest, far north of the arctic circle.
Read the article more closely. It isn't Oslo that is getting the generator, but Hammerfest. Just like most news sources, the place listed in the start of the article is not the place the article is about, but the origin of the story.
As for what I've used to run on a 16MB box, you're right: No Mozilla, Open Office or anything related to Java. That wasn't the point. The point is that even if I just start X on my box at home, with 128MB of RAM, and still have more than 64MB left any heavy disk activity will still make the box slow to a crawl, even for basic things like moving windows. When doing the same on an old, classic Pentium box I used to run, with 16MB RAM and a good SCSI controller and disk, the same thing was never a problem.
You demonstrate very clearly that you don't know what the hell you're talking about. The issue isn't IO throughput - IO throughput is good in my current system, but at the expense of good interactivity.
And most home users DO notice this, whenever they try to start an application. Try starting OpenOffice on a system with a crappy IDE controller, regardless of the amount of RAM. Try the same with Mozilla. The only benefit extra RAM will buy you in a system that hardly ever swaps in the first place is caching of more apps in the buffercache, but that will only help you if you keep starting the same apps. In my case I tend to load Mozilla and Evolution and keep them running more or less permanently (and my system still rarely swaps, in fact to verify that swapping didn't affect the performance much, I turned off my swap partition and had no problems but the performance issues were still as bad).
If you truly believe that anyone would see significant improvements from adding more memory when their system rarely swaps, and you rarely start any large apps more than once after booting the machine, then you're deluded.
Yes, for your example, if someone is running a setup that waste tons of memory at startup, they might benefit. But that's not the case here, and not even that common - I've never come across a manufacturer that don't adapt their setup to perform reasonably with the amount of memory it ships with, including keeping the amount of crap that gets loaded at startup down in "low memory" machines.
And if you think I'm complaining about not wanting to spend $40, you obviously can't even read properly, considering that what I've been saying is that I'd prefer to spend my money on getting a proper SCSI setup, and that will cost significantly more than $40. But contrary to buying more meory, it won't be money straight down the toilet.
Fact is, we don't NEED more memory, unless we keep on running more and more bloated software. I run my share of bloated software, but certainly not enough to need any more than 128MB, and more most of what I do I'd have no problems working with much less without noticing. Same goes for CPU. My current box has a 1GHz Athlon, but I'm practically never close to 100% CPU utilization - even passing 20% would be uncommon. That's not what slow my system down. Disk is. Not primarily transfer rates, though they suck too (though the gap in transfer rates between IDE and SCSI is closing). Whats much more important is latency, and for cheap IDE controllers the way they slow the rest of the system to a crawl whenever I do IO.
This is a growing problem with low cost PC's: All the emphasis is on raising CPU speeds etc., and the way they finance that is by cutting corners all over the place, including junk like Winmodems (which has the same issues - ever tried a Conexant based one under Linux? It makes the box extremely unresponsive while dialling, for instance) and cheap IDE. Those things are quickly becoming much more noticable on low cost PC's than memory limitations.
Your attitude is a good demonstration of what's wrong with advances in computing these days. The main chokepoints for an ordinary home desktop is shitty IO subsystems and bloated software, not RAM or CPU.
When I DO get around to upgrading my machine, a good SCSI controller and a decent disk will be far higher on my list than more memory.
As for stability, the programs I used were stable, and even when I was writing my own assembly programs I rarely crashed the machine. But the very nature of a machine without MMU is that it is trivial to crash and programs needs to be written to a much higher quality to be usable...
But I still like the Amiga. It still has innovative features that I aren't in widespread use, such as Datatypes, a sane, well working DLL system, a standard scripting language for automating applications that made it easy to expose functionality from your app (Arexx), Assigns (aliases for paths that shows up in file selector windows etc.), Screens (thought some version of Enlightenment apparently introduced screens that works approximately like the Amiga, except for the ability to have different resolutions on different screens) and more.
I like Linux, and it's certainly more solid, but from a design point of view the Amiga was much more well structured and clean than any Linux distribution I've used (and don't get me started on Windows ;)
First you transmit a random stream of bits, and you verify which bits weren't eavesdropped on. Then you use the bits you know isn't in the hands of an adversary as a onetime pad for the real message.
The longer the key, the longer it takes to encrypt and decrypt the data.
Often you can choose between a free standing drive that's high quality and expensive, or several cheap drives combined in a RAID at the same price. If the chance of the single drive failing any given day is 1% (which is of course ridiculously high), and the alternative is two cheaper drives mirrored with 5% likelyhood of either of them dying, then provided that the RAID controller is as reliable as the controller for the other drive, the mirror solution is 5%*5% = 0.25% likely to lose data every day, and is indeed the better choice.
RAID doesn't automatically give you more reliability to a single drive, but for mirroring and RAID-5 it gives you more reliability than a single drive of the same type that you build your RAID array with. Sometimes you use RAID in order to get high reliability, sometimes you use it to be able to cut cost but retain reasonable reliability.
Sounds like in the projects you mentioned, they had equated RAID with reliability, and then gone with a cheap solution believing that RAID is RAID regardless of the quality of the individual components, not because of an inherent problem with IDE RAID.
It's depressing to have a 1GHz Athlon system with 128MB RAM and see system responsiveness being worse that classic Pentium servers I used 7-8 years ago whenever there's disk activity...
In the past they've even licensed their technology to their partners before they started using in their own manufacturing in some cases, in order to ensure acceptance in the market ("you buy this tech, and you'll have an advantage over everyone for a while, us included").
The Amiga OS in itself was rock stable. The problems (and the guru meditations :) came whenever you used an application that contained any bugs - which meant quite frequently. So yes, the OS was stable, the system as a whole wasn't thanks to the lack of memory protection. You don't have to go back that far though, to find Amiga users arguing that memory protection was a waste of resources and shouldn't be added to the OS.
Thats bullshit, and shows that you have no idea what you're talking about. No Amiga model was even capable of a text mode, and that was one of the things that really annoyed me when I got my first Amiga, as updating a screenful of text was sometimes slower than on my C64 (got better once people started learning how to exploit the platform, though).
The did walk away from it, but felt obliged to let people know how Microsoft had tried to use them for their PR ploy. Given the way Microsoft treated them I think that's entirely within their rights.
Try asking the American public what they think about Microsoft and Bill Gates, and see the value of good marketing.
As for your question, read the open letter from Schoolnet to Microsoft. Schoolnet got an offer of cheap Acer machines, but Acer's distributor was contractually bound by Microsoft to deliver the machines with Windows preloaded (presumably the same type of deal that Microsoft are no longer allowed to use in the US). When approached, Microsoft would not offer them to donate the licenses, costing about 9000 USD, but instead offered Office licenses worth about 2000 USD instead.
Schoolnet then entered into negotiations with Microsoft as they "would be happy to provide Microsoft with an opportunity to develop a potential alternative to our viable Open Source LTSP refurbished LAN and stand-alone Linux-PC solutions" (from Schoolnets letter). In other words: They'll incure 9000 USD worth of Windows license fees if they buy the Acer PC's, so they wanted to see whether they could actually get any use out of it.
However Microsoft tried stiffing them by offering the software for "free" except for some unspecified R&D costs.
Schoolnet further wrote: " I have, from the very beginning made it VERY clear that SchoolNet has NO desire to REPLACE Linux with Microsoft, but would be happy to accommodate an AFFORDABLE Microsoft diskless refurbished thin-client LAN alternative for potential use in areas where Microsoft distributors would be able to provide technical support to such proprietary Microsoft LAN alternatives."
It should be pretty clear that they're ready to use what they have and try to get just the hardware donated, or even to buy the hardware, but Microsoft tried to trick them into a deal they could gain a PR advantage from but that would end up being costly to Schoolnet without giving them much to show for it.
So "only" 250,000 euros might be more than enough to produce a report that provide workable strategies.
Cambridge most certainly does pay Microsoft for it under Microsofts Select 4 licensing. See this page for some superficial information about (no prices etc).
Sure, many of these projects will be smaller because they can draw on open source components. But similarly many more projects will suddenly make economic sense exactly because the company can cut the projected cost by improving an existing open source project instead of building something from scratch.
For software projects that aren't directly revenue generating, that is a critical factor. Open source has already been the enabler for several projects I've worked on: Thanks to Linux and other open source we've been able to cut the cost of the projects enough that we've been able to spend money elsewhere instead - including on further software development of features we would've gone without if we didn't have access to open source.
We're not creating a lack of jobs - we're widening the market by creating a platform of commodity software that can be customized cheaply enough to enable projects that would otherwise be dead long before getting even to a requirements specification because of cost issues.
Modifying well tested commodity software is also a smaller risk, and less complex, and as such should hopefully in the long run reduce the failure rates of IT projects, which would further increase the chances of getting projects approved.
I'm not saying the end result isn't good, but don't think that he's doing this purely out of good will considering the amount of money he can save by donating shares this way.