An alternative that takes advantage of their existing stores is that the device takes some sort of memory stick which you fill at the store with the content. That way they can rent digital content, and you won't have to return it. It will be quicker to put several movies on a memory stick at a store than download for quite a while. And having the settop box means they don't have to worry about getting it working with a general purpose desktop machine with all the resulting issues, can lock it down to the particular box etc.
Static linking is rare on Windows. Instead you link dynamically and install the needed shared libraries if necessary. Some apps totally botch that up leading to DLL hell.
The problem with static linking is that you don't pick up updates. For example if you statically link to the openssl libraries, then when the system one is updated your app doesn't pick that up. Linux has a boatload of shared libraries. For example Firefox links against 42, xemacs links against 33, xterm links against 21.
While that is the case superficially, every release of every distro comes with different versions of the various shared libraries. That leaves the following possibilities:
Ship software as source and expect customers to compile it
Provide source and get popular enough so that the distros do the compilation for you
Provide the software yourself, and pick some subset of all Linux distros/versions to support
Note that for Windows you can provide a single setup.exe and it will run on everything from Windows 98 through Vista.
Here are some examples of Linux applications:
Opera where they pick a sizable number and versions of distros
Oracle supports RHEL, SUSE Linux Enterprise Server and Asianux
Basically a closed source vendor has to pick a limited number of distros and versions for logistical reasons. You'll also notice that they typically only support x86 processors, but it is rare indeed to see something like PowerPC Linux support. The Linux Standards Base was supposed to address this, but for example they said RPM is the standard packaging mechanism. That annoys anyone using Debian, Slackware etc. If they had picked Deb as the standard then it would have annoyed Redhat/Fedora/Suse etc instead.
So in summary, Linux is only Linux is only Linux if you provide your software as source. Even then, if the distros don't package you then you'll still end up supporting a subset of all versions of all distros of all processors.
Those are far too much work. I just recently got an Apricorn Drivewire which lets you plugin 40 pin PATA, 44 pin PATA (as used in laptops) and SATA drives but without the whole metal case bit. It was $15 at Frys after a $10 rebate.
Lenovo actually make it rather easy to address. They have a piece of software known as base system administrator http://www-307.ibm.com/pc/support/site.wss/TVAN-ADMIN.html#TBSA which when run puts a config file in the recovery partition saying what packages you actually want. You then do a recovery install and you'll have exactly what you wanted.
The reason is because the US does not use dedicated area codes for cellphones like most other countries do. Consequently as a caller you cannot tell the difference between calling someone's home phone vs calling someone's cell phone as they will have the same area code. Conceptually the reason why you have to pay for incoming calls is because the call goes to your home area via conventional means, and then goes by radio to your phone. You have to pay for that last radio hop. (Of course it doesn't really work like this now, but that is how it all started). This also means that you don't get charged extra to call a cellphone as happens in many other countries.
US consumer psychology is also very different. Historically US consumers have always preferred fixed bills versus variable bills, even though many would save money with variable bills. This is the reason that local phones calls are free - the cost is fixed, not actually free. The Internet also took off here early on because of that - plans were almost entirely fixed cost. For cell phones, everyone fixates on the plan with how many bundled minutes it includes (fixed cost). Competition has led to voice minutes being underpriced, so the carriers ding on other services such as data, SMS, sending/receiving picture messages etc. Some carriers (Verizon Wireless) go so far as deliberately crippling features in phones they sell so that the only way to do various things is via them, for a charge. (And in general phones are carrier locked in the US, and cannot be used with another carrier even if unlocked, or can but with significantly reduced functionality). Verizon even went so far as making SMS messages very expensive if you don't buy a bundle to encourage people to sign up for bundles they mostly don't use fully. To put things in perspective, a text message consumes about as much bandwidth as one tenth of a second of voice, but is typically charged the same as 60 to 90 seconds of voice.
Apologies for not being able to cite the consumer preferences for fixed billing source. A story was posted on/. several years about the research paper, but I haven't been able to find it again.
Were you using an external measurement or relying on the software to report it's own usage?
I was relying on the battery reporting via ACPI which is shown in the Gnome Power Manager under Linux and the Lenovo Power Manager under Windows. I believe both read the same underlying data from the battery. Measuring externally on a laptop (I do have the equipment) is harder since you'd have to plug a charger in at which point the software behaves differently.
It seems the GNOME brightness feature doesn't dim the backlight, but rather fades the screen...
That is correct. However I use the xbacklight program which does change the actual backlight brightness and does show power consumption differences. For example 10%=14.2W, 50%=16.7W, 100%=19.6W.
As for the other things, I was using a non-GL desktop (xine won't play movies). I have tried experimenting with various things like turning off the CD polling, removing usb controller modules etc but even combining everything still doesn't bring it close to Windows. Note that I bought the laptop specifically because it uses Intel everything and Intel's support of open source. There will probably have to be a few more rounds of power saving additions in the kernel and user space before Windows is matched (or exceeded).
I did on my dual boot laptop (Lenovo T61). Under Windows with no tuning etc, all radios on, power consumption was 12W. Under the latest Gutsy (32 bit tickless kernel) with no tweaking and same screen brightness consumption was 18W. Turning off the physical radio switch reduced consumption by 1W. Doing all the powertop suggested tweaks brought things down by another watt. (Incidentally there wasn't much difference between the 32 bit tickless kernel and the 64bit tickfull kernel suggesting that most power consumption is in the various peripherals and devices).
Basically Windows consumed significantly less power than Linux (about 30-40% less). From the lesswatts site it is good to see that more power saving will be in the Linux kernel in the 2.6.24/25 timeframe (eg SATA power saving) as well as user space (eg not polling SATA cdroms every 2 seconds looking for media changes). I guess we'll be seeing those updates in Hardy Heron.
I think the best analogy is that mainframes are like large big rigs. They are expensive and slow
but they carry a heck of a lot of stuff. Sure your Toyota could outrun it, park in better spots etc but
it will never haul 40 tons of lumber in one go and is unlikely to do so in multiple gos.
Mainframes will end about the same time they stop using big things to transport large loads.
This is not workable. In fact it is the single most selfish thing you can do.
You have no way of knowing if the message you respond to is spam or not. If it is spam then
you respond to a forged email address which basically means you are spamming an innocent other
person. http://en.wikipedia.org/wiki/Challenge-response_sp am_filtering
I like many other admins consider these auto-responses spam and report them. Ultimately
you will find yourself on email blacklists.
Also don't forget the browser on the Nintendo DS and Wii. The Wii one works pretty well with most sites. I got the DS one this weekend and so far have only found one site that is quick and works exceptionally well - BBC News. Most other sites like to have layers of site links, menus, advertising in every corner etc which are pretty much unworkable on the DS.
Why risk your money and time starting your own venture when you have it made at google?
If Cringely had done even a tiny bit of research he would know about the Google Founders' Awards where Google gives employees multi-million dollars for entrepreneurial achievement. Basically Google pays you almost what you would have got from an external venture, but without you having to leave the company.
Incidentally there are explanations for some of the behaviour. For example the "you want feature X, you code it" is roughly saying that if it is important to you, then it should be important to you! It is quite likely that the person you are talking to already has way more than enough to work on. What they really meant to say was, "I already have more than enough higher priority items to work on. If you want new feature X sooner because it is a higher priority to you, then you should code it or find someone else who will."
As for not responding, it probably wasn't the case where you were doing stuff, but one project I worked on interacted with other hardware (all of which were buggy and very different even between seemingly similar model numbers). There were explicit instructions at the bottom of each mailing list message, at the sign up to the mailing list, in the Help > Support menu item, on the SF help page, in the online doc and everywhere else I could think of about how to make support request. It included stuff like actually saying what operating system, hardware, model, error messages were etc. Yet the *vast* majority of people didn't bother to give *any* of the information requested. I tried to help, but got fed up. In the end I quit the project. Other developers just ignored the messages unless there was something very specific of interest in what they said.
Why not let developers get a taste for the development tools without buying a new system?
Developers also have the issue of how to support the various Mac versions out there. Once 10.5 comes out, there will
be three versions in the wild - 10.3, 10.4 and 10.5 (also add in cpu combinations!). Today a developer would have to get 3 separate machines or triple-boot.
You also can't do snapshots to capture various configurations (eg spotlight turned off completely vs on).
As far as I can tell, most developers just end up supporting the most recent two versions. Of course that helps
Apple in the short term as it forces users to upgrade to get the latest versions of software. But those users
will eventually get very annoyed at being on a "forced" upgrade cycle.
One notable incident was the terrible GNOME file chooser
For some reason, this is actually a UNIX trait. You should have seen the file selection
dialogs in Motif, Athena and various earlier X toolkits. It was as if programmers decided they
hated their users. Many applications even wrote their own choosers. Oh boy did they suck. The Gnome
chooser is way better than the bad old days, but as you rightly point out it isn't something to be proud off.
(Try selecting a file or directory starting with a dot sometime!)
The situation used to be similar on Windows. Every "programming Windows 3.0" book used to have a
chapter on writing your file selection dialog. One of the changes in Windows 3.1 was the provision of a
standard file selection dialog.
I'll add my voice to those asking developers to learn from this:-)
You have the wrong SCO. The original SCO developed Tarantella and sold all of the UNIX business to Caldera in 2001. Old SCO renamed itself to Tarantella. Caldera failed in their Linux venture and after a few years decided to rename themselves SCO. It is that 'new' SCO/old Caldera that is pulling all the stunts suing IBM. Tarantella/Old-SCO was bought by Sun last year.
Tarantella has supported Linux for a long time. Citrix has only ever done enough with Linux in order to not annoy Microsoft. There is no technical reason why Citrix couldn't have released a server product for Linux.
Usually that sort of thing is kind of like a pre-nup. They are trying to make clear what each side had before
getting together. The assumption is that they will have lots and you'll have almost nothing, so consequently
you list all your stuff and assume anything else is theirs. Should there be any "trouble" in the "marriage"
it makes clear who had what.
I've signed something like that 4 times in the last 3 years. I write something like the following in
that space:
A large body of works identified as copyright by me sometimes jointly held with others. Examples
include but are not limited to the following:
Open source project A
Random crap on my web page
etc etc etc
If I get any grief I ask them to write down all their stuff. That usually settles things.
The reason for the "progress" so far is that the folks working at OSAF are all senior people, some already "independently wealthy". Consequently you get lots of high level design (or as Joel calls it, Architecture Astronauts) but not much actual real work.
Python is a good language for writing a standalone PIM. However I question the point of a standalone program.
Today you can't tell if email coming from Amazon.com is important unless you also have been watching my web browsing. If I was there in the last few days then I'd be excited about what is shipping to me. Conversely if I haven't been in years, then it is spam. A good PIM can only be worthwhile if it takes into all of your activities over time with whom you communicate and that must take into account web, blogs, mail etc. The problem today is not storing information, but making sense of it and working out which is more valuable and when.
Are you sure? The printer does tell me pagecounts per cartridge and they are different, as well as page density for each cartridge. I have definitely seen it not decrease the colour counts on using black and white only. In fact right now it has black at 923 pages and CMY at 770. The amount remaining per cartridge is different (eg I have less C left than M or Y). Page densities are 6, 4, 5 and 5% respectively,
There is also a menu option where you can tell it to keep printing - no need to play tricks on the cartridges. There is a good explanation at http://www.bunniestudios.com/wordpress/?p=53#comme nt-977 - the main articile also shows the circuit board.
Slashdot Mirror
An alternative that takes advantage of their existing stores is that the device takes some sort of memory stick which you fill at the store with the content. That way they can rent digital content, and you won't have to return it. It will be quicker to put several movies on a memory stick at a store than download for quite a while. And having the settop box means they don't have to worry about getting it working with a general purpose desktop machine with all the resulting issues, can lock it down to the particular box etc.
Static linking is rare on Windows. Instead you link dynamically and install the needed shared libraries if necessary. Some apps totally botch that up leading to DLL hell.
The problem with static linking is that you don't pick up updates. For example if you statically link to the openssl libraries, then when the system one is updated your app doesn't pick that up. Linux has a boatload of shared libraries. For example Firefox links against 42, xemacs links against 33, xterm links against 21.
While that is the case superficially, every release of every distro comes with different versions of the various shared libraries. That leaves the following possibilities:
Note that for Windows you can provide a single setup.exe and it will run on everything from Windows 98 through Vista.
Here are some examples of Linux applications:
Basically a closed source vendor has to pick a limited number of distros and versions for logistical reasons. You'll also notice that they typically only support x86 processors, but it is rare indeed to see something like PowerPC Linux support. The Linux Standards Base was supposed to address this, but for example they said RPM is the standard packaging mechanism. That annoys anyone using Debian, Slackware etc. If they had picked Deb as the standard then it would have annoyed Redhat/Fedora/Suse etc instead.
So in summary, Linux is only Linux is only Linux if you provide your software as source. Even then, if the distros don't package you then you'll still end up supporting a subset of all versions of all distros of all processors.
Those are far too much work. I just recently got an Apricorn Drivewire which lets you plugin 40 pin PATA, 44 pin PATA (as used in laptops) and SATA drives but without the whole metal case bit. It was $15 at Frys after a $10 rebate.
Lenovo actually make it rather easy to address. They have a piece of software known as base system administrator http://www-307.ibm.com/pc/support/site.wss/TVAN-ADMIN.html#TBSA which when run puts a config file in the recovery partition saying what packages you actually want. You then do a recovery install and you'll have exactly what you wanted.
The reason is because the US does not use dedicated area codes for cellphones like most other countries do. Consequently as a caller you cannot tell the difference between calling someone's home phone vs calling someone's cell phone as they will have the same area code. Conceptually the reason why you have to pay for incoming calls is because the call goes to your home area via conventional means, and then goes by radio to your phone. You have to pay for that last radio hop. (Of course it doesn't really work like this now, but that is how it all started). This also means that you don't get charged extra to call a cellphone as happens in many other countries.
/. several years about the research paper, but I haven't been able to find it again.
US consumer psychology is also very different. Historically US consumers have always preferred fixed bills versus variable bills, even though many would save money with variable bills. This is the reason that local phones calls are free - the cost is fixed, not actually free. The Internet also took off here early on because of that - plans were almost entirely fixed cost. For cell phones, everyone fixates on the plan with how many bundled minutes it includes (fixed cost). Competition has led to voice minutes being underpriced, so the carriers ding on other services such as data, SMS, sending/receiving picture messages etc. Some carriers (Verizon Wireless) go so far as deliberately crippling features in phones they sell so that the only way to do various things is via them, for a charge. (And in general phones are carrier locked in the US, and cannot be used with another carrier even if unlocked, or can but with significantly reduced functionality). Verizon even went so far as making SMS messages very expensive if you don't buy a bundle to encourage people to sign up for bundles they mostly don't use fully. To put things in perspective, a text message consumes about as much bandwidth as one tenth of a second of voice, but is typically charged the same as 60 to 90 seconds of voice.
Apologies for not being able to cite the consumer preferences for fixed billing source. A story was posted on
You can also watch the James Bond film, Goldeneye.
I was relying on the battery reporting via ACPI which is shown in the Gnome Power Manager under Linux and the Lenovo Power Manager under Windows. I believe both read the same underlying data from the battery. Measuring externally on a laptop (I do have the equipment) is harder since you'd have to plug a charger in at which point the software behaves differently.
That is correct. However I use the xbacklight program which does change the actual backlight brightness and does show power consumption differences. For example 10%=14.2W, 50%=16.7W, 100%=19.6W.
As for the other things, I was using a non-GL desktop (xine won't play movies). I have tried experimenting with various things like turning off the CD polling, removing usb controller modules etc but even combining everything still doesn't bring it close to Windows. Note that I bought the laptop specifically because it uses Intel everything and Intel's support of open source. There will probably have to be a few more rounds of power saving additions in the kernel and user space before Windows is matched (or exceeded).
XP Professional.
I did on my dual boot laptop (Lenovo T61). Under Windows with no tuning etc, all radios on, power consumption was 12W. Under the latest Gutsy (32 bit tickless kernel) with no tweaking and same screen brightness consumption was 18W. Turning off the physical radio switch reduced consumption by 1W. Doing all the powertop suggested tweaks brought things down by another watt. (Incidentally there wasn't much difference between the 32 bit tickless kernel and the 64bit tickfull kernel suggesting that most power consumption is in the various peripherals and devices).
Basically Windows consumed significantly less power than Linux (about 30-40% less). From the lesswatts site it is good to see that more power saving will be in the Linux kernel in the 2.6.24/25 timeframe (eg SATA power saving) as well as user space (eg not polling SATA cdroms every 2 seconds looking for media changes). I guess we'll be seeing those updates in Hardy Heron.
I think the best analogy is that mainframes are like large big rigs. They are expensive and slow but they carry a heck of a lot of stuff. Sure your Toyota could outrun it, park in better spots etc but it will never haul 40 tons of lumber in one go and is unlikely to do so in multiple gos.
Mainframes will end about the same time they stop using big things to transport large loads.
This is not workable. In fact it is the single most selfish thing you can do.
You have no way of knowing if the message you respond to is spam or not. If it is spam then you respond to a forged email address which basically means you are spamming an innocent other person. http://en.wikipedia.org/wiki/Challenge-response_sp am_filtering
I like many other admins consider these auto-responses spam and report them. Ultimately you will find yourself on email blacklists.
Also don't forget the browser on the Nintendo DS and Wii. The Wii one works pretty well with most sites. I got the DS one this weekend and so far have only found one site that is quick and works exceptionally well - BBC News. Most other sites like to have layers of site links, menus, advertising in every corner etc which are pretty much unworkable on the DS.
If Cringely had done even a tiny bit of research he would know about the Google Founders' Awards where Google gives employees multi-million dollars for entrepreneurial achievement. Basically Google pays you almost what you would have got from an external venture, but without you having to leave the company.
Incidentally there are explanations for some of the behaviour. For example the "you want feature X, you code it" is roughly saying that if it is important to you, then it should be important to you! It is quite likely that the person you are talking to already has way more than enough to work on. What they really meant to say was, "I already have more than enough higher priority items to work on. If you want new feature X sooner because it is a higher priority to you, then you should code it or find someone else who will."
As for not responding, it probably wasn't the case where you were doing stuff, but one project I worked on interacted with other hardware (all of which were buggy and very different even between seemingly similar model numbers). There were explicit instructions at the bottom of each mailing list message, at the sign up to the mailing list, in the Help > Support menu item, on the SF help page, in the online doc and everywhere else I could think of about how to make support request. It included stuff like actually saying what operating system, hardware, model, error messages were etc. Yet the *vast* majority of people didn't bother to give *any* of the information requested. I tried to help, but got fed up. In the end I quit the project. Other developers just ignored the messages unless there was something very specific of interest in what they said.
Developers also have the issue of how to support the various Mac versions out there. Once 10.5 comes out, there will be three versions in the wild - 10.3, 10.4 and 10.5 (also add in cpu combinations!). Today a developer would have to get 3 separate machines or triple-boot. You also can't do snapshots to capture various configurations (eg spotlight turned off completely vs on).
As far as I can tell, most developers just end up supporting the most recent two versions. Of course that helps Apple in the short term as it forces users to upgrade to get the latest versions of software. But those users will eventually get very annoyed at being on a "forced" upgrade cycle.
Joel has a very good article on this at joelonsoftware.com/articles/fog0000000052.html written several years ago.
I wonder if Google will forward your gmail address if you decide to quit?
Ah, that would be same Messagelabs that inundates me with backscatter spam.
http://www.google.com/search?q=fios+rip+out
You do know that they rip out your existing copper wiring before they will install the fiber? That is one way that they do screw you.
For some reason, this is actually a UNIX trait. You should have seen the file selection dialogs in Motif, Athena and various earlier X toolkits. It was as if programmers decided they hated their users. Many applications even wrote their own choosers. Oh boy did they suck. The Gnome chooser is way better than the bad old days, but as you rightly point out it isn't something to be proud off. (Try selecting a file or directory starting with a dot sometime!)
The situation used to be similar on Windows. Every "programming Windows 3.0" book used to have a chapter on writing your file selection dialog. One of the changes in Windows 3.1 was the provision of a standard file selection dialog.
I'll add my voice to those asking developers to learn from this :-)
You have the wrong SCO. The original SCO developed Tarantella and sold all of the UNIX business to Caldera in 2001. Old SCO renamed itself to Tarantella. Caldera failed in their Linux venture and after a few years decided to rename themselves SCO. It is that 'new' SCO/old Caldera that is pulling all the stunts suing IBM. Tarantella/Old-SCO was bought by Sun last year.
Tarantella has supported Linux for a long time. Citrix has only ever done enough with Linux in order to not annoy Microsoft. There is no technical reason why Citrix couldn't have released a server product for Linux.
Usually that sort of thing is kind of like a pre-nup. They are trying to make clear what each side had before getting together. The assumption is that they will have lots and you'll have almost nothing, so consequently you list all your stuff and assume anything else is theirs. Should there be any "trouble" in the "marriage" it makes clear who had what.
I've signed something like that 4 times in the last 3 years. I write something like the following in that space:
If I get any grief I ask them to write down all their stuff. That usually settles things.
The reason for the "progress" so far is that the folks working at OSAF are all senior people, some already "independently wealthy". Consequently you get lots of high level design (or as Joel calls it, Architecture Astronauts) but not much actual real work.
Python is a good language for writing a standalone PIM. However I question the point of a standalone program.
Today you can't tell if email coming from Amazon.com is important unless you also have been watching my web browsing. If I was there in the last few days then I'd be excited about what is shipping to me. Conversely if I haven't been in years, then it is spam. A good PIM can only be worthwhile if it takes into all of your activities over time with whom you communicate and that must take into account web, blogs, mail etc. The problem today is not storing information, but making sense of it and working out which is more valuable and when.
Are you sure? The printer does tell me pagecounts per cartridge and they are different, as well as page density for each cartridge. I have definitely seen it not decrease the colour counts on using black and white only. In fact right now it has black at 923 pages and CMY at 770. The amount remaining per cartridge is different (eg I have less C left than M or Y). Page densities are 6, 4, 5 and 5% respectively,
e nt-977 - the main articile also shows the circuit board.
There is also a menu option where you can tell it to keep printing - no need to play tricks on the cartridges. There is a good explanation at http://www.bunniestudios.com/wordpress/?p=53#comm