I agree that wired is more reliable, in addition to other factors.
On a related note, using Linux, when my wireless connection is managed by the dhclient utility I find
that it is more likely to immediately reconnect after a failure than when it is managed by plain ifup
at bootup. Presumably it wouldn't be too hard to add the algorithms used by dhclient to ifup, and
this would improve functional wireless reliability on Linux.
CDR replaced floppies for software installation years ago, but the advent of cheap USB pen drives with massive storage, reliability, and convenience benefits over floppies at an acceptable price was the other shoe dropping.
If you look at revenue and profits in their last 10Q compared to the year before, it's clear that PCs and printers are driving the improvement, rather than servers or services.
One effect of buying Compaq - getting a larger retail presence for HP - was much needed. But there were cheaper ways of doing that than buying the whole company. Even though retail has lower margins, it is important for mindshare and understanding customer preferences in different ways. HP should have focused on retail sooner, but better late than never.
A more prosaic, but more generally useful feature that would be relatively straightforward for X11 WMs to implement would be sub-windows that let you group and then rigidly translate on the desktop, as well as scroll within the sub-window, a set of other client windows, typically from different client applications.
I installed from a beta version of the dapper release about a week ago. The machine used has an atheros wireless card in it and no other networking. The install preview had me up and connected to the internet through my wireless router without asking me anything. The install to hard disk did the same with a few non-techy questions like what did I want to call my machine, etc.
I agree with the author's rebuttal to Schwartz (and would also point out the silliness of the premise that McNealy gets credit for anything Sun the company did while he was CEO), but I'll add that Sun did do a lot to fight the mono-culture when it was most threatening and to keep Unix commercially viable for a lot longer than many predicted. It's hard to predict how things would have evolved without that.
I'm not sure what the intended functionality of this $100 laptop is, but I imagine it includes browsing the web, sending email (possibly using the web browser) and word processing. The fact that lots and lots of scientific work was done on much, much less powerful computers than the one Negroponte proposes is probably irrelevant to the intended audience. Unfortunately, many of them will fill that their web experience is being compromised by an inability to run all of the fat plugins demanded by so many web sites that nowadays don't even bother with providing thin alternatives.
Re:They aren't alternatives.
on
Practical Mono
·
· Score: 1
A while back I tried the Java Qt bindings and thought they were pretty nice. So if interface portability is only required for Unix/X11, Win32, and MaxOSX that might be a good alternative as well.
My point isn't to pass out criticism, but rather to encourage decision-makers (I'm using this term literally, not as a euphemism for people controlling corporate purse strings) to read the individual benchmarks and see how they apply to their typical usage patterns. In particular, spending a high percentage time creating the initial filesystem would be a very unusual pattern of usage, and therefore that test skews the analysis against ext? in the geometric average.
It doesn't make sense to give equal weight to the time required for file system creation as for the other types of tasks unless you do it roughly as often (which is highly unlikely).
Other people have mentioned that it is not uncommon to use slower CPUs for fileservers since more CPU is often overkill. But even for workstations, one situation that should be of interest to many people is compiling one or more large source trees - there CPU usage of the filesystem is very relevant.
I recently bought an HP "business" notebook at a good price because it was refurbished. My reason for picking HP was mainly that their keyboards are better than current generation Dells and Sonys.
The best solution for most of these white collar injuries is to make a habit of getting up and stretching every 15-30 minutes (this also explains why drinking a lot of coffee is good for programmer health:) Making this routine will help your hands, eyes, back, etc.
Even more important than bad (or less optimized = more cost effective) programming is the fact that the size of the data we want to process keeps getting bigger. Examples where this is obvious include things like computer games, meteorology, data mining, physical simulation, etc. There is still plenty of need for CPU power because each advance can be met by a desire to work on bigger/harder problems. But these bigger problems don't occur anymore in word processing, web browsing, or doing business forms, so computer gaming is the most activity where more processor can be useful.
I've tested the glossy logitech "performance mouse" MX 510 side by side with its form identical cousin the MX 500 and less snazzy grey mouse gives better performance for click response. Apparently the grey plastic is more flexible than the bright translucent stuff. http://www.logitech.com/index.cfm/products /product list/US/EN,CRID=19
Your point is good, but it's not what the SuSE guy said. If this is the problem, he should have said that RedHat is creating fragmentation of the software interfaces.
Reason: I moved and the my new home is too far from the local switch for DSL.
Other comments: With QWest I had a regular ISP, not MSN. In order to use with Linux howerver, I needed the Cisco external modem and the "deluxe" package. It was around $30/month for speed closer to 640/256. I've found customer service and reliability to be pretty good with both Qwest and Comcast.
It's not relevant to any of the parents - but I completely disagree about the merits of Java vs. its libraries. The language itself is reasonably good but the libraries are, at the same time, too maximalistic in the features vs. performance tradeoffs of what they do provide and too minimalistic in the virtual OS facilities that they provide to the Java programmer who is interested in portability.
The basic problem seems to be that the correctness proof starts from the premises of a logical description of what the program is supposed to do and how the abstract machine it runs on will operate, and it is at least as easy to make mistakes in these logical descriptions as it is to make mistakes in the program to be analyzed.
A 'priviledged' user on the internet is someone with admin access to a host that is always connected, has regular domain name, a static IP address, a fast connection, and runs its own services. The completely priviledged user would be able to run any sort of network server/client that they wished. A completely underpriviledged user would be someone who occasionally uses a laptop/PDA that they don't admin that connects behind a NAT and depends on other servers/domains for e-mail, web, etc. P2P is about technologies for allowing the underpriviledged user to participate in the same type of networking applications, especially content generating ones, that the priviledged user does. There are a set of technologies like http tunneling, virtual DNS, network discovery, etc. that allow simply allow this to happen.
Slashdot Mirror
Is there a port of Konsole or some similarly functioned terminal emulator to Win32?
I agree that wired is more reliable, in addition to other factors. On a related note, using Linux, when my wireless connection is managed by the dhclient utility I find that it is more likely to immediately reconnect after a failure than when it is managed by plain ifup at bootup. Presumably it wouldn't be too hard to add the algorithms used by dhclient to ifup, and this would improve functional wireless reliability on Linux.
CDR replaced floppies for software installation years ago, but the advent of cheap USB pen drives with massive storage, reliability, and convenience benefits over floppies at an acceptable price was the other shoe dropping.
Good leader > no leader >> bad leader
Nothing in this piece convinces that common sense is wrong.
If you look at revenue and profits in their last 10Q compared to the year before, it's clear that PCs and printers are driving the improvement, rather than servers or services.
One effect of buying Compaq - getting a larger retail presence for HP - was much needed. But there were cheaper ways of doing that than buying the whole company. Even though retail has lower margins, it is important for mindshare and understanding customer preferences in different ways. HP should have focused on retail sooner, but better late than never.
A more prosaic, but more generally useful feature that would be relatively straightforward for X11 WMs to implement would be sub-windows that let you group and then rigidly translate on the desktop, as well as scroll within the sub-window, a set of other client windows, typically from different client applications.
I installed from a beta version of the dapper release about a week ago. The machine used has an atheros wireless card in it and no other networking. The install preview had me up and connected to the internet through my wireless router without asking me anything. The install to hard disk did the same with a few non-techy questions like what did I want to call my machine, etc.
I agree with the author's rebuttal to Schwartz (and would also point out the silliness of the premise that McNealy gets credit for anything Sun the company did while he was CEO), but I'll add that Sun did do a lot to fight the mono-culture when it was most threatening and to keep Unix commercially viable for a lot longer than many predicted. It's hard to predict how things would have evolved without that.
I'm not sure what the intended functionality of this $100 laptop is, but I imagine it includes browsing the web, sending email (possibly using the web browser) and word processing. The fact that lots and lots of scientific work was done on much, much less powerful computers than the one Negroponte proposes is probably irrelevant to the intended audience. Unfortunately, many of them will fill that their web experience is being compromised by an inability to run all of the fat plugins demanded by so many web sites that nowadays don't even bother with providing thin alternatives.
A while back I tried the Java Qt bindings and thought they were pretty nice. So if interface
portability is only required for Unix/X11, Win32, and MaxOSX that might be a good alternative as well.
My point isn't to pass out criticism, but rather to encourage decision-makers (I'm using this term literally, not as a euphemism for people controlling corporate purse strings) to read the individual benchmarks and see how they apply to their typical usage patterns. In particular, spending a high percentage time creating the initial filesystem would be a very unusual pattern of usage, and therefore that test skews the analysis against ext? in the geometric average.
It doesn't make sense to give equal weight to the time required for file system creation as for the other types of tasks unless you do it roughly as often (which is highly unlikely).
Other people have mentioned that it is not uncommon to use slower CPUs for fileservers since more CPU is often overkill. But even for workstations, one situation that should be of interest to many people is compiling one or more large source trees - there CPU usage of the filesystem is very relevant.
I recently bought an HP "business" notebook at a good price because it was refurbished. My reason for picking HP was mainly that their keyboards are better than current generation Dells and Sonys.
Same story for me with an nc8000. Ubuntu install was perfect and everything seems to work (also haven't tried the modem or hibernation).
The best solution for most of these white collar injuries is to make a habit of getting up and stretching every 15-30 minutes (this also explains why drinking a lot of coffee is good for programmer health :) Making this routine will help your hands, eyes, back, etc.
Even more important than bad (or less optimized = more cost effective) programming is the fact that the size of the data we want to process keeps getting bigger. Examples where this is obvious include things like computer games, meteorology, data mining, physical simulation, etc. There is still plenty of need for CPU power because each advance can be met by a desire to work on bigger/harder problems. But these bigger problems don't occur anymore in word processing, web browsing, or doing business forms, so computer gaming is the most activity where more processor can be useful.
I've tested the glossy logitech "performance mouse" MX 510 side by side with its form identical cousin the MX 500 and less snazzy grey mouse gives better performance for click response. Apparently the grey plastic is more flexible than the bright translucent stuff.s /product list/US/EN,CRID=19
http://www.logitech.com/index.cfm/product
Your point is good, but it's not what the SuSE guy said. If this is the problem, he should have said that RedHat is creating fragmentation of the software interfaces.
I don't believe that any software library is anything like a security architecture.
But when I looked around 2 years ago for the best crypto library to actually write code with, I settled on Wei Dai's Crypto++
http://www.eskimo.com/~weidai/cryptlib.html
Reason: I moved and the my new home is too far from the local switch for DSL.
Other comments: With QWest I had a regular ISP, not MSN. In order to use with Linux howerver, I needed the Cisco external modem and the "deluxe" package. It was around $30/month for speed closer to 640/256. I've found customer service and reliability to be pretty good with both Qwest and Comcast.
It's not relevant to any of the parents - but I completely disagree about the merits of Java vs. its libraries. The language itself is reasonably good but the libraries are, at the same time, too maximalistic in the features vs. performance tradeoffs of what they do provide and too minimalistic in the virtual OS facilities that they provide to the Java programmer who is interested in portability.
The basic problem seems to be that the correctness proof starts from the premises of a logical description of what the program is supposed to do and how the abstract machine it runs on will operate, and it is at least as easy to make mistakes in these logical descriptions as it is to make mistakes in the program to be analyzed.
A 'priviledged' user on the internet is someone with admin access to a host that is always connected, has regular domain name, a static IP address, a fast connection, and runs its own services. The completely priviledged user would be able to run any sort of network server/client that they wished. A completely underpriviledged user would be someone who occasionally uses a laptop/PDA that they don't admin that connects behind a NAT and depends on other servers/domains for e-mail, web, etc. P2P is about technologies for allowing the underpriviledged user to participate in the same type of networking applications, especially content generating ones, that the priviledged user does. There are a set of technologies like http tunneling, virtual DNS, network discovery, etc. that allow simply allow this to happen.